Lamatic.ai

Hey Product Hunt!

I'm a Founding Applied AI Engineer at Lamatic.ai, and I'm excited to finally share something we've been working hard on: AgentKit – our open-source framework for building full-stack agentic applications. Check it out: https://github.com/Lamatic/AgentKit

AgentKit is basically a ready-to-go boilerplate that lets you build multi-agent applications without starting from scratch. We designed it to be modular and scalable, so you can handle complex workflows with fault tolerance and smart routing. Your agents don't just think – they actually coordinate with each other.

It integrates seamlessly with Lamatic's ecosystem (our reasoning APIs, vector stores, orchestration tools), and we've built it to handle real-world scenarios – from knowledge assistants that reason over documents to agent networks that automate entire workflows.

• Deep search agents: https://lamatic.ai/agentkits/kit-agentic-deep-search

• Enterprise assistants with secure connectors (Drive, OneDrive, S3, Postgres...): https://lamatic.ai/agentkits/bundles-assistants

• Embeddable chat widgets you can ship in minutes: https://lamatic.ai/agentkits/kit-embed-chat

I helped architect the core logic – how agents communicate, handle retries, and pass state between each other. I also worked extensively on our Agent Kit Reasoning system, which enables multi-agent collaboration in structured flows: https://www.lamatic.ai/templates/agentkits/agentic/agent-kit-reasoning

Plus, I made sure everything plays nicely with Lamatic's managed infrastructure so you can run these agents serverless and scale them reliably. Integrated all of these into web applications for the world to simply plug and play products into their system.

Why give it a shot?

• Developers: Get a battle-tested foundation instead of reinventing the wheel

• Startups: Quickly prototype and validate your ideas with real agent workflows

• Enterprises: Build secure, production-ready knowledge systems and assistants

• Open-source first: Fork it, extend it, make it yours

We're actively building new AgentKits for different use cases (like hiring automation: https://www.lamatic.ai/agentkits/kit-automation-hiring), adding more connectors, richer UI components, advanced orchestration patterns and scalable kits for handling not just apps, but devops and complex pipelines with AI too!!

I'd genuinely love your feedback – try it out, open issues, suggest features, or contribute. Your input will directly shape where we go next.

Before I wrap up, I have to give a huge shoutout to our amazing dev team @arun2728 @naitik_kapadia @ian_dsouza. None of this would've been possible without their late nights, creative bug-solving, and relentless commitment to making AgentKit something we're all proud of.

Looking forward to hearing what you all think!

Adding a separate logs/reports section is a win, especially for debugging, but how fast can a user pinpoint the exact node causing a failure?

Clean take on agent tooling. Visual builder + GitHub versions is the bit I wanted. Less glue code. Real-time traces and experiments might save me some 2am debug pain. I'll poke AgentKit this weekend; 5-min ship would be wild. If the federated GraphQL plays nice with my schemas, I'm in.

Interesting Product @chuck_whiteman @arun2728 @naitik_kapadia . I’m curious how the new flow builder changes day-to-day use. Does it actually make complex automations easier to set up without extra guidance?

Hey Product Hunt! 👋

While the rest of the team focused on the core infrastructure and visual flow, my mission for Lamatic 3.0 was to make the actual building experience effortless using—you guessed it—AI.

I’m thrilled to introduce the Family of AI Assistants living inside the Studio. I built these to act as your co-pilot, so you never get stuck on a blank screen.

We shipped:
✨ Vibe Builder: Generate full AI agents from a single screen just by describing what you want.
✨ Flow & Node Assistants: Get real-time guidance and a mini-expert inside every node to optimize inputs and fix errors.
✨ Code & Prompt Assistants: Write, refine, and debug code or prompts with contextual suggestions without leaving your workspace.

A massive shoutout to @arun2728 for the guidance and help in bringing this architecture to life!

I would love to know: which part of agent building do you usually find the hardest? Hopefully, our new assistants can help solve that! 👇

Hey Product Hunt! 👋

I'm a Founding Engineer at Lamatic.ai, and I’m the one behind our new Version Control & Environments (Branches) system in 3.0.

Our goal with this release is simple: make the entire user journey smoother, more reliable, and production-ready.

🤔 Why do you need Version Control Systems?

Building production-grade AI agents demands the same engineering discipline as any real-world software system - and that starts with version control. When you're transforming domain expertise into dependable AI flows, you need clarity on every change:

• Which prompt performed best?

• What model configuration worked?

• Who modified what—and when?

That’s why we brought GitHub integration directly into Lamatic Studio. It gives teams a familiar workflow while adding the ability to automatically audit changes, roll back flows or deployments, and maintain full accountability across the board.

🌱 Why environments (branching) matters?

Branching is especially powerful for agent development. It lets you cleanly separate production from development - so you can experiment freely without risking live systems. Spin up a branch, test safely, iterate fast, and merge only when the flow is stable.

In AI systems - where even a tiny prompt tweak can shift behavior dramatically - version control isn’t a nice-to-have. It’s the foundation for building agents that actually perform reliably in production.

Happy to answer questions or dive deeper into how it works!

Specific agents with domain expertise are definitely game-changer. How does Lamatic AI measure "reliable"?