Skip to content

AI applications

Multi-agent: work that one agent cannot finish goes to a team of them

A complex task split across several agents: one gathers data, one decides, one actually writes to your systems, one reviews. Each agent only gets the permissions and tools it needs, inside an environment with spend limits and a full record of what happened.

A good fit when

  • The task spans several systems and several steps, more than one exchange can handle
  • You need the AI to call real APIs and write real data, not just produce text
  • You need human review, and a record of why something was done

Not a fit when

  • A single ChatGPT account already solves it
  • There is no data yet, and no decision about what problem to solve

What you get

  1. 01Agent design: who does what, who takes over when one is stuck, who is allowed to finish
  2. 02Tool interfaces (MCP/API) that let agents read and write your systems safely
  3. 03Evaluation sets and quality reports, so you can compare before and after a change
  4. 04Usage, cost and trace dashboards: every call is inspectable

Related articles

Write-ups on our founder's blog, covering how it was done and the problems along the way.

Related work

Systems we run ourselves or have published that are the same kind of thing as this service.

  • v-terminal developer platform

    Harness / internal tools

    v-terminal developer platform

    Our own browser-based platform for building and operating software: terminals, Kubernetes, Helm GitOps deploys, CI, registry and an AI coding agent in one screen. This website was shipped from zero using it.

    • Git push to self-hosted runner to Harbor to Helm sync, one pipeline
    • AI console: send a task from your phone, the agent reads the spec, edits and tests

    Next.js · Node.js · SQLite · Kubernetes client · Claude Agent SDK

    See insideHarness
  • AI video studio

    AI applications

    AI video studio

    Turns a PDF or a paragraph into a short video with narration, subtitles and motion: AI writes the script, generates the images and the voice, Remotion renders the file.

    • PDF or text to storyboard to narration to finished file, unattended
    • Narration and subtitles in several languages, with proper CJK line breaking

    Remotion · Vertex AI · Qwen TTS · MinIO

    See insideAI applicationsVisit site
See all work

Frequently asked

Control. When one prompt goes wrong your only move is to rewrite the whole thing. Split the work up and each agent has its own inputs, outputs and evaluation, so you replace the broken part and leave the rest alone. Permissions separate too: the agent that reads data never holds the token that writes.