Skip to content

AI that ships changes you can account for.

ChangeFabric scopes the work under your rules, lets AI agents build it and open the pull request, and keeps a full record of every run: who asked, what it cost, what was checked, and who approved it.

Request accessSee how a run works
Retry invoice export when the ERP times outPublished
1/1 tickets done · $0.72 spent of $5.00 budget · claude-sonnet-5
Rule applied
Changes to billing code need a named reviewerFrame · control record
Pull request
#212 Retry the invoice export on ERP timeoutMerged
Checks and cost
  • Governance check · 0 violationsPassed
  • Unit testsPassed
  • Read ticket, draft plan$0.09
  • Write the change$0.46
  • Run checks, open PR$0.17
  1. 1Frame ruleThe rule this change had to follow, from your project's controls.
  2. 2Weave pull requestAn agent built the change and opened the PR.
  3. 3Trace recordThe checks it passed and what each step cost.
Product view with demo data. Not a customer's run.
Difference

The same change, with and without a record

AI already writes code and scripts in most companies. What is usually missing is everything around the change.

  • Who asked for it
    AI change todayUnclear
    With ChangeFabricNamed on the ticket
  • Plan before code
    AI change todayNone
    With ChangeFabricAccepted, or rejected with a reason
  • Rules it had to follow
    AI change todayWhatever the prompt said
    With ChangeFabricYour project's control records
  • Checks before merge
    AI change todayIf someone remembered
    With ChangeFabricA governance check, beside your tests
  • What it cost
    AI change todayUnknown
    With ChangeFabricPer model call, against a budget
  • Where it lives
    AI change todayA chat log or a laptop
    With ChangeFabricA pull request and a run record
Model

Three questions, one record

Start with a Gauge assessment of where AI already runs in your business. Then each change moves through three products, and every one of them writes to the same record.

  • Frame

    Should we build this?

    Scope the work and set the rules it must follow, per project. A plan is accepted, or rejected with a reason and a remedy, before any code is written.

  • Weave

    Is it safe to ship?

    Agents build the change in a sandbox and open a pull request. Governance checks report on the PR, and it merges through your own review rules.

  • Trace

    Is it still worth running?

    Every run keeps its record: each model call, its cost and time, the checks, and who approved what.

Walkthrough

One change, start to finish

Step through a single change. The record on the right fills in as it goes.

A ticket arrives with who asked and why. ChangeFabric reads it from GitHub and never edits the fields your people own.

Asked by
Finance operations, ticket #418This step
Plan
Not yet
Agent run
Not yet
Checks
Not yet
Pull request
Not yet
Cost
Not yet
Step 1 of 6: the record so far, with demo data.
Limits

Agents act only inside limits you set

Each automation has a ceiling: what it may spend and how many runs it may start in a period, how many at once, and whether a plan must be accepted first. Where no ceiling is set, nothing runs. Absence means zero, not unlimited.

  • Spend is tracked per model call, against the run's budget
  • The model is chosen per automation, as a spending decision
  • When ChangeFabric rejects its own plan, a person decides what happens next
Spend per 24h
$20
Runs per 24h
10
At once
2
Accepted plan required
Yes
The limits panel, with demo values.
Roles

One record, read four ways

The same run answers a different question for each person who has to sign off on AI work.

  • Engineering

    CTO / Head of AI

    What is being built with AI, and who owns it once it matters?

  • Finance

    CFO / Audit

    What did each AI change cost, and against which budget?

  • Operations

    COO / Operations

    Which AI work runs the business, and who answers for it?

  • Security

    CISO / Risk

    Which change touched what, under which rule, and who approved it?

See the platform
Proof

No logos, no borrowed numbers

ChangeFabric is in a closed beta with a small group of design partners. We don't show customer logos or results we can't stand behind. What you see on this site is the product, shown with demo data and labelled as such.

How we work with design partners Security posture
Access

Request access

ChangeFabric is in a closed beta with a small group of design partners. Tell us where AI work already runs in your business.

  • A person on our team reads every request.
  • The first conversation is about the one workflow you name, not a generic demo.
  • If it fits, we scope a design-partner engagement and price it with you.

So the first call is about your work, not a generic demo.