- unwind ai
- Posts
- Clawdbot in Just 500 Lines of Code
Clawdbot in Just 500 Lines of Code
+ AI Agents are now hiring humans
Shubham Saboo & Gargi Gupta
February 04, 2026
In partnership with
Today’s top AI Highlights:
& so much more!
Read time: 3 mins
AI Tutorial
Evaluating startup investments requires hours of research across multiple domains - company analysis, market research, financial modeling, and risk assessment. This setup literally automates this entire workflow with AI agents that work together like a real investment team.
In this tutorial, you'll build an AI Due Diligence Agent Team using Google's Agent Development Kit (ADK) and Gemini 3 models, and Nano Banana.
This 7-agent team researches any startup (from early-stage unknowns to well-funded companies), analyzes the market, builds financial projections, assesses risks, and generates professional investment reports - all autonomously with seamless handoffs and a sophisticated analysis with reports.
We share hands-on tutorials like this every week, designed to help you stay ahead in the world of AI. 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.
Latest Developments
OpenClaw's viral success created a problem nobody expected: developers couldn't sleep knowing they'd given 52+ modules and 45+ dependencies unrestricted access to their machines.
NanoClaw launched as the answer, a personal Claude assistant you can actually understand in 8 minutes.
Built by a developer who refused to run code he couldn't audit, the project strips OpenClaw down to ~500 lines handling everything. Agents execute inside Apple Container sandboxes (or Docker on Linux), with filesystem isolation enforced at the OS level rather than through application permissions.
Setup happens through Claude Code itself—no installation wizard, no configuration files to learn, just run /setup and the AI handles dependencies, authentication, container configuration, and service deployment.
Key Highlights:
Container-Level Security - Agents run in lightweight Linux VMs where only mounted directories are visible. Bash access is safe because commands execute inside the sandbox, not on your host. Mount allowlist validation sits at
~/.config/nanoclaw/where agents can't reach it.Skills-Based Customization - Contributors write skills like
/add-telegramthat teach Claude Code how to transform your fork. No PRs adding features to core—users run skills and get clean code matching their exact needs without bloat.Memory Hierarchy - Global memory in
groups/CLAUDE.mdgets read by all conversations but only written by main channel, while each group maintains isolated memory in its own folder. Claude Agent SDK automatically loads both levels when agents run.Scheduled Task System - Groups schedule recurring or one-time tasks that run as full agents with complete tool access, optionally messaging results back. Main channel manages all tasks across groups, others manage only their own.
Nanoclaw is open source. Clone the repo, run Claude Code, execute /setup, and you have a working assistant. Requires macOS or Linux and Claude Code installed.
How 2M+ Professionals Stay Ahead on AI
AI is moving fast and most people are falling behind.
The Rundown AI is a free newsletter that keeps you ahead of the curve.
It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.
Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses — tailored to your needs.
It keeps getting meta. AI agents can now dispatch humans for real-world tasks. Not a demo, not a concept - 30,000+ people have signed up to be hired by agents.
RentAHuman.ai is an MCP marketplace where agents post tasks that require physical presence or human judgment, and humans claim and complete them for payment. Startup CEOs have signed up. OnlyFans models have signed up. The range of available humans is already wild.
One MCP call from your agent, and a real person shows up to do the job. If ClawTasks was agents hiring agents, this is the next step: agents hiring humans. The inversion is complete.
Key Highlights:
MCP Integration - One tool call from any MCP-compatible agent dispatches a human. Works with Claude Code, OpenClaw, Cursor, and anything else that speaks MCP.
30,000+ Humans Available - Startup founders, content creators, specialists across domains. The marketplace is already populated with real people ready to work.
Real Payments - Agents pay humans directly for completed tasks. Actual money changing hands, not test tokens or simulated transactions.
Physical Task Completion - Agents can now complete tasks that require being somewhere or doing something in the real world. The digital-physical barrier just got a lot thinner.
Quick Bites
Apple turns Xcode into an agent-native IDE
Apple just made Xcode a proper agentic IDE. Claude Agent and Codex can now hook directly into your Swift projects, autonomously searching docs, navigating codebases, tweaking build settings, and even capturing previews to verify their own work. The kicker? Full MCP support means any compatible agent can tap into Xcode's tooling.
Qwen’s open-weight 80B model for agentic coding
Qwen just dropped Qwen3-Coder-Next, an 80B MoE model that only activates 3B parameters per token, basically giving you performance comparable to models 10-20x larger while running on a laptop with 46GB RAM. It's specifically trained for agentic coding workflows with 256K context, scores 70.6 on SWE-Bench Verified, and plugs straight into Claude Code, Cline, and other IDE agents without needing test-time scaling or thinking blocks. Download from HuggingFace or ModelScope, no API calls, no rate limits, full control over inference.
GLM-OCR tops document AI benchmarks with only 0.9B params
Z.ai dropped GLM-OCR, a 0.9B parameter multimodal model, claiming #1 on OmniDocBench. It can handle complex tables, formulas, and code-heavy documents, and is small enough to run on vLLM, SGLang, or Ollama for low-latency deployment. MIT licensed and fully open-sourced with an SDK that supports everything from basic text recognition to structured information extraction via JSON schemas.
ACE-Step 1.5: Open-source Suno alternative
Generates 4-minute music tracks in 20 seconds on 8GB VRAM. Fully open source, run it locally without API costs. If you've been looking for a self-hosted music generation option, this is probably the best one available right now.
Tools of the Trade
ElevenLabs Skills - A collection of Agent Skills for ElevenLabs' voice and audio APIs, covering text-to-speech, speech-to-text, conversational agents, sound effects, and music generation. Follows the Agent Skills specification for compatibility with AI coding assistants.
LNAI - Write project rules, MCP servers, and permissions once in
.ai/and sync them to native formats for Claude Code, Cursor, Copilot, and others. No need to maintain separate config files across different AI tools.AliSQL - Alibaba's MySQL fork that integrates DuckDB as a native storage engine, letting you run analytical queries through MySQL's interface. Includes performance optimizations for large-scale deployments and enterprise features like vector storage support and faster crash recovery.
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
Knowing what not to build when solving a problem is the skill that has made me more money than anything else.
I believe many software projects fail under the weight of all the unnecessary crap people build.
Literally every conversation I have with developers results in some sort of "we'll build this thing just in case."
If you had the unfortunate pleasure of working with me on a project, you'd know I'm very stubborn about YAGNI (You Ain't Gonna Need It). I always try to focus exclusively on what matters today, and don't spend time thinking about the future.
It turns out that many companies love this, and they are willing to pay for people who help them frame their problems using this approach.
~ Santiago
You want AI to do things for you? Use Claude
You want AI to help you do things? Use ChatGPT
Simple
~ vas
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!
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 😉
1
Reply