Your hardware
One workstation, server, or lab machine is used for the pilot.
10-business-day private-agent evaluation
OpenJet Private Agent Pilot
Evaluate OpenJet on one real repository and one machine you control. We agree 8–12 representative tasks, run them under your constraints, and deliver evidence for a go, no-go, or targeted follow-up decision.
Boundary: this is not production enterprise software yet. It is a paid pilot for evaluating OpenJet on controlled hardware before a wider rollout.
Starter pilot: £2,500. First 3 pilots only.
Fixed-scope private-agent deployment
The starter pilot is scoped around one machine, one repository, and 8–12 agreed tasks. The outcome is a working OpenJet setup plus task-level evidence and a decision report, not a production enterprise deployment.
Pilot scope
Your repo
One repository is evaluated against 8–12 representative tasks and agreed pass criteria.
Your model path
A local or self-hosted model runtime is selected, configured, and documented for the pilot machine.
Decision report
The pilot ends with task-level evidence, observed failure modes, configuration notes, and a go/no-go recommendation.
Paid pilot
OpenJet Starter Pilot
A paid pilot for teams that want to evaluate OpenJet on private code and controlled hardware before planning a wider rollout.
£2,500. First 3 pilots only.
50% upfront, 50% on handover.
Book a paid pilotOne customer-controlled machine used for the pilot
8–12 representative tasks and pass criteria agreed
Local or self-hosted model/runtime path selected
Task-level completion, intervention, time, and failure evidence
Approval-gated command flow demonstrated
Decision report, runbook, and findings readout
30 days of business-hours support for the delivered pilot setup
What you provide
A machine or server to run on
One repo or workflow to test against
A named technical contact
Access for setup/debugging
Clear acceptance criteria for what working means
Not included
Not production enterprise software yet
No production SLA
No custom model training
No guaranteed model quality beyond the agreed pilot tasks
No autonomous production actions
No broad enterprise rollout in the starter pilot
Who this is for
AI consultancies
Private-AI consultancies that need a working agent layer for client pilots.
Local-AI teams
Teams already running self-hosted models that need a practical coding-agent workflow.
Labs
Labs that need coding-agent evaluation on controlled machines and internal code.
Jetson/GPU workstation owners
Owners of Jetson boxes, GPU workstations, or local AI PCs that need a usable developer workflow.
Privacy-sensitive small dev teams
Small engineering teams evaluating local coding agents for private repositories and sensitive client work.
Controlled local setup
On the local path, repository access, prompts, tool output, approvals, logs, and checkpoints stay on the controlled machine.
runs on customer-controlled hardware
uses a local or self-hosted model runtime
requires approval before shell commands run
keeps logs, traces, and diagnostics available for handover
documents the hardware/model configuration used during the pilot
includes business-hours support for the pilot period
OpenJet core is AGPL-3.0-only. Commercial pilots are delivered under a separate written services/support agreement.
Deploy OpenJet on your hardware and repo.
Starter pilot: one machine, one repo or workflow, one handover.