5 Things Ann Arbor Companies Should Do When AI Slows a Deal

The hardest part of a first large deal often comes after the customer says yes. They want the product they saw, but the contract comes back from their legal team with a clause forbidding the vendor from using AI to deliver the service. That can happen whether AI is the whole product or one feature inside it.

Joe Morrison spends his days helping clients with problems like this. He is a corporate attorney who worked as an IT consultant before law school and has practiced technology and venture capital law for more than a decade, representing both investors and startups, including Ann Arbor SPARK companies. He calls the work a continuing education, because no two clients use or commercialize AI the same way.

Below are five things any company using AI can do to get a deal through a customer's legal review, drawn from the contracts Morrison deals with on a daily basis.

 

1. Plan for the customer's legal team to start at no

Assume the first contract you get back will prohibit you from using AI at all, and build that into your sales timeline. Morrison says this is the default position that many large companies take in their form contracts, and it has nothing to do with your product. Customers, including Fortune 500 buyers, generally do not have contracts that account for the bespoke ways vendors use AI or commercialize technology that integrates AI, so their starting position is to forbid vendor AI use outright.

That leaves founders selling a product that leverages AI to a business team that wants it while the customer's own counsel writes the AI out of the contract. "An in-house lawyer's job is not to do a 'good business' deal. An in-house lawyer's job is just to mitigate risk and protect their client or employer," Morrison says.

Morrison reads this as a repeat of an argument the industry already worked through once. "It's very similar to 10 years ago when everybody said everything has to be on prem because they didn't trust the security or functionality of cloud hosting and architecture. Then we got over that. We're still in the early stage right now where everybody's saying no to AI use by vendors."

For founders, the deal is usually still winnable. Morrison counsels his clients not to sign contracts with clauses forbidding AI use if AI is a core part of their technology, and in his experience, the customer drops or revises those terms once the company explains how it uses AI and how its technology actually works. But it costs time. Founders often build a sales cycle with no room for several weeks of back-and-forth with a legal department. Planning for it from the start takes some pressure off.

 

2. Build education into your sales process

Don't wait for the contract to explain your technology. Morrison tells clients to treat educating the customer's legal team as part of the sale, starting with the first sales conversation.

Your sales material has to work twice. "Your sales-y brochure about how great your platform is also needs to be paired with education about how you do what you do at a relatively granular level, because we're going to take that from the sales process and then use it in the legal process," Morrison says.

Done well, the material that convinced the business team is the material their lawyers read first. Legal review then starts from a specific description of how your product handles data, instead of a general assumption that AI is a risk to be written out.

 

3. Put your salesperson, your attorney, and your data owner in the same room

Review contracts with three people present: the attorney, the salesperson, and the person who actually owns the data at your company. One of Morrison's clients has built this into how it sells. Everyone involved in the deal works from the same tight timeline, which cuts down on internal back-and-forth.

Including everyone from the start makes it easier to get on the same page. Salespeople often don't know the specifics of how their own company governs data, and the contract will be written to match whatever was sold. "The data folks and their chief data officer will say, 'That's not how we do that,'" Morrison says. When the data owner is in the room, those questions get answered, and the contract gets corrected before signing.

4. Write down how you handle data before a customer asks

Build a clear, technical account of how your product handles customer data and have it ready before contract review starts. Morrison says scrutiny lands unevenly, and that smaller vendors absorb more of it than the large platforms do.

"Large, best-in-class providers like Anthropic, Microsoft, Google, are often totally fine," he says. "It is rare that clients have any problem with those cloud tools, but people are really skeptical of non-bulge-bracket, brand-name AI tools."

What contract reviewers want is a company that can account for how it handles data: where it goes, who touches it, what the AI model does with it, what the model retains, and what your agreements with your own vendors allow. Most small vendors cannot answer that in writing on short notice. Prepare in advance for a faster review.

5. Ban consumer AI tools for company data

Morrison advises clients not to put company information into consumer AI tools, and to write that rule down so they can hand it to anyone who asks. Customer legal teams ask about it, and client data privacy and AI policies can cover the same ground. A company that is loose with its own data has a hard time arguing it will be careful with a customer's.

"Just forbid your employees from using off-the-shelf, public AI [GARBLED: "toolqueries"], because a lot of those tools are unrestricted environments that will commercially use your data to train their products or systems," Morrison says. "You're giving this counterparty pretty heavy use of your data and sometimes, client data, if the team is not conscientious."

The exposure he has in mind is confidential material and HR records, handed to a counterparty with broad rights to use it. For anything sensitive, his position is that the work has to happen inside a private cloud environment.

Where Ann Arbor founders can get help paying for this

Founders trying to close a deal don't have to absorb the legal cost alone. Ann Arbor SPARK offers grant programs that can be applied toward professional services, including legal and intellectual property work, and additional statewide funding is available to bring in third-party specialists for early-stage technology companies. Each program has its own eligibility criteria (company stage, industry, location, and how the money can be spent all vary), so the first step is usually a conversation about which programs a company qualifies for.

SPARK sits at the center of a dense network of support organizations, university programs, and investors, and for most founders building here, SPARK is the first call. It also administers the Michigan Angel Fund, alongside its SPARK Capital pre-seed fund. Companies that reach those programs have usually come through earlier-stage support first, including the eight-week Entrepreneur Boot Camp and the Entrepreneur-in-Residence program, which pairs founders with experienced operators for ongoing mentorship.

The same regional network that helps with a first contract is where founders find their next customers, hires, and investors.