A useful AI contract review does not start with the words “review our contracts”. It starts with a brief that tells the AI exactly what to look for and, in our case, exactly what to attack. We briefed an AI agent to act as our customer’s lawyer and redline our own Master Service Agreement. The verdict came back in one line: heavily supplier-favourable, would advise the client to negotiate. The same agent that had just spent weeks improving those contracts then found five serious risks a smart client should push back on. Here is what an adversarial AI contract review found in our own agreements, why the brief mattered more than the tool, and how to run the same check on yours.

Utilizing an AI contract review can uncover aspects that might be overlooked in traditional reviews.

Why “review our contracts” gets you nothing useful

Anyone can paste a contract into an AI tool and get something back. It will be generic, surface-level, and it will miss half of what matters. That is the version most people try, and it is why they conclude AI is not up to serious legal work.

The tools themselves are capable. Thomson Reuters reports that AI is already cutting document review from hours to minutes, and that most lawyers now expect AI to transform their profession within five years (see the Thomson Reuters analysis of how AI is transforming the legal profession). Capability is not the constraint. The brief is.

AI contract review technology employs sophisticated algorithms to analyze risks efficiently.

What good AI contract review actually requires

Our contract review worked for one reason: it was briefed properly. Before the agent touched a single document, we defined its scope. What is its authority? What can it decide and what must it escalate? What is the quality standard for a review? What legislation applies? What is the risk threshold? What does “done” look like?

An effective AI contract review can save time and increase accuracy in identifying issues.

By the time it started reading, it knew exactly what it was looking for, which framework to apply, and what standard to hold itself to. The guardrails were not constraints. They were what made the output trustworthy. The agent is not the hard part. The brief is the hard part, and the brief is only as good as the thinking behind it.

The dangerous question: what if it switched sides?

The adversarial nature of an AI contract review can provide insights from both sides of a negotiation.

Once the agent had finished improving our contracts, we had a thought. What if we asked it to attack them, not as our lawyer, but as our customer’s lawyer?

We briefed it plainly: “You are an experienced commercial lawyer advising a mid-sized company. iAhub has just sent you their Master Service Agreement. Your client is about to sign. Review this contract and tell your client exactly what to worry about.”

The results were uncomfortable.

The five risks it found in our own contracts

Completing an AI contract review can help ensure that contracts are fair and balanced.

Our agent, the same one that had just spent weeks strengthening these agreements, identified five serious risks from the customer’s side.

  • Unbalanced liability. Our liability was capped. The client’s was not. We had limited what we would pay if things went wrong, while the client’s exposure to us was effectively unlimited in several clauses. Legal commentators make the same point: courts tend to scrutinise one-sided liability caps and favour mutual clauses that reflect genuine risk allocation (a useful primer is Sirion’s guide to limitation of liability clauses).
  • No set-off rights. If a client had a legitimate dispute about an invoice, they still had to pay first and argue later.
  • A double disclaimer. A blanket warranty disclaimer sat alongside a blanket data-loss exclusion. In plain terms: we do not guarantee the work, and if your data is lost, that is not on us either.
  • Background IP defined too broadly. Our definition of the intellectual property we bring to a project was wide enough that a client could reasonably argue it captured work that should belong to them.
  • No exit strategy. If a client wanted to leave, there was no clean mechanism at the top level of the agreement.

It also flagged clauses a client should try to strike entirely, and a list of protections that were simply missing: minimum insurance requirements, service level commitments, transition assistance, and fee escalation controls.

With an AI contract review, you can proactively address gaps before they become issues.

The point is not that the contracts were illegal

They were not. They were not even unusual for a technology services company. But that is not the test.

Ultimately, conducting an AI contract review serves to enhance the integrity of your agreements.

The point is that we had never looked at our own work through our customer’s eyes. We had built contracts that protected us without stress-testing whether they were fair. Our contracts said “we will protect ourselves”. Our values said “fairness matters, partnership over extraction”. Those two things were not aligned.

We are fixing it. Not because a client complained, and not because we were sued. Because we checked. And we only got an honest answer because the agent was briefed well enough to give one. Ask it to “review our contracts” and you get a generic summary. Invest in defining the standard, the scope and the adversarial brief, and you get an AI contract review that changes how you operate. The investment was not in the AI. It was in the thinking.

Investing in an AI contract review aligns your contracts with your commitment to fairness.

How to run an AI contract review on your own agreements

Pick your most important client-facing document: your contract, your terms of service, your proposal template, whatever governs the relationship. Read it as the client seeing it for the first time, not as the person who wrote it. Then ask three questions.

  • Where is the imbalance? What protections do you have that the client does not? Is that justified, or is it just there because nobody pushed back?
  • What is missing? What would a client reasonably expect to see that is not there? Exit provisions? Service commitments? Data protections?
  • Does it reflect your values? If your company says “partnership” and “trust”, does the contract say the same thing, or does it say “we will protect ourselves and you are on your own”?

If there is a gap between what you say and what your contracts say, your clients will eventually notice. Better that you notice first. And remember the rule that makes AI safe to use here: a person reads every finding and checks it against the actual clauses before anything changes. AI does the analysis. People make the decisions.

By incorporating insights from an AI contract review, you can elevate your contract management practices.

Want us to redline yours?

This is where a tailored AI contract review can provide significant value to your business.

This is not a “send us your contract and we will run it through a chatbot” exercise. To do it properly we need context: what your business does, who your clients are, what the agreement is meant to govern, and what matters most to you. That context is what makes the output useful rather than generic.

We build this kind of adversarial review with agentic AI and intelligent automation, always with a senior human in the loop making the final calls. If you would like an honest assessment of what a smart client would challenge in your agreements before they do, book a scoping conversation and we will run the same AI contract review process on your contracts that we ran on our own.

Our approach leverages AI contract review processes to ensure thorough assessments of agreements.