What matters What you want AI to do for the business, and what you would fix first.
02 When you think about AI in your product, which sentence sounds most like you?'AI can bring real extra value to our customers, but only if it's good enough that they trust it and use it. Losing that trust is what I worry about.' 'AI is where the market is going. Competitors are moving, and we can't be the ones falling behind.' 'I'm open to adding AI. I need to see that it pays for itself.' 'AI can hurt us if it goes wrong, with a customer or with a regulator.'
Which risks of adding AI capabilities to your product worry you most? Pick up to 3 that apply. Optional.
It gives a customer a wrong or harmful answer, and we lose their trust. A change breaks something that was working, and nobody notices until a customer does. Regulation, like the EU AI Act, or a large customer's security review asking for proof we don't have. Spending on AI that doesn't pay back. Falling behind competitors who deliver better quality. None of these worry me much right now.
03 If you could fix one thing about AI in your product this quarter, what would it be?Finally get it in front of customers. Know for certain whether it's good enough to trust. Ship improvements faster. Every change takes too long. Stop hearing about problems from customers first. Know what it costs us and whether it's worth it. Be able to show proof it works when a customer or a regulator asks. Build or bring in expertise, so the team can ship AI faster and at higher quality.
Quality and trust Whether the AI delivers for customers, and how you know.
04 How do you measure the value AI brings to your product?Pick up to two.
Usage. How much customers use the AI features. Quality of outcomes. Whether customers get the right answer or result from it. Time saved. Customers get to their answer or finish their task faster. Business metrics. Conversion, activation, retention or engagement moved because of it. Revenue increase. Customers pay for it directly, or it wins deals we would have lost. Cost reduction. Fewer support tickets, less manual work for our team. Positioning. Being seen as an AI-forward product by the market, customers and investors. We don't measure it yet. It's too early to tell.
05 How confident are you in the value the AI brings to your customers?Not very. Customers have complained, or they aren't really using it. Cautiously. The team tells me it works, and I go by that and by what customers say. Fairly. I see usage numbers and customer feedback, and they look fine. Very. I get regular numbers on how well it performs, not just on how much it's used.
Speed How fast AI capabilities and improvements reach customers.
06 How fast can your team deliver AI capabilities, or improvements to them?We haven't delivered anything yet. We're still preparing: data, infrastructure, working it through. We've been in prototype or beta for months. Every change is checked by hand and takes a long time. We ship changes, but we can't tell whether each one made things better or worse. We ship in days, and we know each change was an improvement.
07 What does the team show you before an AI update goes to customers?I don't see it. The team makes the call. A demo, or a few example outputs. A short summary of what was tested and how it went. Results on real customer cases, with numbers I can compare from one release to the next.
Cost What AI costs to run, and how you weigh that against quality.
08 How does the cost of your AI products or features show up for you today?It's small enough that it isn't on my radar yet. I see the total. I can't tell what drives it. The product team has a breakdown of what each capability costs. We know what it costs per customer or per request, and we manage it actively.
09 How do you think about the trade-off between AI cost and quality?Cost isn't a concern yet. Getting quality right comes first. Cost matters, but I wouldn't trade quality to lower it. Reducing AI cost is a priority for the product team, but we can't tell where we can cut without hurting quality. We know where cheaper is good enough and where it isn't, and we decide on that.
