When I spoke with Steve, he told me he was hearing from at least one AI-native go-to-market company a day that had raised funding, sold to customers, and was struggling to get traction.
They had made it far enough to convince someone to buy. Getting people to keep using the product was proving harder.
“Adoption and usage seems to be the biggest challenge,” he said.
At the time of our conversation, Steve was leading the integration of Clari and Salesloft, so he was seeing this from both sides: customers trying to make a growing collection of tools work, and smaller companies looking for a way forward.
What stayed with me from our conversation on Inside the Silicon Mind was the gap between a product finding its first customers and earning a lasting place in their work. A signed contract gives a founder something important to build on. What happens after the sale tells you much more about whether the business can last.
Understand what the customer has to change
Steve described customers who had bought several AI tools to solve different parts of a business problem. Each tool might have a useful capability. The customer still had to manage the collection, connect the work, and get people to use it.
That last part can be easy to underestimate when you are building.
You know why your product is better. The person using it has an existing routine, other responsibilities, and a limited amount of patience for another system. They may agree with your pitch and still return to the way they worked before.
My takeaway is to study that return to the old routine. Choose a customer whose usage fell after the first few weeks and ask them to walk through a recent example of the work your product was meant to help with.
Where did they start? When did they reach for your product? If they stopped using it, what happened at that point? What did they use instead?
A founder may hear that the output needed too much checking, that information had to be entered twice, or that a colleague could not use the result. Those are different problems. A general request for more training will miss some of them.
The more precisely you understand the interruption, the better your chance of removing it.
Expand around the work that already matters
Steve’s view was that successful companies would grow beyond the first specific problem they solved and cover more of a customer’s workflow.
For an early founder, I would start by looking closely at the steps immediately around the product.
Suppose your tool helps a salesperson research an account. The research may be useful, but the salesperson still has to decide whom to contact, prepare a relevant message, and record what happened. Ask which of those steps is preventing the research from becoming useful action.
That is an illustrative example, but the question is practical: where does the value you have created get lost before the customer can use it?
The answer might be a small integration. It might be a clearer handoff to another person. It might justify a new capability. You need evidence from the customer before deciding how much to build.
I would also look at who is asking. Several active customers struggling at the same next step tell you something different from a prospect describing an entirely new use case.
That distinction can keep expansion connected to the reason people chose you in the first place. You are learning how to complete more of a valuable job, with customers who can show you whether the change helped.
Count the work your product leaves behind
Steve questioned how much of the activity around AI was actually giving people time back.
He described people spending more time at their computers because they could now build agents and attempt tasks they had not done before. Some of that work would be valuable. He wanted to know which parts produced a useful result and which parts simply created more activity.
The same question belongs in a customer review.
If your product produces an answer quickly, how long does someone spend checking it? What happens when it is wrong? Does the result arrive somewhere the customer can act on it, or does another round of copying and coordination begin?
For a founder, I would measure one completed piece of work from beginning to end. Include the preparation, review, corrections, and handoffs. Compare that with how the customer did it previously.
Then ask what they do with the capacity they get back. Can they serve more customers, respond sooner, or spend more time on a conversation that needs their judgment?
Those observations give you a more credible account of value than the speed of the model alone. They also help you decide which product improvement is worth making next.
Put retention alongside new bookings
Steve connected weak usage to a sequence he was seeing in smaller companies: retention becomes difficult, growth slows, and the next funding round becomes harder.
He expected some of those companies to look for acquisitions or acquihires. That was his view of the market, rather than a prediction about every AI startup. But the underlying operating problem deserves attention well before a company needs another round.
A founder can be busy winning new accounts while earlier customers quietly stop depending on the product.
I would bring those earlier customers into the same discussion as the new bookings. Who is using the product regularly? Who has reduced their usage? What happened to the problem they bought it to solve?
Look for differences between groups of customers too. If one type of team keeps using the product while another needs repeated encouragement, that may tell you where to focus your sales effort and development time.
The commercial question becomes more useful when it is tied to a specific customer experience: what would need to improve for this team to want another year?
Choose one result to improve this week
Near the end of our conversation, Steve recommended Will It Make the Boat Go Faster? He liked the discipline of testing daily decisions against a small number of priorities.
His own priorities were product innovation, bookings and revenue, and retention. He also described his team using its own sales software every day, giving them a direct view of where the product helped and where it needed work.
For this week, choose one customer whose initial enthusiasm has not become regular use. Reconstruct the last time they tried to complete the relevant task. Find one point where progress stopped or unnecessary work appeared.
Agree on a change small enough to test. Decide what a better result would look like for that customer, then go back and watch what happens.
The next conversation should give you something concrete: the customer completed the work, needed less help, or came back to do it again. If the change did not help, you have a more precise problem to investigate.
What would have to change for a customer who tried your product last month to depend on it next month?
Firas.