QuantMD

Isn’t it just crazy that one needs so much programming expertise just to write the code to orchestrate two AI agents, using existing libraries?

CrewAI: 15,000 lines of code.
LangGraph: 50,000 lines.
The thing I actually needed: ~150 lines.

So I built litecrew.

Here's a complete multi-agent workflow:

researcher = Agent("researcher", model="gpt-4o-mini")
writer = Agent("writer", model="claude-3-5-sonnet")

(researcher, writer)
def write_article(topic):
research = researcher(f"Research {topic}")
return writer(f"Write about: {research}")

That's it. No YAML. No config files. No framework to learn.

What it does:
→ Sequential handoffs (A → B → C)
→ Parallel ex*****on (fan out, collect)
→ Tool calling
→ Token tracking
→ Works with OpenAI, Anthropic, or Ollama

What it doesn't do (on purpose):
→ No hierarchical agent management
→ No state machines
→ No streaming

The philosophy: SQLite doesn't try to be PostgreSQL. Sometimes simple is the feature.

If you need the full power of CrewAI, use CrewAI. This is for when you just want agents to pass data without reading 200 pages of docs.

Open source. MIT licensed. Your API keys stay on your machine.

Link in comments 👇

Just shipped Ollama integration for SoulSearch, an open-source Chrome extension that gives you an AI assistant on every webpage.

The key difference from Comet/etc: Your memory lives in a Git repo you own — not on someone else's servers. Every save is versioned. You can restore any previous state. You control your data. (Context my earlier post on SoulSearch: https://www.reddit.com/r/Perplexity/comments/1rvlej3/comment/obdaakb/?context=3)

What's new in v0.3:
• Ollama support — Run llama3.2, qwen2.5, or any local model. Fully offline capable.
• Session-specific memory — Quick notes that stay with one session, separate from your Git-backed global memory
• Agent mode — Browser automation with tool calling
• Web search — Brave API integration (optional, BYOK)

Ollama setup:
Start with CORS enabled for Chrome extensions:

OLLAMA_ORIGINS="chrome-extension://*" ollama serve

Tool calling note: Vision models (llama3.2-vision) don't support tools! There's a separate "Agent Model" setting — use vision for chat, text model for browser automation features of SoulSearch (or change to another LLM provider with models that do support tool calling; anthropic, openai and gemini are supported!)

Memory architecture:
• SOUL.md — Your agent's identity/personality
• MEMORY.md — Accumulated knowledge, Git-versioned
• Session memory — Per-conversation, not synced

Everything stays local with Ollama. Git sync is optional and goes to YOUR repo (GitHub, GitLab, Gitea, self-hosted — whatever).

MIT licensed. This Ollama compatible version is not in Chrome Web Store yet — load unpacked from GitHub.

Links:

• GitHub: https://github.com/menonpg/soulsearch/tree/feat/ollama-support (feat/ollama-support branch)

• Blog post with details: https://menonlab-blog-production.up.railway.app/blog/soulsearch-v03-ollama-session-memory-brave-search

• Pre-Ollama SoulSearch version is also live on the Chrome Web Store! Approved and published March 19, 2026. Install it here: https://chromewebstore.google.com/detail/soulsearch/jfcbpgdapidbpobibghabfpdoipknoig

On fixing the browser automation bits for Ollama, what Ollama models are you all using for tool calling?

SoulSearch - Chrome Web Store AI research assistant with persistent memory, powered by soul.py

🧠 Excited to share something I've been building — soul.py!

Most AI conversations end and the agent forgets everything. Next session? Back to zero. No continuity, no growth, no memory of who you are or what you've talked about.

We built soul.py to fix that — persistent identity and memory for any LLM agent, in about 10 lines of code.

It uses two plain markdown files:
📄 SOUL.md — the agent's identity, values, personality
📄 MEMORY.md — a timestamped log of everything it learns

Works with Claude, GPT, Ollama — whatever you're using. Zero required infrastructure.

pip install soul-agent

I’ve been pushing it further and further — v2.0 (released today) adds a query router that automatically decides whether to use fast semantic search (RAG) or deep recursive synthesis (RLM) depending on what you ask. The router gets it right ~90% of the time without you touching anything.

Here’s a littele piece on the philosophy behind it — framing agent memory through the lens of Darwinian evolution. Worth a read 👇
🔗 Blog: https://menonlab-blog-production.up.railway.app/blog/soul-py-darwin-evolution-agent-identity/

⭐ GitHub: https://github.com/menonpg/soul.py
📦 PyPI: https://pypi.org/project/soul-agent/
🎮 Live demo: https://soulv2.themenonlab.com

Would love to hear what you think — especially from anyone building agents or working with persistent AI systems.

soul.py v2.0 — RAG + RLM Demo

🚀 Excited about my upcoming talk titled, "Impact of GenAI on Machine Learning Education: Empowering New Programmers," at the Incorporating Generative AI into Learning Experiences Virtual Showcase of University of Maryland.

📅 Date: April 26, 2024
⏰ Time: 11:10 a.m. - 11:30 a.m. ET
📍 Location: Room D, Virtual Showcase

We will explore the transformative role of Generative AI in enhancing scientific programming and machine learning education, and discover how tools like Retrieval Augmented Generation (RAG) and innovative, enabling cloud-based IDEs for developers on the cloud (eg: Google Colab, Lightning.ai), are revolutionizing the way we teach and learn, making complex subjects more accessible and engaging for new programmers.

👉 We'll dive into the architecture of a Virtual Teaching Assistant that I've adopted in my courses, view a live demonstration of the same, discuss real-world applications, and share insights on the profound impact of GenAI on student success and engagement.

https://www.usmd.edu/cai/incorporating-generative-ai-learning-experiences-virtual-showcase

Incorporating Generative AI into Learning Experiences Virtual Showcase | USM Center for Academic Innovation Sponsored by MarylandOnline, Montgomery College, the University System of Maryland William E. Kirwan Center for Academic Innovation, and the USM Council of University System Faculty

Every now and then we get to do something truly groundbreaking:

Our COVID-19 test analyzes volatile organic compounds in breath samples using proton transfer mass spectrometry, powered by QuantMD AI🙂

See our publication on E Clinical Medicine (The Lancet)

https://doi.org/10.1016/j.eclinm.2021.101207

QuantMD is excited to be the AI behind this groundbreaking endeavor to diagnose covid19 earlier and less invasively than ever before!

Mercyhealth, RJ Lee Group Study Real Time COVID-19 Breath Test in Janesville - Mercyhealth JANESVILLE, Wisconsin — Mercyhealth and RJ Lee Group Inc, based in Monroeville, PA, is excited to announce it has recently concluded COVID-19-related research at Mercyhealth North in Janesville. This research is the first step in developing a real time, non-invasive breath test for COVID-19 that p...

Meet new collaborators and learn Complexity Science by doing @ Complexity Weekend!

Complexity Weekend is a virtual conference focused on developing innovative solutions that correlate with the COVID19 pandemic. This is a virtual call to action for our community of computational and data scientists. Registration is free!

This event will feature Complexity Science-inspired lectures, discussions, and workshops on Friday night and Saturday. All attendees will then engage in a collective brainstorming and team formation process Saturday afternoon, followed by a facilitated hackathon experience with these teams on Sunday.

Check it out here:
https://www.complexityweekend.com

#

Complexity Weekend 2020 | May 22, 2020 - May 24, 2020 Join our facilitated online hackathon May 22-24. Learn Complexity Science by solving problems related to COVID-19 and beyond. Work with an interdisciplinary team over the weekend with expert guidance

Always touching to receive appreciation from clients (especially when cupcakes are involved) 🙂

A little interactive app from QuantMD to offer a sense for daily deaths and cumulative deaths owing to COVID-19:
https://quantmd.shinyapps.io/visualizeCorona/

Data is updated daily (last update 3/14)