← All tech signals
Infrastructure

Data Under Contract

Schema drift used to be a nuisance a person caught eventually, but an AI agent propagates it instantly, which is turning data contracts into a merge-blocking requirement.

What is it?

A data contract is a machine-readable agreement about what a dataset should look like — schema, freshness, quality — built so an AI agent can query it, because a person once caught drift eventually but an agent propagates it instantly. Whether writing one means anyone enforces it stays open.

Why now?

AI agents remove the human checkpoint that used to catch a broken schema before it spread. Feed an agent instead of a person, and an unenforced contract stops being documentation and becomes a live failure inherited downstream. The force is agents acting faster than anyone can verify.

Who’s building it?

Platform vendors and open-source spec maintainers, wiring enforcement into the pipelines engineering teams already run.

Contract vendorsOpen-source specsEngineering pipelinesCompliance teamsAI agents
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — vendors like Atlan, plus open-source contract specification maintainers.
Buyers — engineering teams shipping AI pipelines, and the compliance functions now watching them.
Gatekeepers — CI systems that block a merge when a contract breaks.
What’s at stake
Contracts only earn their audit-trail value where enforcement is real — written but unchecked contracts pay compliance nothing.
Vendors compete to own the enforcement layer itself.
Regulation could make enforced contracts a legal requirement.
What’s unsettled
Whether merge-blocking rules actually change behaviour, or just get bypassed.
Who is liable when an enforced contract still ships bad data.
Whether a contract works as an audit trail before regulators demand one.
02 · How the demand matureda progression, not a timeline
Documentation
Teams write contracts by hand, mostly as internal reference, rarely checked automatically.
Enforcement
Open-source tooling wires contracts into CI, blocking a merge on breach.
Compliance layer
AI-governance rules start treating an enforced contract as evidence a pipeline met its own standard.
Standard practice
Contracts move from optional documentation to a default expectation across engineering teams.
03 · Where the market goesthe forks, not a prediction
→Enforcement deepens or stays symbolic — either merge-blocking tooling becomes standard practice everywhere, or most contracts stay written but never actually checked.
→Compliance regulation could force enforcement everyone currently treats as optional, changing the incentive overnight.
→Value could concentrate in a handful of enforcement vendors, or stay fragmented across many open specifications.
→Liability could land on whoever wrote the contract, or whoever skipped checking it.
The enforcement rails are maturing regardless; the unsettled part is whether writing a contract without ever checking it still counts as real governance.
04 · The AI Dev Process deskadjacent markets, moving now
Synthetic Data
The raw material AI training needs.
Public training text may run out within 6 years.
South China Morning Post (citing Epoch AI)
Keeping the Model Honest
Tooling that keeps watching a model after launch.
An observability platform absorbed.
InfoWorld
Tracing the Reasoning
Watching what an agent actually decided to do.
A platform to govern agents.
TechCrunch
05 · The Wire
01
Market path to $3.64B.
Research and Markets ↗
02
Contracts agents can query.
Atlan ↗
03
Open-data-contract-standard 3.2.0 adds context for AI systems
Bitol (open-data-contract-standard release notes) ↗
04
Consultant says data contracts can demonstrate AI compliance
TechTarget ↗
The panel on this signal
They agreed on
–Contracts are shifting from optional documentation into enforced, merge-blocking infrastructure that AI pipelines now depend on.
–Agents specifically need machine-readable definitions they can query directly.
–Value is concentrating in the vendors who build the enforcement and observability tooling.
The ML systems engineer disagreed on
–A schema agreement only functions as an audit trail if violations are automatically blocked or logged; a written contract without enforced validation checks in the pipeline provides no more assurance than the informal norms it replaced.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.