Your corporate terms, checked on every contract

Every procurement team carries a list it cannot let slip. Liability caps sized to the contract value. A data processing addendum on anything that touches personal data. Termination for convenience with a notice window you can live with. Audit rights, price protection, an assignment clause that survives an acquisition. The list is real, it is written down somewhere, and yet it gets checked by whoever happens to read the paper that week, against whatever version of the standard lives in their head. The gap between the terms you require and the terms you actually verify is where risk hides. We shipped a feature that closes it, and this post explains what changes for the buyer.

The named problem: your standard is a memory, not a check

Ask five people on a procurement team what the required liability cap is and you get four answers and one shrug. The corporate standard exists, but it lives in a playbook PDF, a legal wiki, and the recollection of the person who negotiated the last big deal. When a contract arrives, the reader compares it against that fuzzy memory, and the comparison is only as good as their attention on a Thursday afternoon. Terms that should be non negotiable slip through because nobody held the paper against the standard line by line. As we argued in the answers sit on the paper, the fix is not more meetings, it is reading the actual clause against the actual requirement, every time, without relying on who is in the room.

The new feature turns your standard into something the Contract Agent applies. You choose the terms your organization requires, either from the platform's list or written in your own words. You say which kinds of paper each term should be read on, so a data processing requirement fires on anything that touches personal data and not on a simple order form. From then on, every contract the agent reads is checked against your terms the moment the read completes.

Four states, a passage, and the language to ask for

Every required term lands in one of four states. Met means the contract carries the term as you specified it. Partly met means the term is present but weaker than your standard, a cap that exists but sits too low, a notice window that is shorter than you require. Missing means the clause is absent. Unclear means the language is ambiguous enough that a reasonable reader could argue both ways, which is itself a finding worth flagging. Each state comes with the passage it rests on, so you are never trusting a label alone, you can read the exact sentence the agent scored against. And where a term is partly met or missing, the agent supplies the language to ask for, so the gap ships with its own redline instead of a note that says fix this.

That last part matters for the flow into negotiation. When the term check and the redline speak the same language, the handoff from review to counterparty is short. The head of the redline itself now shows how many of your required terms are met, so before you send anything you know the shape of the ask. This connects directly to the work in the AI redline, where the challenge to vendor paper is written in its own margins.

The rate ladder now reads against your own MSA

On a statement of work, terms are only half the risk. The other half is money. A SOW can honor every clause in your MSA and still price the engagement above your contracted rates, because the person building the SOW quoted from a rate sheet nobody checked against the master agreement. The rate ladder now reads each SOW line against your own MSA rate card. It shows the money over the engagement and applies a cap at your contracted rates, so a senior engineer billed above the rate you negotiated shows up as an overage, not a surprise on the third invoice. This is the same discipline we described in invoice intelligence, moved earlier, to the SOW stage where you can still change the number before it becomes a commitment.

How it sits in the workflow

Nothing here asks you to change how contracts arrive. You still drop paper into the platform, or route it through your normal intake, and the Contract Agent reads it the way it always has. The term check is not a separate task you remember to run, it fires automatically on every read, so a contract that lands overnight is scored by the time you open it. You can point the terms at specific paper types, so an NDA is not scored against a rate card requirement and a SOW is not scored against a data processing addendum unless it needs one. The result is a review that starts from a scored position rather than a blank read, which is the difference between checking work and doing work. If you want the fuller picture of how a read becomes a workspace beside the clause, see decode any contract and the Contract Agent tooling.

Honest limits

The feature reads the paper you give it. It does not know a term you never told it to require, and it cannot score a requirement you have not written down. If your standard is genuinely ambiguous, the agent will often mark it unclear, which is honest, but it means the quality of the check tracks the quality of your term definitions. Spend time on the terms and the output sharpens. The four state scoring is a reading, not a legal opinion, and unclear is a real answer that means a person should look, not a failure to decide. On the rate side, the cap and overage are only as accurate as the MSA rate card you loaded, so a stale rate card produces stale math. And the suggested language is a starting position for negotiation, not a guarantee the counterparty accepts it. What the feature removes is the risk that a required term slips through unread. What it does not remove is your judgment on the terms that sit at the edge. Use it to clear the clear cases fast, so your attention lands on the ones that actually need it.

Want to be updated when major licensing and pricing changes land? One analyst brief a week: the price rises, metric changes and audit campaigns that move software costs. Work email only.