Antinuous
Antinuous
Platform
Use cases
Our mission
Sign up

Infrastructure for AI research.

Antinuous builds the compute, orchestration, simulation, and artifact layers for AI to work on hard research problems. Our agent runs on them, and any agent can call the same primitives or hand work to ours.

Sign up
Our agent and other agents at the top call into swarms of agents, which save results as tables and run simulations, such as a beam coloured by stress, on a cluster of computers

Four layers

Our agent and the app sit on top of four layers. Outside agents reach the same layers through the same tools.

  • A cluster of computers, most of them running work, with more being added at the edge as demand grows

    Compute

    Elastic compute for research. Each run gets its own isolated computer, and large jobs split across thousands of workers at once.

    • A computer per run

      Each run's computer comes with simulation, engineering, and data tools installed, and can be resized for heavy work.

    • Worker pools

      One job runs on many fresh workers at once, each with its own copy of the inputs. Extra workers queue until others finish.

    • GPU clusters

      GPU jobs run on clusters through Antinuous, or on your own Modal account.

  • A lead agent starts a swarm. Each round, weaker branches are dropped after review, and the rest return their results.

    Orchestration

    Agents work together as swarms. A large question is split across many agents, which work in rounds and check results against their sources.

    • Swarms

      When a question can be split up, the agent starts a swarm, and each agent takes part of it.

    • Rounds

      Agents share a board of scored approaches. Each round, the best continue and the rest are dropped.

    • Review

      A second model checks each fact against its source.

  • Four simulations side by side: a loaded plate with a hole coloured by stress, a crystal of atoms, simulated price paths that end up or down, and a storm on a weather grid

    Simulation

    Agents test ideas by running simulations instead of guessing, across engineering, science, medicine, finance, and earth science. They use our own kernels and solvers alongside established solvers.

    • Geometry and CAE

      CAD geometry, meshing, and CAE solvers for structures, fluids, heat, and circuits.

    • Science and medicine

      Models for life science, chemistry, medicine, and earth science, from molecules to climate.

    • Math and finance

      Math and statistics engines, plus pricing, risk, and trading models.

    • Vendor libraries

      Kernels call the hardware vendors' own libraries on GPUs and CPUs.

  • A notebook, a chart, a deck, a molecule, a table, a CAD bracket, and a sequence, each in its own form

    Artifacts

    Results arrive in the formats your field already uses, such as notebooks, documents, decks, charts, maps, tables, CAD models, and protein structures. Each value links to its source or code.

    • Notebooks and documents

      Analysis notebooks, protocols, literature reviews, manuscripts, and reports, with each section cited.

    • Decks

      Slide decks with a sources slide that export to PowerPoint, PDF, or Google Slides.

    • Charts, maps, and tables

      Charts, maps, timelines, and tables that keep their sources.

    • Models and structures

      CAD models with STEP files and drawings, and protein structures, sequences, and molecules that open in 3D.

Connect your agent and steer ours

A local or cloud agent can start our agent, steer it while it runs, and read what it finds. It can also start swarms and clusters itself.

On its own, an agent runs a handful of experiments. The same agent calling Antinuous can start a swarm that runs thousands at once.

Connect

  • Over MCP

    Claude, ChatGPT, Claude Code, Codex, Cursor, and other MCP clients connect to the Antinuous MCP server.

  • Your account

    Your agent signs in with your Antinuous account. Settings in the app shows the setup steps under Coding agents.

  • Over A2A

    Agents that speak A2A find Antinuous from its agent card and start or continue research the same way.

  • Same rules

    Every tool calls the same handlers as the app, so your agent sees only what you can see.

Start and steer

  • Start our agent

    A call to start research hands the question to our agent. When massively parallel computation can speed up the work, it starts a swarm.

  • Steer it

    Follow-ups add questions, more research, or changes while it runs. They wait in the queue until the current run ends.

  • Check or stop it

    Your agent can check a run's status, or stop it and keep the results saved so far.

  • Open it

    Each run returns a link to the run in the app, so you can watch the same work there.

  • Start a swarm or cluster

    Your agent can also start a swarm or a compute cluster directly. Calls go through a fair queue.

  • Change results

    Your agent can change results directly, without sending a follow-up.

Limits

  • An account is required

    Research needs an Antinuous account, and an agent acts as the person who signed it in.

  • Runs take minutes

    Research runs in the background, usually for several minutes, so agents check back on it.

Connect your agent to Antinuous.

Over MCP or A2A, your agent uses the same compute, orchestration, simulation, and artifacts as ours, or hands the work to our agent.

Sign up