Manage AI
token spend.
See what your AI token spend bought, team by team, and route each task to the cheapest model that clears your quality bar.
Trusted by engineering teams
1,284 tasks routed · 2 held at frontier
- Haiku 4.5Mechanical, fully specified$0.02−94%
Scaffold CRUD endpoints
navigaracom/vision · ENG-2917
- Sonnet 5Touches security, needs care$0.28−61%
Refactor auth middleware
navigaracom/vision · ENG-2884
- Opus 5High blast radius, no downgrade$1.42
Payments architecture spike
navigaracom/billing · ENG-2902
- Haiku 4.5Pattern repetition$0.03−91%
Backfill unit tests
navigaracom/identity · ENG-2871
- Opus 5Ambiguous, spans four services$1.18
Incident root cause, checkout
navigaracom/vision · INC-441
Chameleon picks the model. You teach it what best means.
Token spend optimization starts with the task. Chameleon reads the task and its blast radius, then routes it to a cheaper model only when your policy allows. It logs every decision and its cost against your AI token spend.
Start with the bills by team. Then test token spend optimization on real tasks, inside the quality rules you set.
Scaffold CRUD endpoints
navigaracom/vision · ENG-2917
- Haiku 4.592$0.02
- Sonnet 597$0.11
- Opus 599$0.98
Haiku 4.5 at $0.02. 98% cheaper than defaulting to the frontier model.
Per-task, not per-seat
The unit of routing is a task with a known shape, not a developer with a licence.
Trainable, not fixed
Every accept, revert and rollback is a label. Your policy diverges from our default on purpose.
Refusal is a feature
Work with real blast radius stays on the frontier model, and the router records why it declined to save money.
The default policy is ours. The one you run is yours.
Every codebase has its own idea of which tasks are dangerous, so the policy is trainable rather than a vendor default.
Decide migrations always run on the frontier model and that becomes policy. The router argues with the bill, never with you.
Token spend optimization needs delivery context.
Two questions sit behind every AI token spend review: which models you run, and what work the bill paid for. Chameleon answers the first. Token spend optimization needs the second too, because a cheaper model running work nobody asked for is still waste.
Compare token spend with engineering throughput.
Divide AI token spend by merged pull requests and the cost per unit drops 65%. Divide it by graded throughput, and it drops 38%. ETV scores each code change in repository context, so token spend optimization is measured against work that met your quality bar.
- ActivityPull requests mergedWhat the rest of the category divides byDenominator74210 PRs+184% growth$3,066$1,080per pull requestOverstates the gain
- Graded valueETV, scored per commitA language model reads each diff and grades what it was worthDenominator46.675.4 ETV+62% growth$4,869$3,009per ETVThe honest number
Review AI token spend alongside graded throughput and work mix. Token spend optimization counts only if it preserves useful delivery.
Compare AI token spend with ETV, a per-commit score derived from a language model reading the diff in repository context.
Review AI spend by team and objective.
Break AI token spend down by team and objective, then read features, maintenance, and fixes separately. A larger bill can mean more delivery. Token spend optimization evidence lets you show which before the budget meeting.
Cost per ETV down 38%
Absolute spend rose over the same period. What fell is the price of a unit of graded output, which is the number that settles the argument.
- Features$9,48552%Shipped against a roadmap objective
- Maintenance$4,01322%Keeping what exists running
- Tests$2,18912%Coverage written alongside the change
- Docs$9125%Runbooks and API reference
- Fixes$1,6419%Bugs, regressions, edge cases
At today’s rate, another $40,000 a quarter buys roughly 165 ETV of additional shipped work.
Follow a dollar from the initiative to the commit.
Spend flows from an initiative down to a commit, and at every link it either still answers to the mandate above it or it does not. The band at the bottom left is the 21% with no issue behind it.
Green clears the mandate above it. Amber is questionable. Red has no mandate behind it. Band width is cost.
Export AI token spend by team, with the billing and commit evidence attached, so finance sees the cost behind each token spend optimization decision.
Token spend optimization starts with how your team works with AI.
The usual answer is one maturity score that tells nobody what to do on Monday. We score five dimensions from evidence, compare each against your org median, and name the one gap worth closing next.
Five independent levels for how this team works with AI agents. Each level comes from its own evidence, so the levels are never added together.
Ahead of the median on 3 of 5 dimensions. Largest lead: Delegation depth (+1.8). Largest gap: Context leverage (−1.5).
- Agent fluency4.4AI-active 7 of 8 weeks
- Delegation depth3.9Deep delegation on 21 of 28 agent days
- Roadmap alignment3.572% of work credited to roadmap objectives
- Context leverage0.83 of 12 artifacts state a why and outcome
- Agentic autonomy0.52% of work owned end-to-end by agents
- Require a why and an expected outcome on every agent task, not just a title
- Attach the ticket and the failing test to the prompt so agents stop rediscovering context
- Promote the three artifacts that already do this into templates for the rest of the team
The levels are never added together. A team that delegates deeply but writes no context has a specific, fixable problem, and an average hides it.
Token spend optimization in budget reviews.
Answer the CFO’s question about the AI line item with a cost per unit of shipped work, not a per-seat count
Walk into the budget review knowing what another $40,000 a quarter actually buys
Show which teams turn tokens into roadmap work and which turn them into waste
Give a team one named thing to improve next quarter, with the evidence behind it
AI token spend questions.
What data does an AI token spend review use?
Your provider bills, your repositories, and your tracker. Setup checks which bills are in scope, because token spend optimization built on half the bills can mistake missing data for savings.
Does reviewing AI token spend require routing?
No. Measuring AI token spend reads work and billing data only. Chameleon is a separate opt-in that sits in the request path and uses your own provider accounts. Test token spend optimization on a sample of real tasks before you trust a cheaper model with more.
How is token spend optimization different from a budget cap?
A cap stops spending at a number. Token spend optimization checks whether the AI token spend produced useful throughput and whether a cheaper model can hold the quality bar. Adaptive limits then apply the rules you choose.
How do we compare AI token spend across teams?
By team, work type, and recorded objective, over the same period. The team running payments migrations will cost more per task than the team writing docs. Token spend optimization starts with that context, so the team with the largest AI token spend gets a conversation before any budget cut.
Can token spend optimization account for maintenance?
Yes. Maintenance and fixes are reported alongside features, so necessary maintenance is visible. Navigara links each slice of AI token spend to the work behind it, so the team can explain it.
Where should we start with AI token spend?
Pick one month and list the tools and bills behind your AI token spend. Check a sample of linked work, then make one token spend optimization decision and measure the next month against it. Read the token spend guide.
See what your tokens
actually bought.
Start token spend optimization with read-only measurement of your repositories, boards and provider bills. Routing is the one place Navigara sits in the request path, and it is opt-in, per team.
Your keys, your providers
Chameleon routes through your own provider accounts. No resale, no markup.
Nothing retained
Prompts and completions are not stored once a routing decision is made.
Every decision logged
The model chosen, the reason, the cost, and the cheaper option it rejected.