Today we’re sharing Asimov: our first product milestone on the path to superintelligence. Building highly capable coding agents is a crucial step toward superintelligence more broadly.
In large codebases, engineering teams spend up to 70% of their time understanding and designing code and only 10% writing it.¹ But existing products focus most of their effort on code generation, which results in limited comprehension abilities.
Coding agents share a similar problem. Without deeply understanding large codebases and the business logic around them, coding agents will be stuck with shallow capabilities, unable to solve problems on the critical path to an engineering organization.
That’s why we built Asimov, the code research agent that is best-in-class in codebase comprehension, built for teams and organizations.
In a blind testing with maintainers of some of the largest open source projects, Asimov’s answers to complex questions were preferred 60-80% of the time relative to other coding products.
There are three key components that make Asimov work.
1. Asimov builds a single source of truth for engineering knowledge
The knowledge for your engineering org doesn't just sit in your codebase. It’s scattered across organizational systems. Asimov ingests entire codebases, architecture docs, GitHub threads, chat history, and more. It builds persistent memory of your systems, remembers key decisions, and acts as a trusted brain for an engineering organization.
2. Asimov captures team-wide tribal knowledge
While other products focus on individual developer preferences through rules or README files with instructions for agents, Asimov Memories enable engineering teams to capture team-wide tribal knowledge.
Engineers can update Asimov’s knowledge, e.g. “@asimov remember X works in Y way.” This allows the most senior engineers to offload the context stored in their heads to Asimov, which benefits the team at large. Memories come with a permissioned RBAC (Role-Based Access Control) system to allow an organization to control who can edit Asimov’s knowledge.
3. Asimov is designed to ingest a lot of information
Asimov uses a multi-agent architecture to achieve state-of-the-art performance on code comprehension. The architecture consists of many small long context agents (retrievers) that retrieve relevant information from a large codebase, among other sources, and one large short-context reasoning agent (combiner) that synthesizes this information into a coherent answer to the user query.
We built Asimov for two reasons:
- It is useful. We use Asimov internally to accelerate our own engineering velocity. And it's getting smarter every day.
- It is mission aligned. We believe superintelligent code understanding is the prerequisite for superintelligent code generation, not the other way around.
We’re selecting teams for early access. Join the waitlist here.
We’re also hiring across both product and research roles, see open roles here.