Three ways to work with us more deeply.
Most teams get value from the self-serve flow on the landing page. A few want something more structured — early access to capabilities we're testing, a co-design seat at the product table, or a tailored evaluation suite for a production-grade rollout. These are the three programs we run.
Research preview
Early access to capabilities we're testing — surrogate-gradient prioritisation, the optimization plan, prompt-cache instrumentation, new judges — before they hit the main product. In exchange, you give us honest feedback on what worked and what didn't.
- Feature-flagged access weeks before public release.
- A direct line to the engineer who shipped it (one shared Slack channel).
- Cost capped at the standard wallet limit so a preview can't blow your budget.
- Free to join. You can leave any time.
Good fit if: you run cycles weekly, you're comfortable with rough edges, and a 30-minute follow-up call once a month feels useful, not annoying.
Design partner
A smaller cohort that shapes what we build next. Design partners get the research preview plus a seat at the product table — quarterly review of the roadmap, dedicated time with the product team, and the option to co-fund work that unblocks your specific use case.
- Quarterly roadmap review with the founders.
- Named contact for product input — you write the brief, we ship it.
- Early input on pricing, packaging, and contract terms.
- Co-funding option for work that requires bespoke capability.
Good fit if: your team has unusual or vertical-specific requirements (regulated industries, on-prem hosting, custom evaluation criteria) and you'd rather shape the product than file feature requests.
Enterprise pilot
Structured trial for serious deployments: extended-trust sandboxes (longer-lived, larger budgets), a tailored evaluation suite for your domain, security review on our side, and a deployment plan negotiated before any change hits your production agent.
- Up-front threat model + security review (we ship the artefacts your procurement team needs).
- Custom evaluation suite calibrated to your domain (we co-design the scenarios).
- Higher per-cycle budgets and parallel sandboxes for serious throughput.
- DPA, subprocessor briefing, and SOC 2 readiness packet on request.
- 30/60/90-day success plan with concrete checkpoints.
Good fit if: the agent you're improving runs at production scale, procurement is involved, and the cost of regression is materially higher than the cost of caution.
How to apply
Email partners@agentetna.com with which program you're interested in, a one-paragraph description of the agent you'd bring, and a sentence on what "success" would look like for your team. We aim to reply within two business days; if it's a fit, we'll get on a 30-minute call from there.
Not sure which program fits?
Start with self-serve — connect a repo and run a cycle. We're happy to recommend a program once we see what you're working on.