Our technology

Fullstack engineering

We are a TypeScript-first company. We build frontend and backend systems with a shared language — from interfaces and APIs to data access and production services.

  • TypeScript
  • Node.js
  • Hono
  • Better Auth
  • Drizzle ORM
  • PostgreSQL
  • Next.js
  • Vite
  • React
  • shadcn/ui
  • SWR

Agent engineering

AI Agents

We build agents that connect to your data and tools — embedded in your systems or working alongside your team.

System agents

Agents integrated into software workflows to process data and carry out recurring tasks with defined tools, permissions, and rules.

Supervised agents

Agents that help your team research, write, and complete tasks, with people guiding the work, reviewing results, and approving actions.

  • TypeScript
  • eve
  • Vercel AI SDK
  • MCP
  • OpenCode
  • OpenClaw
  • Qwen
  • GLM

Data architecture

PostgreSQL at the core.

We keep persistent state in PostgreSQL and S3-compatible storage. Other services stay stateless, so we can redeploy them independently while keeping data in place.

  • Queues & workflows

    We use pgmq for queues in PostgreSQL, coordinating background jobs and workflows close to the data.

  • Vector & full-text search

    We combine pgvector for vector search with tsvector for full-text search, supporting hybrid retrieval for AI systems.

  • CDC & realtime

    We use change-data capture (CDC) and PostgreSQL notifications to connect data changes to realtime updates and event-driven workflows.

  • PostgreSQL
  • Supabase
  • RustFS

Local AI infrastructure

Model inference

We host model inference locally on our own infrastructure. We select and configure the hardware and inference engine around the model and workload — from consumer hardware to dedicated servers and large-scale clusters.

Consumer hardware → Dedicated servers → Large-scale clusters

  • vLLM
  • SGLang
  • llama.cpp

Production operations

Infrastructure

We manage infrastructure for all our workloads — from stateful and stateless services to storage, databases, and AI agent sandboxes with code execution capabilities.

  • Application services

    We deploy and operate stateful services that retain data and stateless services that can be independently replaced and scaled.

  • Storage & databases

    We manage the databases and storage that keep application data and files available across deployments.

  • AI agent sandboxes

    We provide isolated execution environments where AI agents can run code and complete tasks with defined resource limits and access permissions.

  • Kubernetes
  • Docker
  • WebAssembly