Open-Source Full-stack Agentic RAG

Today’s top AI Highlights:

1. Open-source full-stack RAG with agentic capabilities

2. Google Code Wiki to generate docs and visualize code

3. API to monitor the web for state changes

4. Google Colab now runs inside VS Code

5. Open-source agent in Rust that can't delete your database

& so much more!

Read time: 3 mins

AI Tutorial

Google's free 5-day AI Agents Intensive course ends today!

This course by Google's ML researchers covers everything from agent foundations to production-ready systems.

And they released 5 whitepapers (~300 pages total) that are staying up for free.

Here's what the whitepapers cover:

Introduction to Agents (50 pages)

• The 5 levels of agents: from pure reasoning to self-evolving systems

• Core components: Model (the brain), Tools (the hands), Orchestration (the nervous system)

• Moving from prototype to enterprise-grade systems

Agent Tools & Interoperability with MCP (50 pages)

• How tools actually work and why descriptions matter more than code

• The Model Context Protocol architecture explained

• Security risks nobody talks about: dynamic capability injection, tool shadowing, confused deputy problems

Context Engineering: Sessions, Memory (70 pages)

• The difference between prompt engineering vs context engineering

• Session management: conversation history + working memory

• Memory as an active curation system, not just "save the conversation"

Agent Quality (50 pages)

• The four pillars: Effectiveness, Efficiency, Robustness, Safety

• Process evaluation: judging reasoning, not just outputs

• Building agents that learn from production failures

Prototype to Production (40 pages)

• Evaluation gates, circuit breakers, and evolution loops

• Turning demos into production systems

• Real-time monitoring and continuous evaluation

The best part?

All whitepapers are 100% free and packed with zero fluff.

If you're serious about leveling up your AI skills and staying ahead of the curve, subscribe now and be the first to access our latest tutorials.

Don’t forget to share this newsletter on your social channels and tag Unwind AI (X, LinkedIn, Threads) to support us!

Latest Developments

Open-Source Full-Stack RAG with Agentic Capabilities 🤖📚

Here's an open-source RAG platform that actually ships with the pieces you'd build yourself anyway.

AgentSet gives you a complete RAG-as-a-service setup with hybrid search, reranking, and agentic capabilities already wired up. Everything is MIT-licensed and ready to be self-hosted.

The platform processes documents across 22+ formats while preserving structure and metadata, then uses multi-vector embeddings and semantic search to handle retrieval. What makes it practical is the developer experience: TypeScript and Python SDKs with full type safety, REST APIs with OpenAPI specs, and namespace isolation for managing multiple document collections. AgentSet also includes query expansion and answer validation built into the agent layer, so you can deploy sophisticated RAG apps that reason about results before responding.

Key Highlights:

1. Document processing - Handles PDF, DOCX, HTML, TXT, and 18 other formats with automatic content extraction, intelligent chunking that preserves structure, and metadata-based filtering for partitioned retrieval.

2. Agentic Deep Research mode - Agents can plan multi-step queries and validate answers before returning results, giving you accuracy beyond standard retrieval patterns when you need deeper reasoning on complex questions.

3. Hybrid retrieval with reranking - Combines semantic search with multiple embedding strategies and sophisticated reranking algorithms, so you get high-accuracy results without manual tuning of retrieval parameters.

4. Built-in source citations - Automatically tracks and surfaces document sources for every answer, giving users direct access to verify information.

5. Full self-hosting - Deploy on your own infrastructure with complete access to the codebase, or use the managed service. Both options include production monitoring and scalability features.

An AI scheduling assistant that lives up to the hype.

Skej is an AI scheduling assistant that works just like a human. You can CC Skej on any email, and watch it book all your meetings. It also handles scheduling, rescheduling, and event reminders.

Imagine life with a 24/7 assistant who responds so naturally, you’ll forget it’s AI.

• Smart Scheduling
  Skej handles time zones and can scan booking links

• Customizable
  Create assistants with their own names and personalities.

• Flexible
  Connect to multiple calendars and email addresses.

• Works Everywhere
  Write to Skej on email, text, WhatsApp, and Slack.

Whether you’re scheduling a quick team call or coordinating a sales pitch across the globe, Skej gets it done fast and effortlessly. You’ll never want to schedule a meeting yourself, ever again.

The best part? You can try Skej for free right now.

Sign up for free

Google Code Wiki to Generate Docs and Visualize Code 📑📊

Google just turned code repositories into self-updating Wikipedia pages.

Google just launched Code Wiki, a platform that generates interactive docs, visualizes code relationships, and maintains living documentation for entire code repositories. No outdated README files or guessing what that mysterious function does.

The system ingests public repos and creates interactive wikis where you can jump from conceptual explanations straight to the actual code implementations. It's powered by Gemini, and you get a chat agent that answers specific questions using the always-current wiki as context. Plus, it auto-generates architecture, class, and sequence diagrams that stay synced with your code changes.

Code Wiki is currently in public preview, freely available for many public repos. Google is also building a Gemini CLI extension so teams can run this on their private repos locally.

Key Highlights:

1. Interactive Navigation - Navigate from high-level concept explanations directly to specific code files, classes, and functions they reference.

2. Context-Aware Chat Agent - Ask the Gemini-powered agent specific questions about your repository and get answers grounded in the always-updated wiki.

3. Auto-Generated Diagrams - Code Wiki automatically creates and maintains architecture, class, and sequence diagrams that reflect the current state of your code, so you can visualize complex relationships without manual diagramming.

4. Private Repo Support Coming - A Gemini CLI extension is in development for teams to run Code Wiki locally on internal repositories, bringing the same automated documentation capabilities to private codebases.

Code Wiki is quite like DeepWiki by Cognition AI. But here are some major differences:

Quick Bites

API to monitor the web for state changes
Instead of asking "what's new?" every hour, just set a query that tells you when something's new. Parallel just launched Monitor API, essentially a persistent web search that pushes updates whenever new information matching your query appears online. Think of it as a webhook for the entire internet: define what you're tracking (competitor launches, market signals, research developments) and the API streams updates as they appear, triggering agents or workflows automatically. Try it for free on their Developer Platform.

Google Colab now runs inside VS Code
Google just released an official Colab extension for VS Code. You can now spin up Colab runtimes with GPU/TPU access without leaving your local editor. Install it from the Marketplace, connect to a runtime, and suddenly your local notebooks are running on Colab's GPUs and TPUs. Train models, run inference workloads, and execute compute-heavy code while you stay in your customized VS Code setup.

Execute code with Sandboxes for DeepAgents
LangChain just shipped sandbox integrations for DeepAgents, letting you execute agent-generated code safely in remote environments from Runloop, Daytona, and Modal. This keeps your agent logic local while pushing all execution to configurable remote environments, which means you can spin up multiple agents in parallel, each with its own dependencies, and tear them down when you're done.

xAI releases faster and more intelligent Grok 4.1
xAI released Grok 4.1 after running two weeks of silent A/B tests across production traffic, where it won 64.78% of blind comparisons against its predecessor. The model is optimized heavily for conversational quality and emotional intelligence. It ranks #1 on LMArena's Text Arena with 1483 Elo in thinking mode, and remarkably, its non-reasoning variant at #2 (1465 Elo) beats every other model's full reasoning configuration. Grok 4.1 is now available to all users on grok.com, X, and the iOS and Android apps. It is rolling out immediately in Auto mode.

Tools of the Trade

1. Stakpak - The most secure CLI agent in Rust, built for operations & DevOps. Designed for the grittiest parts of software development with enterprise-grade security features including mutual TLS (mTLS) encryption, dynamic secret redaction, and privacy-first architecture.

2. Anytool - Lets LLMs create and execute tools on-demand rather than pre-defining dozens of static functions. When your agent needs a capability (QR generator, CSV parser, API client), Anytool generates executable Bun code with npm access, caches it by hash, and runs it in an isolated sandbox with persistent KV/SQLite state. First run takes 3-5 seconds, subsequent calls under 100ms.

3. GitHub Repo Visibility - Scans your repo and shows how easily it can be found across GitHub search, Google, and LLMs. It reviews tags, descriptions, structure, and metadata, then gives clear fixes to raise your repo’s visibility score. It’s basically a fast audit that tells you what’s missing and how to improve it.

4. Orion - A visual agent API that combines VLM reasoning with specialized computer vision tools to execute tasks like object detection, segmentation, OCR, and image/video editing, that frontier models typically can't execute reliably. It handles images, videos, and documents through an OpenAI SDK-compatible interface with auditable, structured outputs.

5. Awesome LLM Apps - A curated collection of LLM apps with RAG, AI Agents, multi-agent teams, MCP, voice agents, and more. The apps use models from OpenAI, Anthropic, Google, and open-source models like DeepSeek, Qwen, and Llama that you can run locally on your computer.
   (Now accepting GitHub sponsorships)

Hot Takes

1. There are people becoming millionaires writing code in COBOL, and here you are wondering whether Python is better than TypeScript.

   ~ Santiago
   
   

2. If you're having any personal life problems my advice is to hold off on solving them until Gemini 3, the benchmarks are looking good enough that it'll probably be better at solving them than you

   ~ Andy Masley

That’s all for today! See you tomorrow with more such AI-filled content.

Don’t forget to share this newsletter on your social channels and tag Unwind AI to support us!

Unwind AI - X | LinkedIn | Threads

Awesome LLM Apps | Sponsor Us

PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one, two (or 20) of your friends 😉