Map your AI spend to your roadmap — in two weeks.
A focused, read-only pilot: connect one tracker and your AI tools, and we'll show you exactly what your AI compute costs per ticket, epic, and team — with a budget forecast and guardrails — without a single prompt leaving your environment. Free to get started.
A fit if this sounds like you.
You'll get a real answer in two weeks.
- ✓An eng org of ~20–150 shipping real product.
- ✓You use Jira, Linear, or GitHub Issues with reasonable PR/ticket hygiene.
- ✓You run Claude Code and/or Cursor daily — and the AI bill is now business-visible.
- ✓Someone — a VP/Director of Eng or a business/FinOps lead — feels the pain.
Probably too early if…
- ·You're a solo founder or very small team.
- ·You don't track work in a ticketing system.
- ·Your AI spend is still trivial — there's nothing to govern yet.
Not sure? Ask us — a 10-minute look at your setup tells us fast.
What you connect, what you get back.
Read-only and metadata-only. Outlay runs on categories, token counts, and ticket IDs — prompt content and keys never leave your environment.
~30 minutes of setup.
- →One tracker — GitHub Issues, Jira, or Linear.
- →Your AI usage — Claude Code transcripts, Cursor admin export, and/or the Anthropic admin API.
- →That's it. No app rewrite, no proxy required to see the picture.
Your spend, on the roadmap.
- ✓Spend attributed to tickets, epics, and teams — with a coverage number.
- ✓A quarter forecast from your open scope + a budget burndown.
- ✓Anomaly flags on outlier tickets, and a measured forecast-accuracy read on your data.
Two to three weeks, four steps.
Kickoff
20-minute call to confirm fit, pick the tracker + usage source, and agree what "good" looks like.
Connect
Read-only access; we ingest the metadata and resolve spend to your tickets. ~30 min on your side.
Review
We walk you through your real numbers — coverage, forecast, burndown, and anomalies.
Decide
Keep going (set budgets and roll Outlay out wider) or part as friends. Your data, deleted on request.
We agree the bar up front.
The practical questions.
What does it cost?
Nothing — design-partner pilots are free. In exchange we ask for candid feedback, and (if it lands) a reference. You also get influence over the roadmap and early pricing.
What access do you need?
Read-only: one tracker (GitHub/Jira/Linear) and a way to read your AI usage (Claude Code transcripts, Cursor admin export, or the Anthropic admin API). No write access, no app changes.
Do prompts or customer data leave our environment?
No. Outlay runs on metadata only — task categories, token counts, ticket IDs. Prompt text, model outputs, and API keys never reach us, and our endpoints reject any payload containing them.
What if our branches aren't named after tickets?
Then coverage from branch inference alone is low — which is exactly why Outlay also resolves the ticket from PR→issue links and an explicit one-line task tag (works even for remote/CI agents). We'll tell you your real coverage number in week one either way.
What happens to our data after the pilot?
Deleted on request at the end. Per-deployment isolation throughout; you can leave anytime.
Two weeks to a real number.
Map your AI spend to your roadmap and see exactly where it's going — and where it's about to go over.
Read-only · prompts never leave your environment · free while you get started