Artificial intelligence tools that draft commercial contracts have become a genuine part of modern legal practice — and for good reason. They are fast, they are increasingly sophisticated, and they have made a first draft of a workable agreement accessible to businesses that previously had no practical way to get one. At Promethean Legal Services, we use AI drafting tools as part of our own practice, and we think they represent a real step forward.

But like any tool, they work best when used with knowledge of their limitations. After more than forty years in corporate and commercial law, I have come to think of AI-generated contracts the way I think of a very good first draft from a talented junior associate: it gives you an excellent structure to work from, and it raises most of the right issues. The value an experienced lawyer then adds is knowing which clauses need adjusting for this particular deal, which provisions are missing, and where the standard language doesn’t quite fit the commercial reality of the transaction in front of you.

1. Tailoring to Your Specific Deal

AI tools are trained on large bodies of existing contracts and are very good at producing well-structured, professionally formatted agreements. Where they naturally have less to work with is the specific commercial context of your deal — the relative size of the parties, the nature of the relationship, the risk profile of the transaction.

A supply agreement between two large corporations with in-house legal teams on both sides calls for different drafting choices than an agreement between a growing mid-market company and its most important vendor. The standard clauses may all be present, but the right weighting — where to push back, where to accept market terms, where a bespoke provision is genuinely needed — comes from understanding the deal, not just the template.

The value an experienced lawyer adds is knowing which clauses need adjusting for this particular deal — and where the standard language doesn’t quite fit the commercial reality in front of you.

 2. Indemnification and Liability Provisions

Indemnification and limitation of liability clauses are among the most heavily negotiated provisions in any serious commercial contract, for good reason — they determine who bears the financial consequences when something goes wrong. AI tools handle these clauses competently at a standard market level, but standard market positions are not always the right positions for every party.

A mutual indemnification clause that looks balanced on its face may warrant adjustment when one party is significantly larger or better capitalised than the other. A liability cap set at the value of the contract is a common and often reasonable starting point, but in transactions where a vendor failure could produce downstream losses well in excess of the contract value, it deserves a closer look. These are judgment calls that benefit from experienced review rather than template defaults.

3. Jurisdiction, Governing Law, and Dispute Resolution

Governing law and dispute resolution provisions are easy to overlook — they sit at the back of the agreement, they rarely come up in the commercial negotiation, and AI tools handle them with apparent confidence. But the choices embedded in these clauses have real practical consequences if a dispute arises.

AI tools will typically default to governing law and jurisdiction provisions that reflect common market practice — often Delaware or New York — without necessarily considering whether those defaults make sense for the particular parties and transaction. Similarly, mandatory arbitration clauses are inserted or omitted based on what is common in the training data, rather than on an analysis of what best serves the parties’ interests in this specific context. It is worth a moment’s attention to make sure these provisions reflect a deliberate choice.

4. Provisions That May Warrant Addition

A well-trained AI tool will produce a contract that looks complete — and in most respects it will be. But there are provisions that experienced lawyers routinely add to protect their clients that do not always make it into a standard AI draft, simply because they are more deal-specific or because they benefit one party more than the other.

Change of control provisions, assignment restrictions, audit rights, step-in rights, and termination for convenience clauses are examples. None of them are exotic — all are standard in sophisticated commercial practice. Whether they belong in a particular agreement, and how they should be structured, is a question of commercial judgment that is worth thinking through on each deal.

5. Renewal and Escalation Provisions

Automatic renewal clauses and price escalation provisions are common features of vendor agreements, and AI tools reproduce them faithfully from market templates. That is not a problem in itself — they are legitimate commercial terms. The issue is simply one of awareness: it is easy to sign an agreement with an automatic renewal and a sixty-day cancellation window without fully registering the practical implications.

The AI draft gives you an excellent structure. The experienced review ensures that structure fits your deal.

A quick review of these provisions at the time of signing — and a calendar reminder set for the appropriate notice date — is a simple discipline that avoids a common source of commercial friction. It is the kind of practical detail that gets attention when a lawyer is involved in the transaction.

6. Intellectual Property Ownership

In any agreement where a contractor or vendor will be creating work product — software, content, designs, or other deliverables — the question of who owns that output needs to be addressed clearly and explicitly. The default rules in most jurisdictions favour the creator rather than the commissioning party, which means that an agreement that is silent on IP ownership may not reflect what either party actually intended. AI tools address IP ownership and work-for-hire provisions with varying degrees of precision.

The Right Combination

The emergence of capable AI drafting tools is genuinely good news for businesses that previously had to choose between expensive law firm fees and going without proper legal documentation. The most effective approach, in our experience, is to use AI as a starting point and bring experienced legal judgment to bear on the result.