Make every AI claim clickable

Customers told me the same thing in workshop after workshop. They want to check the AI’s answer rather than just be handed it. I heard it running AI study-analysis sessions at Lyssna, in rooms full of researchers whose entire job is to avoid being fooled by a confident summary. The output that impressed them in the demo was the same output they wouldn’t put their name on. Most AI features die quietly in the gap between those two reactions.

What closes the gap is small and unglamorous. Make every claim clickable. When hey anna says a change is statistically significant, that claim is a link. Click it and you get the test behind it, the two groups, the sample sizes, the p-value, the rows it ran on. Not a restatement that it held, the working itself. When it says revenue rose 18%, the number resolves to the 340 orders it was computed from, and you get the orders rather than a summary of them. It’s the same move deep-research tools made when they started footnoting every sentence back to a source you can open.

There’s a further point under the citations. Because every claim resolves to your own rows, the agent stops being a box you send questions to and becomes an interface to your data. It works inside your dataset and updates it as you go. You can stay hands-off and let it run the analysis, or drop into the sheet and work the numbers yourself, and it’s the same data in the same place either way.

What clickable means

The bar is higher than a citation icon that opens a vague sources panel. A claim is clickable when three things are true. The link points at the exact evidence rather than the general neighbourhood, so the rows, the source document, the calculation, not “the sales table”. The path is short enough that a sceptical user will actually take it, which in practice means one click rather than a five-step drill-down. And the evidence makes sense when they arrive, so they can look at it and agree the claim is fair, rather than landing in a wall of raw data they now have to analyse themselves.

Miss any of the three and what you have is decoration. A citation that resolves to “rows 1 through 9,000” is honest and useless. A claim that takes four clicks to verify gets verified exactly once, by you, in the demo. The work is in making verification cheaper than doubt.

Why it changes adoption

The obvious reading is that citations make people trust the AI more. True, and the bigger effect is on what the user does next.

An AI feature gets adopted when it becomes a step in someone’s real workflow. A summary that lands in their inbox is something they read, nod at, and route around when the real decision gets made, because they can’t defend it to whoever asks where it came from. A clickable claim survives that question. They can forward it, put it in a deck, drop it into a board pack, because the evidence travels with the assertion.

You can see the split in usage data. The uncited version gets opened, admired and abandoned, with short sessions that don’t repeat. The cited version gets opened, clicked into and returned to, because the user has learned that the claims hold up when they pull on them. Verification is what lets someone put their own credibility behind your output, and people only do that with tools they can check.

What it costs to build

This isn’t free, and pretending otherwise sets you up to cut it under deadline. Clickable claims force an architecture where every assertion is traceable back to its evidence by construction. The number has to be computed deterministically, carry a reference to its inputs, and survive the trip into the sentence the user reads.

In practice the model never does the counting. It renders facts that were settled before it saw them. You need a UI layer that keeps the link between a phrase and its source intact instead of flattening everything into prose. And you have to resist the demo-friendly shortcut of letting the model freestyle a narrative, because a narrative with no anchors is the exact thing your customer has already learned not to trust.

Why it’s worth it

That cost buys two things that are hard to get any other way. It buys adoption, because a checkable claim is one a professional can act on without putting their own judgement at risk. And it buys you time, because traceability is slow to copy and gets harder to fake as your product touches more of a customer’s real data. A competitor can match your model in a weekend. Matching a system where every claim resolves to its evidence takes them as long as it took you.

The economics point the same way. hey anna runs at under a dollar a day against analyst alternatives that start north of $400 a month, and what makes the cheap version trustworthy enough to replace the expensive one is that you can click on what it tells you.

Build the feature so a sceptic can audit it in one click. The sceptic is your most valuable user, because they’re the one who decides whether anyone else gets to depend on it.