Most of the teams I talk to have turned AI on. People have accounts, a few have built something clever, and someone senior is asking: now what? Where’s the benefit?
Last month I argued that we measure AI by usage when we should measure the decisions and results it changes. The numbers agree. Research by Apollo Global Management and chief economist, Torsten Slok, shows that artificial intelligence has delivered near zero measurable productivity gains or financial payoff across mainstream corporate America outside the technology sector. Revenue per worker is rising for the Mag 7 but is flat for the S&P 493.
The worry I hear more often is a different one, and it’s where many teams find themselves right now. Teams don’t know how to get to impact or how to demonstrate the value of their AI work. Real impact sounds like it needs a big programme first: a company-wide platform, new roles, a knowledge layer, heavy investments. Until then, it's too early to expect a number to move.
In one 2026 survey by Product Focus, a product management training firm, of 677 product people, 97 per cent reported better personal productivity. Only 64 per cent reported better product outcomes.
What I’ve seen is the opposite: you don't need a transformation first. Teams can realise value better in ways that executives and stakeholders will understand if they have already developed a habit of building a chain from their work to the company’s key metrics. The teams I’ve helped realise these changes often pick one way of working or focus, and name the number or metric it should move.
Today, I’m giving away a tool kit I call Impact First AI. I've built two free tools to run it with your team. Thanks to Matt LeMay’s Impact First Product Teams, and our collaboration on the coming: Impact First Leadership. These are some of the tools I use to facilitate impact for teams working with AI products. So make sure you download and try them at the bottom of this post.
The FIRST AI Maturity Model
Matt’s book Impact First, looks at how teams set themselves up for big picture business impact and the value of doing hard things. On top of that, I built a framework I’ve been using with clients to unpack practice maturity.
FIRST rates one way of working, such as discovery, planning, or build and testing. Consultancy ladders like BCG's Deploy, Reshape, Invent rate a whole company, or focus on too broad a level to actually drive impact into daily practices.
By contrast, FIRST focuses on practices and their impact in ways product leaders can use. Impact First AI can be driven into a company by teams who take the time to focus and understand their practices and product discipline. Rating a practice keeps the conversation close to the work, where people can show it, and keeps teams focused on a single discipline as they mature.
FIRST looks at these stages of maturity across practices. In the face of rapidly disrupted and changing ways of working, FIRST helps teams articulate and set maturity targets.
Faster: the task takes less time; the result is the same. Evidence: time per task drops.
Improved: the output is more accurate or complete. Evidence: less rework, and the people using it notice.
Reshaped: steps disappear, and decisions move to new people or to the AI. Evidence: fewer handoffs, measured end to end.
Scaled: the new way spreads beyond one team. Evidence: others use it unprompted, it's written down, someone owns it.
Transformed: decisions, roles and what customers get are redesigned around AI. Evidence: it improves itself, nobody goes back, the customer's job is different.
Download it below to check it out and have a FIRST conversation with your team. Everyone rates alone first, then you compare. The FIRST AI Maturity model helps teams align on the questions “where are we now,” and “what should we focus on next?”
Size the bet, then come back with evidence
The rungs are different sizes of bet. A team that writes tests faster with AI is at Faster: one team, a few weeks, a number you can watch. Getting to Improved means fewer errors someone notices. Getting to Reshaped means a handoff disappears. People are slow to change however much AI you add, so fund Reshaped and Transformed as medium and long bets.
Then check in every two to four weeks, for thirty minutes, with the people doing the work. Bring a number, a before and after, or the work itself, and move up a rung only when its evidence is true.
Shipping isn't the finish line. A merged change tells you the work is done. It doesn't tell you it helped. Mark it "shipped, awaiting evidence" and keep the bet open until the review date. That leaves room for an honest answer, like "we got faster, but rework hasn't dropped yet."
In a previous role, we set a target to move 30 per cent of data engineering time from maintenance to customer-facing work. We moved 44 per cent. The target came first.
What has to be true? Understanding the Impact Chain
The Impact Chain is a single page worksheet that helps teams determine how to make better AI product bets.
It starts with the job we’re hiring AI to do and works through how to track value across the product development journey. Chains are measured by their weakest link, wherever a team struggles to clearly articulate aligned assumptions in a single sentence or two, is a weak link. There’s the place a team can strengthen the chain.
Try it yourself
Both tools are free under CC BY 4.0: intentionaut.com/tools/impact-first-ai. If you'd like me to run it with you, that's what an AI strategy workshop is for.
Start with one practice. What has AI changed in it, and what would you need to see before calling that an improvement?
Reply and tell me. I'd like to know which practice you'd rate first.
Thanks for reading.
Intentionaut is a fortnightly letter from Saielle DaSilva on design, product, data and AI. What's working, what isn't, and how to keep your bearings.
If this was useful, forward it to one person who'd get something from it. That's how it grows.
Saielle DaSilva
Putting the soft back in software
