Your AI bill,
itemized.
Outlay maps every AI dollar to the ticket, team, and person that drove it — so you can forecast it, budget it, and stop the spend that isn't work. Prompts and keys never leave your environment.
Works with Anthropic · OpenAI · Azure OpenAI · AWS Bedrock · Google Vertex · Claude Code · Cursor · Jira · Linear · GitHub
Attribute. Forecast. Govern.
Connect the tracker and AI tools you already use — read-only, no app rewrite. Every other tool stops at an infrastructure tag. Outlay meters the work.
Every dollar finds its ticket.
Each AI call resolves to the ticket, epic, and person behind it — from branch, PR, and session metadata. Prompt content is never read.
fix/PROJ-123 → PROJ-123 → Q3 Stability
fidelity: call · branch · session · team
Price the quarter before you build it.
Forecast from your open backlog, the way you estimate effort — back-tested on your own closed tickets, with the error shown. Never a vendor benchmark.
Q3 forecast $48k ± $9k
back-test on 188 closed tickets · error shown
Budgets that act, not just alert.
Set budgets per team, program, or epic. Pace-based alerts flag overspend before it lands — or set a hard cap the opt-in gateway enforces, blocking or routing down.
Platform $61k / $50k · over → route-down ×1,240
Launch Q3 $28k / $40k · on track
Decisions no dashboard can make.
Budget by what matters.
Group teams and projects into programs with one budget. Over the cap? New calls block or route down automatically — and budget moves where priorities move.
See governance →Work vs. non-work.
Tag a side-project key once and see what the company key is really buying. Stop non-work usage per team — without reading a single prompt.
How the split works →Buy compute the cheap way.
Per-person spend shows who belongs on a flat seat and who belongs on API credits. Outlay computes the cheapest mix — analysis nobody buying blind can run.
Procurement mix →Built to pass yours.
Most spend tools see everything you send. Outlay is built so sensitive data physically can't reach us — attribution runs on metadata: task categories, token counts, ticket IDs, dollar figures.
SSO/SAML & SCIM · phishing-resistant passkeys + MFA · full audit log with SIEM export · configurable retention & self-serve erasure · WCAG 2.1 AA · read the security overview
Read-only to start. No app rewrite.
Put AI compute
on a budget.
Start with a free read-only pilot: connect one tracker and your AI usage, and see your real spend itemized in days.
Become a customer →How does Outlay attribute spend to a ticket?
It resolves the work context of each AI call from the most reliable signal available — an explicit task tag from your agent launcher or CI, the git branch, the PR's closing-issue link, or a commit trailer — and maps it to the ticket, epic, and roadmap in Jira, Linear, or GitHub. Where teams already link work to tickets it's automatic; where they don't, a one-line tag makes it reliable, even for remote/CI agents.
Do you see my prompts or my customers' data?
No. Outlay connects with read-only tokens and reads metadata — task category, token counts, ticket IDs, and your provider's usage data. Prompt content, model outputs, and your API keys never reach our servers. That's an architecture, not a policy.
Are budgets hard caps, or just alerts?
Your choice, per program. By default they're pace-based projections that flag a scope trending over — Outlay never touches your traffic. For a true hard cap, the opt-in gateway blocks or routes over-budget calls down to a cheaper model. It fails open: our downtime can never block your traffic, only a budget you set being exceeded can.
How accurate is the forecast?
We back-test it on your own closed tickets — hide one, predict it from the rest, compare to what it actually cost — and show you the measured error and sample size by work type. As more work closes, the number sharpens. We never quote a vendor benchmark.