About Atlas
What does knowledge work look like when we have AGI?
Atlas is an internal team at Basis dedicated to finding the answer. We’re a group of builders, former founders, and engineers that embed directly within our teams, so we can build intelligence deeper into their workflows. As models continue to improve, they change how companies are built and how teams operate. Basis is building the first AI-native company at scale. Atlas is responsible for making that possible.
We’ve already shipped agent teammates that push code directly to production, systems that keep company knowledge current automatically, and services that diagnose and repair themselves. These systems do real work inside Basis today. Every advance in model capability opens another set of questions about how work should be organized and another set of things we can build.
Our ambition is much larger than making each person 10% more productive. We’re building zero-human loops that handle entire jobs end-to-end and free up our team to take on higher level work. Getting there means inventing new infrastructure, interfaces, operating models, and ways for humans and agents to work together.
Atlas began when Basis had fewer than forty people. Since then, Basis has raised $100M. Atlas has the resources, access, and mandate to pursue this seriously. Basis intends to grow at extraordinary speed, and doing so requires a fundamentally different operating model—one in which agents take on an increasing share of the work. If “What is the future of knowledge work?” is the question you most want to spend the next several years answering—and you want to make the answer real at a fast-growing company—reach out.
What will you be doing?
Building and shipping internal AI agents that push code to production, manage company knowledge, and handle entire jobs end-to-end
Owning context architecture and management—defining sources of truth and ensuring documentation stays synchronized with reality
Designing and building integrations and MCPs that make internal tooling accessible to agents across the company
Creating evaluation suites, monitoring agent trajectories, and iterating to improve agent performance
Inventing new infrastructure, interfaces, and operating models for human-agent collaboration as Basis scales at extraordinary speed
What we're looking for
First principles thinker
Builder mindset
Track record of launching technical B2B SaaS or AI products.
Excited about working with coding agents (we give every engineer and operator an unlimited token budget)
Excited about working at the frontier of applying agents to real economy problems
We started Basis with the core belief that agents would become an integral part of knowledge work in the near future, and that accounting would be among the first and most important domains where this happened. Accounting is structured, high-stakes, and essential to every business on earth. It's also one of the most underbuilt areas in technology.
Here we are 3 years later, and Basis can now complete a partnership tax workbook end to end. Basis' abilities improve monthly and the rate of improvement itself is increasing. Every day accountants use Basis to create complex journal entries, debug difficult reconciliations, prepare technical accounting memos, and much more. We're deploying real agents to do real work in the real economy.
The world is going to look very different in 2030, and accounting will be at the center of that transformation in ways most people don't yet appreciate. Read more about how we're thinking about the work.
Team size
120 employeesFounded
Website
www.getbasis.ai/Total funding
Company locations
The Basis team has grown to 80 strong, fully in-person office around the corner from Madison Square Park in Flatiron.
Whiteboards everywhere. The whole team in one room. Not a lot of formal meetings, but interesting conversations happening constantly, about agent architectures, accounting edge cases, what the engineering looks like in five years.
You'll have a ton of responsibility from day one and be surrounded by people who are very good at what they do. The standards are high and the feedback is frequent. People here are kind and fun, but they also push each other to be better.
Most of the team eats lunch and dinner together. Many people hang out outside of work. Nobody has to. Everyone is passionate about something, whether that's accounting, ML, climbing, chess, meditation, philosophy, or ping pong. That energy carries into the work.
Basis builds AI agents that learn accounting the same way accountants do — by being agentic and doing the work. Rather than rebuilding accounting software, Basis's agents operate within the existing accounting software ecosystem, completing complex tasks end-to-end for some of the largest accounting firms in the world. Agents can work for hours at a time, performing tasks like completing partnership tax workbooks, creating complex journal entries, debugging difficult reconciliations, and preparing technical accounting memos.
Think of it as Cursor for accounting — AI that dramatically multiplies accountant productivity by 10x to 100x. Basis's capabilities improve monthly, with the rate of improvement itself accelerating, deploying production-grade, long-horizon agents at scale in the real economy.