What is Terminal Use?

Terminal Use is an orchestration and deployment platform purpose-built for AI agents that operate on filesystems. Positioning itself as "Vercel for background agents," the company provides infrastructure to deploy, manage, version, and scale long-running agents. Part of YC W2026, it was founded by three ex-Palantir engineers — Vivek Raja (CEO), Filip Balucha, and Stavros Filosidis — and has raised M in funding.

The platform is framework-agnostic with native adapters for Claude Agent SDK, Codex SDK, and Vercel AI SDK v6. Developers define agents with a config.yaml and Dockerfile, specify three lifecycle endpoints (on_create, on_event, on_cancel), and deploy via CLI. The platform handles sandboxed execution, persistent filesystem storage across conversation turns, file sharing between agents, versioning with rollback, and environment-specific secret injection.

A key architectural innovation is treating filesystems as independent primitives separate from task lifecycles, enabling workspace persistence, filesystem forking for parallel agent testing, and presigned URLs for direct file transfers. Terminal Use supports preview/production environments with git-based targeting, scheduled execution, and enterprise features including physical data/compute isolation.

Key features

Core capabilities this platform advertises.

Agent hosting & deployment
Sandboxed compute
Scheduling
Message streaming
Long-running agent support

Strengths and tradeoffs

What this tool does well, and the limitations to keep in mind.

Pros

Massive M funding provides significant runway compared to competitors in the space
All three founders have directly relevant Palantir experience building agent infrastructure at enterprise scale
Framework-agnostic design with Claude SDK, Codex SDK, and Vercel AI SDK adapters avoids vendor lock-in
Filesystem-as-primitive architecture is a genuinely novel approach solving real stateful agent problems
Strong developer experience with CLI-first design, git-native versioning, and markdown scenario testing

Cons

Plans & pricing

What's included in each plan, and how the tiers compare.

Starter

Contact for pricing

SDK and CLI access
Agent deployment
Sandboxed execution
Filesystem persistence

Enterprise

Custom

Contact sales for a quote

Common use cases

Developers deploying long-running AI agents

Agent deployment
Background agent hosting
Scheduled agent tasks
Production agent infrastructure

Using Terminal Use with Respan

Terminal Use deploys and manages long-running AI agents, while Respan monitors the LLM calls those agents make. Together they provide both infrastructure management and AI observability for production agent deployments.

Monitor LLM costs and performance for agents deployed on Terminal Use via Respan
Track agent behavior patterns across Terminal Use deployments through Respan traces
Optimize agent LLM spending using Respan analytics for Terminal Use-hosted agents