Fit Reviewby Crux DigitalRequest an audit

A Crux Digital managed service

Not less AI.
The right amount of it.

Much of every enterprise AI bill pays for work a smaller model, a script, or an overnight job would do as well. Fit Review reads the usage records you already keep, flags where AI was the wrong tool or a bigger model than needed, prices the fix, and measures what it saved.

Built for companies spending six or seven figures a month on AI APIs.

Top fixes this week · cost per yearSample
  1. 01Product's AI agent keeps working long after the task is doneFix: Cap research-agent at 29 steps per task$1.2M$147k
  2. 02Legal's tool re-sends the whole document with every questionFix: Cache the repeated context in contract-qa$590k$177k
  3. 03Data's automated job pays AI to do a spreadsheet's jobFix: Replace the LLM call in ticket-tagger with a script$280k$992
  4. 04Growth's automated job pays rush rates for overnight workFix: Move nightly-enrichment to the batch API$304k$152k
  5. 05Operations' tool answers the same question hundreds of times a dayFix: Cache shared answers in standup-digest$96k$1,297
Recoverable per year$2.2M of $7.1M AI spend (30%)

What the 2026 research says

AI budgets are growing faster than anyone can account for them.

One finding, start to finish.

From the sample report: a sales team's email tool, from the first signal to savings confirmed after the fix.

  1. 1 · What we saw

    Sales' email tool sent 1.2M short rewrites a month to a premium AI model.

  2. 2 · Why it's waste
    Premium model, simple task
    Oversized model · confident

    Polishing a short email doesn't need the most capable model. A smaller one does it for about 1/50 the price.

  3. 3 · The fix
    # litellm_config.yamlmodel_group_alias:  claude-opus-4-8: email-rewriter-default

    Recommended: one routing change the platform team can make centrally. The sales team's tool works exactly as before.

  4. 4 · The result
    $104k $2,086
    per year · measured after it shipped

    $102k a year saved, confirmed from the next two weeks of usage records.

The missing question

Your tools show how much you spend.
None show whether you should.

Gateways route traffic. FinOps tools allocate spend. Both are good at what they do, and Fit Review replaces neither. It sits on top and judges the work itself.

Questions answered by gateways, FinOps tools, and Fit Review
QuestionAI gatewayFinOps toolFit Review
How much did we spend, and on which model?●●uses yours
Which team or key spent it?●●uses yours
Should this work have used AI at all?––●
Was it the right size of model, called the right way?––●
What would the better approach cost?––●
Did the fix actually save money?––●

How it works

Your AI usage records go in.
A ranked list of fixes comes out.

Every day, Fit Review reads the usage records your AI systems already keep. It looks for patterns across thousands of requests rather than judging single ones, because waste shows up as a habit: the same job, tool, or team, day after day.

What Fit Review does with your usage records

  1. 01

    Sort usage into workloads

    Millions of individual requests become a list you recognize: each app, agent, scheduled job, and team.

  2. 02

    Spot the waste

    Each workload is checked against eight known waste patterns, with the evidence and how certain the finding is.

  3. 03

    Put a price on the fix

    Your actual usage at current list prices, compared with what the better approach would cost.

  4. 04

    Check before anyone acts

    Uncertain findings are held for a person to review. Nothing is shared until your team decides.

What each team gets

  • AI platform ownerRanked list of findingsEvery waste finding in one place, ordered by dollars saved, each with an owner and a status. You know what to fix first.
  • EngineeringA specific fix for each findingThe exact change to make: a routing rule, cache setting, retry policy, batch job, or script outline. No research needed.
  • Team leadsA short summary for their teamTheir team's top findings and suggested fixes, ready to share in Slack or email when you choose.
  • FinanceA savings reportProjected and realized savings side by side, per fix and per team, ready for budget reviews and FinOps tools.

The pattern catalog

Eight ways AI spend goes wrong.

Each one is a plain business problem with a measurable fix. Underneath, each is a precise rule over your usage records.

  1. 01Wrong tool

    Paying AI to do a spreadsheet's job

    Routine, rule-based work like tagging tickets or reformatting records is sent to an AI model thousands of times a day.

    A simple script handles most of it; AI only sees the cases that need judgment.

    Deterministic work on an LLM: thousands of near-identical prompts with structured input and short structured output

  2. 02Wrong size

    The most expensive model on the simplest tasks

    Quick rewrites and yes/no checks go to a premium model when a smaller one gives the same answer for a fraction of the price.

    Default to the right-sized model; keep the premium one for hard cases.

    Oversized model: top-tier model on short prompts with short, simple outputs

  3. 03Wasteful approach

    Re-sending the whole document with every question

    Every question about a contract or report re-sends the entire document, and you pay for all of it every time.

    Send the document once and reuse it, or send only the relevant pages.

    Context stuffing: large, mostly repeated input on every call

  4. 04Wasteful approach

    AI agents that don't know when to stop

    Most automated tasks finish in a few steps, but some loop for hundreds, repeating the same work and running up the bill.

    Put a step limit and a clear finish line on every agent.

    Runaway agent loop: one task with 100+ calls, repeated tool calls, or rising context per turn

  5. 05Wasteful approach

    Paying rush rates for overnight work

    Large scheduled jobs use the instant, full-price service even though nobody is waiting on the results.

    Run them through the overnight batch service at half the price.

    Real-time bulk job: large volume sent in a burst, no user waiting on results

  6. 06Bug

    Paying again and again for the same failure

    When a request fails, a system immediately retries it, often several times, and each failed attempt is billed.

    Retry patiently and fix the underlying error.

    Retry storm: same request repeated after errors or timeouts

  7. 07Wrong tool

    Asking a chatbot for facts it doesn't have

    Employees ask a general AI about internal systems and policies it was never connected to: costly, and often wrong.

    Point those questions at the company's own search or system of record.

    Lookup with no source: factual q&a with no retrieval or tool use

  8. 08Wasteful approach

    The same question, answered hundreds of times

    Many people and systems ask for the same summary or answer each day, and each one is generated and paid for from scratch.

    Generate it once and share it.

    Duplicate work: same or near-same prompt run by many people or jobs

Proof, not promises

Projected savings don't count.

The headline number is realized savings: what shipped fixes actually saved, measured from your logs after the ship date. We compare cost per unit of work before and after, so usage growth can't hide savings or fake them.

Realized savings
5–10× the product's cost in year one
Findings accepted
Over 60%
Time from finding to fix
Under 30 days
AI adoption
Flat or rising. Always.

Privacy, trust, and tone

Built so nobody moves to a personal account.

If people feel watched, they route around you and you lose visibility entirely. Every default is the least invasive one.

  • Metadata first

    Token counts, timing, model, and call structure. Prompt text stays opt-in per deployment.

  • Hashed, then dropped

    When an export carries message bodies, they're hashed on import and never stored.

  • Workloads, not people

    Findings attach to apps and teams. Individual attribution stays with the platform team, only to route a fix.

  • No leaderboards

    No per-person scores. Nothing in the product ranks employees.

  • You decide what's shared

    Findings stay with your team. Nothing reaches other teams or employees unless you choose to share it.

  • A tone rule

    “This job could run as a batch,” never “you're wasting money.” Every output follows it.

Getting started

Start with an audit. Keep what works.

Crux Digital engineers run the first review on your own logs, with your team. After that we keep watching every week, and the fixes stay yours.

Step one

Assisted audit

4 to 6 weeks

  • 30 to 90 days of gateway and provider logs
  • Findings report ranked by recoverable dollars
  • A fix pack for each finding, ready for your gateway
  • A read-out with your platform and finance leads
Request an audit

Step two

Managed Fit Review

Ongoing, every week

  • Weekly findings, checked by a Crux engineer before your team sees them
  • Status tracking and realized savings, measured
  • Team summaries and training material from real cases
  • Your team's own login to the findings queue
  • Priced as a monthly fee, a share of realized savings, or both
Try it on sample data

Request an audit

Tell us where your AI traffic goes today and we'll reply with exactly what we'd need from your logs.

Questions

Common questions.

Who runs it?

Crux Digital, a Denver software and AI engineering firm. Crux engineers run your audit, check each finding before your team sees it, and help ship the fixes if you want a hand. The Fit Review platform is the tool we use to do it, and your team gets its own login.

Does Fit Review sit in the request path?

No. It reads logs your gateway or provider already keeps, once a day. It recommends changes; your team decides whether and how to make them. Nothing about how traffic is routed changes to use it.

Do you read our prompts?

Not by default. Most patterns show up in token counts, timing, model, and call structure. When a log export includes message bodies, they are hashed on import (to match retries and duplicates) and dropped. Prompt classification is an opt-in for later, per deployment.

Which gateways and providers do you read?

Today: LiteLLM spend logs, Portkey log exports, the OpenAI Usage API, and any gateway that can export one row per request as CSV or JSON Lines. If you don't have a gateway yet, setting one up can be part of onboarding.

Won't this make people use AI less?

It shouldn't, and we measure it. AI adoption is a guardrail metric: calls and active users should hold or rise while cost per unit of work falls. If total usage drops, the product is doing something wrong.

What happens when a finding is wrong?

Nothing is acted on or shared without your team. Every finding comes with a confidence level and the evidence behind it, and low-confidence ones are flagged for a closer look. Your team decides what's accurate, what to fix, and who hears about it.

Find out what your AI should cost.