GetProfile

GetProfile is an open-source, drop-in LLM proxy designed to solve the "statelessness" of modern AI. It provides your agents with long-term memory and a structured understanding of every user, allowing for personalized interactions without rewriting your existing codebase.

Key Features

• 🔌 Zero-Friction Proxy Integration GetProfile acts as an OpenAI-compatible gateway. By simply updating your baseURL and adding a user ID header, you can add persistent memory to your application instantly, with full support for streaming and existing OpenAI SDKs.
• 📊 Structured User Profiles Unlike standard memory solutions that store unstructured text, GetProfile extracts specific, typed traits—such as expertise level or communication style—complete with confidence scores. This allows your agent to "know" its user rather than just searching through past transcripts.
• 🧠 Automated Context Injection The proxy automatically enriches every outgoing prompt with a relevant system message containing user summaries, traits, and specific memories. This ensures the LLM has the necessary context to provide personalized responses without overloading the context window.
• 🛠️ Fully Customizable Trait Schema You define exactly which user attributes matter for your application. Using a JSON configuration, you can specify how traits should be extracted from conversations and how they should be formatted when injected back into prompts.
• 🛡️ Open-Source & Privacy-First Licensed under Apache 2.0, GetProfile is fully self-hostable via Docker. Because it uses your own PostgreSQL database, your sensitive user data remains entirely under your control and within your infrastructure.

How GetProfile Solves Your Problems

Most LLM applications treat every interaction as a "first meeting," forcing users to repeat preferences and context. GetProfile automates the cycle of Capture → Extract → Inject → Update.

1. The Request: Your app sends a standard prompt to the proxy.
2. The Enrichment: GetProfile fetches the user’s profile from your database and injects it into the system prompt.
3. The Execution: The LLM receives a rich, context-aware prompt and generates a personalized response.
4. The Background Update: After the response is sent, GetProfile asynchronously analyzes the exchange to update traits and store new memories, ensuring zero impact on user-facing latency.

Use Cases

• Personalized Technical Assistants: An AI coding assistant can remember that you prefer Python, work with microservices, and favor concise, technical explanations, eliminating the need to specify these preferences in every session.
• Adaptive Learning Platforms: A tutor bot can track a student's expertise level across different subjects, adjusting the complexity of its explanations based on the "confidence scores" of the student's stored traits.
• Context-Aware Customer Support: Support agents can automatically see a summary of a user's past issues and specific "facts" (e.g., "Uses Kubernetes at work"), providing a seamless experience that feels continuous rather than fragmented.

Conclusion

GetProfile transforms generic AI interactions into personalized user experiences by bridging the gap between stateless LLMs and persistent user data. It is the ideal solution for developers who need to implement sophisticated memory systems without the overhead of building a custom backend.