Tell us about your company
Atoms AI is the product. DeepWisdom is the company behind it. We launched Atoms AI in January 2026, with the aim of helping startups to take the next step from “vibe coding” to complete “vibe business” creation, through the deployment of autonomous AI teams. Instead of pinning their hopes on simple prototypes, with Atoms AI, entrepreneurs can build a fully functional business, capable of active operation, rather than simply pitching for funding. As such, we’ve already attracted more than 1 million customers from more than 100 countries.
We’re also the team behind MetaGPT and OpenManus, two open-source multi-agent projects on GitHub.
We’re backed by Ant Group and Cathay Capital, who led a $31 million round earlier this year.
What do you think makes your offering unique?
We’ve moved beyond simple AI coding. The ability to bypass expensive professionals and use AI to code and product/platform build using natural language prompts completely revolutionised business, helping
startups of all levels to actually begin. But the vibe business is so much more than that. Atoms AI supports the full cycle, from idea generation and launch to iteration and marketing.
We’re there not just to build prototypes but to help businesses launch, operate, and iterate. Our multi-agent approach more closely mirrors real-world workflows, which isn’t just more scientific but more helpful and widely deployable than single-model coding tools. Ultimately, we’re giving people access to specialist skills with the previously needed budget.
Traditionally, if someone had no experience in marketing, for example, it took time and money to build the requisite knowledge – search engine optimisation, Google Ads campaign management, build a backend. Now, with Atoms AI, anyone can do that quickly and efficiently.
How has your offering and approach evolved?
The first step in the journey was MetaGPT. It’s a concept that I’ve had in mind since high school. I’d always wanted to create a machine that creates itself. As a teenager, I expected that it would solve many problems in the world. And throughout my career, I’ve tried to achieve that in many ways. As an open-source multi-agent framework, that’s what MetaGPT is becoming. And it’s led to the creation of OpenManus.
OpenManus is a project built by a group of young researchers in our team. It’s basically a fully open-source, community-driven AI agent framework designed to replicate the features of the viral Manus AI agent. Atoms AI was built upon that framework, but is more specialised.
Rather than providing a general purpose platform, Atoms AI is an end-to-end platform designed to autonomously build, launch, and scale full-stack businesses, empowering solopreneurs to do more.
More from Interviews
- Meet Jack Steiner, Head Of End-To-End Managed Booking Experience: Checkatrade Express
- A Chat With Nik Kairinos, CEO Of RAIDS On Who Owns AI Risk Within Businesses
- Interview With Tobias Cummins, COO At Pencil On Agentic AI In The Marketing Context
- A Conversation With Dr James Gupta, CEO And Founder Of Online Exam Platform Synap On Student AI Use
- A Chat With Joe Crist, CEO Of Transform 42 Inc. On Why Persistent AI Will Split Business Into Two Worlds
- Systems Thinking in Design: Uliana Salo, Platform Design Leader
- Meet Ahmed Hessam, CEO And Founder Of OSAA Innovation
- A Chat With François Bitouzet, Managing Director On The Importance Of Global Tech Events Like VivaTech
What can we see from your company in the future?
We have a few exciting projects in the pipeline. The first, that I can actually mention, is our Seedance 2.0 Partnership Launch, with ByteDance, which will hopefully take us into new markets, new verticals, and new customer segments.
More generally, in the next 12 months, our goal is to build a product that helps users generate income automatically. So, by the end of this year, users will gain access to the first abstraction layer of MetaGPT: a system that can translate ideas into executable business workflows.
Then, in the next three years, we aim to enable AI to evolve autonomously. Users will only need to define the rubrics: the goals, evaluation criteria, and strategic direction. The AI will then build, test, and improve itself over time. In this model, humans no longer need to manage every execution detail. Their role shifts toward validating ideas, identifying promising directions, and deciding how deep each opportunity should go.