Best Free Open Patents & Innovation Resources

Persona Pattern

Prompt Pattern

Free

Persona Pattern is a prompt pattern used in AI and large language model systems. Instructing the model to adopt a specific role for better output quality. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to role-playing, content generation, tone control. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,role,control, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Chain of Thought (CoT)

Prompt Pattern

Free

Chain of Thought (CoT) is a prompt pattern used in AI and large language model systems. Guiding the model to verbalize its reasoning steps before providing the final answer. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to complex reasoning, mathematical problem-solving, debugging. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,reasoning,logic, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Retrieval Augmented Generation (RAG)

Agent Pattern

Free

Retrieval Augmented Generation (RAG) is a agent pattern used in AI and large language model systems. Augmenting the LLM's knowledge base with external, relevant documents for grounded answers. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to grounded qa, document summarization, private data query. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,knowledge,search, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Flipped Interaction Pattern

Prompt Pattern

Free

Flipped Interaction Pattern is a prompt pattern used in AI and large language model systems. Asking the user to act as the AI's validator or expert for improved initial inputs. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to requirements gathering, complex topic exploration, creative briefing. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are [URL], which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Sequential Tasking Workflow

Workflow

Free

Sequential Tasking Workflow is a workflow used in AI and large language model systems. Breaking a complex goal into smaller, dependent steps executed one after the other by agents. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to multi-step data processing, complex itinerary planning, software deployment. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,automation,tasks, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Reflection Agent Pattern

Agent Pattern

Free

Reflection Agent Pattern is a agent pattern used in AI and large language model systems. An agent evaluates its own output against criteria and initiates self-correction loops. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to code review, complex document drafting, reducing hallucination. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,selfcorrection,quality, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Zero-Shot Prompting

Prompt Pattern

Free

Zero-Shot Prompting is a prompt pattern used in AI and large language model systems. Providing the task directly without any examples or demonstrations. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to simple classification, basic fact retrieval, standard q&a. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,basic,qa, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Few-Shot Prompting

Prompt Pattern

Free

Few-Shot Prompting is a prompt pattern used in AI and large language model systems. Providing several input/output examples to teach the model the desired behavior. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to style transfer, pattern replication, custom data extraction. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,examples,learning, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Tool-Use Agent Pattern

Agent Pattern

Free

Tool-Use Agent Pattern is a agent pattern used in AI and large language model systems. Enabling an agent to select and execute external functions or APIs (e.g., calculator, web search). The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to real-time data access, calculation, external system integration. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,apis,functioncalling, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Human-in-the-Loop Workflow

Workflow

Free

Human-in-the-Loop Workflow is a workflow used in AI and large language model systems. Incorporating mandatory human review or approval steps into the automated process. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to high-stakes decision making, sensitive content filtering, final legal review. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,review,safety, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Recursive Summarization

Workflow

Free

Recursive Summarization is a workflow used in AI and large language model systems. Summarizing a long document iteratively by chunking and summarizing the summaries. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to handling very long documents, generating multi-level abstracts. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,nlp,documents, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Tree of Thought (ToT)

Prompt Pattern

Free

Tree of Thought (ToT) is a prompt pattern used in AI and large language model systems. Exploring multiple parallel reasoning paths before selecting the best one. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to strategic planning, counterfactual analysis, difficult creative writing. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,strategy,exploration, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Template Injection Pattern

Prompt Pattern

Free

Template Injection Pattern is a prompt pattern used in AI and large language model systems. Providing a structured JSON or XML template for the model to fill in for predictable output. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to structured data extraction, api response generation, report formatting. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,structure,extraction, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

System Design: Microservice Architecture (LLM)

System Design

Free

System Design: Microservice Architecture (LLM) is a system design used in AI and large language model systems. Structuring an application where different LLM tasks (RAG, generation, moderation) are separate services. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to scalability, maintainability, independent deployment. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,architecture,scalability, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Parallel Processing Workflow

Workflow

Free

Parallel Processing Workflow is a workflow used in AI and large language model systems. Executing multiple independent tasks or function calls simultaneously to save time. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to bulk data analysis, concurrent api calls, multi-model evaluation. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,speed,parallel, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

The Critic Pattern

Prompt Pattern

Free

The Critic Pattern is a prompt pattern used in AI and large language model systems. Instructing the model to evaluate a previous answer based on specific criteria and offer critique. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to quality assurance, self-refinement, detailed feedback generation. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,feedback,critique, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Self-Correction Prompting

Prompt Pattern

Free

Self-Correction Prompting is a prompt pattern used in AI and large language model systems. Providing the model's own incorrect answer and asking it to fix the mistake. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to error handling, iterative improvement, knowledge refinement. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,correction,error, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

API Gateway Pattern (AI Context)

System Design

Free

API Gateway Pattern (AI Context) is a system design used in AI and large language model systems. A single entry point for routing and managing requests to multiple specialized AI models. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to access control, rate limiting, traffic routing to llms/vision models. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,api,routing, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Consensus Agent Pattern

Agent Pattern

Free

Consensus Agent Pattern is a agent pattern used in AI and large language model systems. Multiple independent agents answer a query, and a final agent synthesizes the common answer. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to reliability, reducing bias, high-confidence responses. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,reliability,consensus, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Abstractive Summarization Prompt

Prompt Pattern

Free

Abstractive Summarization Prompt is a prompt pattern used in AI and large language model systems. Requesting a summary that generates new sentences and captures the core meaning, rather than extraction. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to generating articles, policy briefs, creative abstracts. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,summarization,nlp, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Extract-Then-Generate Pattern

Workflow

Free

Extract-Then-Generate Pattern is a workflow used in AI and large language model systems. First, extract key facts or entities (structured data); then, use those facts to generate the final text. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to reducing hallucinations, grounded writing, data-driven content. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,data,extraction, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Data Validation Agent

Agent Pattern

Free

Data Validation Agent is a agent pattern used in AI and large language model systems. An agent whose sole purpose is to verify inputs or intermediate outputs against predefined rules. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to input cleansing, schema adherence, safety checks. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,validation,security, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Iterative Refinement Pattern

Prompt Pattern

Free

Iterative Refinement Pattern is a prompt pattern used in AI and large language model systems. Continuously feeding the model's output back to itself with new constraints or context for improvement. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to detailed design creation, long-form writing, complex legal drafting. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,refinement,iterative, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

System Design: Feature Store

System Design

Free

System Design: Feature Store is a system design used in AI and large language model systems. A dedicated centralized service for storing and serving curated machine learning features, including embeddings. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to consistency, reproducibility, latency reduction. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,mlops,data, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

External API Call Workflow

Workflow

Free

External API Call Workflow is a workflow used in AI and large language model systems. A flow that includes authentication, request formatting, external API execution, and parsing the response. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to real-time stock quotes, weather forecasts, database lookup. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,api,integration, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Step-Back Prompting

Prompt Pattern

Free

Step-Back Prompting is a prompt pattern used in AI and large language model systems. Asking the model to first generate a foundational, abstract question before answering the original query. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to complex reasoning, finding underlying concepts, academic research. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,abstraction,reasoning, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Moderation Agent

Agent Pattern

Free

Moderation Agent is a agent pattern used in AI and large language model systems. An agent specialized in flagging or filtering content based on toxicity, bias, or policy violation. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to content safety, compliance, pre-screening outputs. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,safety,moderation, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Semantic Search Workflow

Workflow

Free

Semantic Search Workflow is a workflow used in AI and large language model systems. Using embedding models for vector search to find relevant content before generating a final answer. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to building rag systems, knowledge base search, legal discovery. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,search,embeddings, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

System Design: Event Sourcing (ML)

System Design

Free

System Design: Event Sourcing (ML) is a system design used in AI and large language model systems. Capturing all changes to the AI application state as a sequence of events for auditability and model retraining. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to audit logs, model versioning, feature engineering. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,mlops,events, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Audience Persona Pattern

Prompt Pattern

Free

Audience Persona Pattern is a prompt pattern used in AI and large language model systems. Defining the target audience of the output to tailor language, tone, and complexity. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to marketing copy, technical documentation, educational content. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,audience,marketing, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Code Generation Agent

Agent Pattern

Free

Code Generation Agent is a agent pattern used in AI and large language model systems. An agent specialized in writing, testing, and debugging code based on natural language requirements. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to software development, rapid prototyping, bug fixing. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,code,developer, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Conditional Prompting

Prompt Pattern

Free

Conditional Prompting is a prompt pattern used in AI and large language model systems. Modifying the prompt based on intermediate model output or external conditions. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to dynamic content generation, personalized interaction flows. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,dynamic,control, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Model Delegation Workflow

Workflow

Free

Model Delegation Workflow is a workflow used in AI and large language model systems. Routing requests to different specialized LLM models based on the intent of the user query. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to intent classification, specialized qa, model cost optimization. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,routing,optimization, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

System Design: Load Balancing (Model)

System Design

Free

System Design: Load Balancing (Model) is a system design used in AI and large language model systems. Distributing incoming requests across multiple identical model instances to ensure high availability and speed. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to high traffic, latency reduction, throughput maximization. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,networking,performance, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Abstractive-Extractive Hybrid

Workflow

Free

Abstractive-Extractive Hybrid is a workflow used in AI and large language model systems. Combining extraction of key phrases with abstractive generation for a balanced summary. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to news summarization, legal case overviews. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,summarization,hybrid, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Chain-of-Verification (CoV)

Prompt Pattern

Free

Chain-of-Verification (CoV) is a prompt pattern used in AI and large language model systems. Asking the model to generate multiple drafts, then verify each one against sources. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to fact-checking, high-accuracy reporting, scientific drafting. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,verification,accuracy, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Data Enrichment Agent

Agent Pattern

Free

Data Enrichment Agent is a agent pattern used in AI and large language model systems. An agent that looks up and adds relevant metadata or information to an input before processing. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to crm data update, entity resolution, context provisioning. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,data,enrichment, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Role Reversal Prompting

Prompt Pattern

Free

Role Reversal Prompting is a prompt pattern used in AI and large language model systems. Asking the model to critique its own prompt to reveal ambiguity or incompleteness. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to prompt engineering, clarity improvement, constraint discovery. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,metaprompt,engineering, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Monitoring & Alerting System

System Design

Free

Monitoring & Alerting System is a system design used in AI and large language model systems. Implementing systems to track model drift, latency, and error rates with automated alerts. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to model maintenance, operational health, anomaly detection. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,mlops,monitoring, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Parallel Brainstorming Workflow

Workflow

Free

Parallel Brainstorming Workflow is a workflow used in AI and large language model systems. Generating multiple creative solutions or ideas simultaneously before filtering and selecting the best. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to ideation, creative writing, design concepts. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,creativity,ideation, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

The Refuser Pattern

Prompt Pattern

Free

The Refuser Pattern is a prompt pattern used in AI and large language model systems. Explicitly defining negative constraints or unacceptable output types in the prompt. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to safety filters, avoiding specific styles, legal compliance. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,constraint,safety, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Guardrails Agent

Agent Pattern

Free

Guardrails Agent is a agent pattern used in AI and large language model systems. An agent running before and after the LLM call to enforce policy and format compliance. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to input/output filtering, pii removal, response structuring. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are agent,security,compliance, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Adaptive Few-Shot Prompting

Prompt Pattern

Free

Adaptive Few-Shot Prompting is a prompt pattern used in AI and large language model systems. Dynamically selecting the most relevant examples (shots) from a library based on the current query's embedding. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to context-aware response, improved task generalization. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,dynamic,rtr, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

System Design: Data Versioning

System Design

Free

System Design: Data Versioning is a system design used in AI and large language model systems. Tracking and managing different versions of training data, feature data, and models for reproducibility. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to reproducibility, auditability, regulatory compliance. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,mlops,data, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

RAG with Semantic Cache

Workflow

Free

RAG with Semantic Cache is a workflow used in AI and large language model systems. Using a vector database to cache frequently asked questions and their grounded answers to bypass LLM calls. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to cost optimization, latency reduction, high volume qa. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as an advanced pattern and is generally more appropriate for users designing sophisticated LLM systems, agent architectures, or production-oriented AI workflows. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,cache,optimization, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

The Question Refinement Pattern

Prompt Pattern

Free

The Question Refinement Pattern is a prompt pattern used in AI and large language model systems. Asking the model to simplify or rephrase a complex user query before searching or answering. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to improved rag performance, user intent clarity. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are prompt,clarity,refinement, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

Fine-Tuning Workflow (LoRA)

Workflow

Free

Fine-Tuning Workflow (LoRA) is a workflow used in AI and large language model systems. The process of lightweight fine-tuning a base model on custom data using parameter-efficient methods. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to customization, domain specialization, cost efficiency. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized at an intermediate level and is most useful once the basic concepts of prompting, LLM applications, or agent workflows are understood. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are workflow,training,mlops, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.

System Design: Model Registry

System Design

Free

System Design: Model Registry is a system design used in AI and large language model systems. A centralized repository for storing, versioning, and managing different trained model artifacts. The core purpose of this pattern is to provide a repeatable structure that developers, researchers, students, and AI application designers can apply when building or improving interactions with language models. In practical use, it is especially relevant to deployment, governance, model lifecycle. Rather than treating every AI request or workflow as an isolated implementation, this pattern provides a recognizable approach that can be adapted to different applications, models, domains, and levels of system complexity. It is categorized as beginner-friendly, making it suitable for users who are learning how structured AI interaction and reusable design techniques work. Implementation details can vary depending on the selected model, prompt structure, available tools, external data sources, orchestration framework, and application requirements. When applying the pattern, users should define the intended objective, inputs, expected outputs, constraints, validation method, and failure-handling behavior so that the resulting workflow remains understandable and testable. The associated dataset tags are design,mlops,governance, which can be used to group and discover this pattern alongside related techniques. The source included with this record should be consulted for the original reference or additional technical context, while real-world implementations should be evaluated with representative tasks before production deployment.