marshal Launch marshal

Build · Govern · Deploy — the enterprise AI agent factory, self-hosted on AWS

marshal.build

Turn plain-English ideas into governed, deployed AI agents on AWS.

Describe what you need in chat. marshal generates a structured specification — requirements, design, tasks — builds the agent, and deploys it to a governed AWS Enclave: budget-capped, risk-gated, auditable end to end.

Launch marshal See how it works
  • Build a document summarizer for our claims team
  • requirements.md — user stories & acceptance criteria
  • design.md — architecture & data flow
  • tasks.md — implementation plan
  • Deployed to a governed Enclave · risk-gated · budget-capped · audited

How it works

Build. Govern. Deploy.

The end-to-end promise: no AI/ML expertise required, and governance is a stage — not an afterthought.

  1. From plain English to deployed AI agent

    Describe your agent in chat. marshal shapes it into a structured specification — requirements, design, tasks — then generates and deploys a serverless AI agent from it. The specification is the auditable contract; the agent is the running product.

  2. Guardrails on every step

    Templates constrain models and architecture, every AI call is logged and budget-capped, and an AI risk gate routes medium and high-risk apps through maker-checker review before anything ships.

  3. One click into a governed Enclave

    Approved apps deploy into an isolated AWS account vended on demand — tagged, budget-capped, and recycled on teardown. Live status streams the whole way.

Who it’s for

One platform, every role covered

From first idea to enterprise-wide policy, marshal keeps everyone on the same governed rails.

Business Users

Build AI agents by describing requirements in plain English — no AI or ML expertise needed.

Power Users

Full control over model selection, prompt tuning, orchestration logic, and deployment parameters.

Platform Admins

Centralized governance: model access controls, cost guardrails, audit logging, and AI risk-assessment gates.

Enterprise CIO

Governed AI innovation at scale without shadow-IT risk, compliant with your internal AI policies.

Features

Everything between the idea and the audit log

Chat, spec engine, templates, and deployments — one governed pipeline.

Streaming AI chat

Converse with Claude on Amazon Bedrock. Responses stream in real time while marshal shapes your requirements.

Full spec generation

One idea becomes a complete spec — requirements, design, tasks — with live preview and automatic versioning.

Spec editor with diff & rollback

Edit any section directly, compare versions side by side, and roll back the moment a draft goes sideways.

Templates & guardrails

Start from curated templates with model allowlists and budget caps baked in by your platform team.

One-click Enclave deployments

Ship into an isolated AWS account vended on demand — leases, budget caps, and automatic recycling built in.

Governance & cost attribution

Every deployment is tagged, budget-capped, and attributed to its owner — so spend and risk stay visible.

Built for enterprises that answer to auditors

  • Runs in your AWS account
  • Cognito SSO
  • Private-by-default backend
  • Every deployment tagged and budget-capped

Ready to turn an idea into an agent?

Open marshal, describe what you need, and ship it to a governed Enclave today.

Launch marshal