A short agreement. Then we get you running.
Evaluation access is private and granted under a short written evaluation agreement with Agiliti Technologies. Once it is signed, you receive the product, its documentation and the demonstration projects.
We help you get the first project running: we walk through the setup with you, answer questions as they come up, and help you shape a project around the work you want to try.
Everything runs on infrastructure you control. Model use is billed to your own Anthropic account at Anthropic's prices, and nothing is sent to Agiliti.
One machine, one key, one hour.
One machine
Windows or Linux, with Node.js 22A workstation or server you can set aside for the evaluation. Alchemy runs as a single self-hosted process.
An Anthropic API key
Billed directly by AnthropicYour key pays for model use directly. We never mark it up, and a spend ceiling on every task keeps the evaluation within the budget you choose.
An hour
To a first governed runEnough time to install, run a demonstration project and approve its first action.
From install to your first approval. Five steps.
- 01
Install Alchemy
Install it on your chosen machine by following the setup guide.
- 02
Run the health check
The built-in health check confirms your setup and shows which API key will pay for model use.
- 03
Open the control center
Create your administrator account, then look around the five views: Team, Work, Floor, Score and Board.
- 04
Run a demonstration project
Start a complete job, such as a code change delivered as a change record and a pull request. Watch members plan, delegate and deliver, and an independent member check every deliverable against its definition of done.
- 05
Try an approval
When the run reaches an action marked for approval, it waits for you. Approve it in the control center, then export the run's record and verify it.
Everything you need to judge it properly.
Alchemy
The runtime, the control center, approvals, spend ceilings, the verifiable run record and the governed integrations.
Complete demonstration projects
Finished jobs to run and inspect, from a code change with a pull request to an export-control release memo and an AI-use disclosure pack.
Guides and reference
Setup, writing a manifest, approvals, spend limits, integrations and the run record.
An operator's runbook
Clear procedures for running Alchemy day to day and handling anything that needs attention.
Ask for an evaluation. We'll help you start.
Tell us who would run it and what you want to try first.