Why 'AI-powered' is dead positioning (and what works instead)

Investors from Emergence Capital, AltaIR Capital, and 645 Ventures told TechCrunch which phrases make them stop reading a pitch.

The list. "AI-powered." "10x better." "Vertical SaaS for X." "Purpose-built." "Seamless integrations." "End-to-end platform."

The article stopped one step short of the useful part. A VC gets tired of those phrases after five hundred pitches. A buyer gets tired after fifty homepages.

Which means our buyers quit reading long before anybody wrote the story.

We're still talking to people who stopped listening.

---

The signable sentence

A claim is dead when three competitors could sign their name to it without lying.

Call it a signable sentence. It's true, it's accurate, it's ours, and it's equally theirs.

"AI-powered" describes how we built the thing. It says nothing about the problem, the person, or what changes in their week. Anything with an API call to a language model can claim it.

"10x better" is a claim with no referent. Better than what, on which dimension, for whom?

The number sounds precise and answers nothing.

"Vertical SaaS for X" meant something when category-specific products were rare. That window closed. Every category now has five vendors claiming vertical focus.

Jake Saper at Emergence Capital put the funding version of this plainly. "If your differentiation lives mostly in UI and automation, that's no longer enough. The barrier to entry has dropped."

Buyer decisions run on the same logic. Somebody staring at a dozen vendors who all say the same thing needs a reason to shorten the list, and a signable sentence hands them nothing.

---

Written from the inside out

April Dunford's "Obviously Awesome" names feature-list positioning as the most common mistake in the discipline. Buyers are shopping for escape from a problem. A capability is something they have to translate first.

Most won't do that work. They'll leave.

"End-to-end security review platform" hands the buyer the translation job. "We help mid-market security teams cut questionnaire turnaround from fourteen days to two" does the job for them.

The second one lands in about a second. That's my problem.

Every dead phrase has the same origin. We spend eighteen months building, so we know the architecture, the integrations, the retrieval setup, the model sitting behind the AI.

Then we describe what the product is made of, because that's the part we know best.

At that distance, writing about our own product from the outside is genuinely hard. We can't unsee what we know.

A buyer reads the headline and wonders what it means for them. We read the same headline and fill in everything standing behind it, because we built everything standing behind it.

That gap is the difference between a homepage that bounces and one that converts.

---

Words our customers already found

The escape is embarrassingly cheap.

Find the three most enthusiastic paying customers. Ask what they say when they explain the product to a colleague. Ask for the hallway version, in their words, with no feature list in it.

Their words beat ours almost every time. They translated the product into terms their own team understood, and they did it knowing nothing about our architecture.

Use theirs.

What comes back usually starts from the worst week they were having before they found us. Get specific about that week. What did it cost, what was at risk, and what was happening to the business while it went unsolved?

A security team taking fourteen days to return a questionnaire isn't mildly inconvenienced. Deals are stalling in review, the VP of Sales is escalating, and the CFO is watching signed-in-principle work slide past quarter end.

That's the before state. "Fourteen days to two" is the after state, and it hands over a measure while it's at it.

The number does as much work as the claim. "Faster reviews" evaporates on contact. "Fourteen days to two" is something a reader can hold up against their own calendar.

---

The filter before it ships

Run every positioning claim through three questions before it goes live.

Who, specifically. Not "enterprise companies," not "SaaS teams." Name the title, the stage, and the context of the person who feels this most.

"Series B security leads fielding fifteen-plus questionnaires a quarter" passes. "B2B companies" doesn't get close.

What changes, specifically. "Save time" and "improve efficiency" describe every product ever built. Name the thing that measurably moves after they implement, because a number beats an adjective every time.

Then the question we skip. So what?

Questionnaire turnaround going from fourteen days to two means deals clear review faster, which means revenue lands in the quarter it was forecast in, which means the CFO stops asking.

Position at the level of consequence the buyer is personally answerable for. Answer all three inside the headline and no competitor can sign it.

---

What's actually hard to copy

Four things resist replication, and the claims that work now describe them.

Proprietary data comes first. If the product improves as customers use it, because we're learning from a data set nobody else can reach, say so plainly. It's the most defensible claim available to most teams and almost nobody states it clearly.

Workflow depth is second. "Integrates with your workflow" tells a buyer nothing at all. Which exact step, in which exact workflow, do we own?

The more precisely we can describe the job, the better we hold up when someone asks whether AI could just do this.

Switching cost is third. What does the buyer lose the day they rip us out?

Integration work, audit history, institutional data, team habit. All legitimate, all badly underused.

The outcome guarantee is last and hardest. What business result do our best customers get, and can we put a number on it? If we can't, finding that number is the most important positioning project on the list.

Descript is worth looking at here. Every page, post, and line of its copy speaks to one audience, people who want to edit podcasts without knowing how to edit video.

Adobe could serve that audience with vastly more resources. It has never claimed the outcome the same way.

That consistency matters more every quarter, because the systems assembling a buyer's shortlist have to resolve who we're for before they'll name us. A sentence three competitors could sign gives a machine nothing to work with either.

---

Read the homepage headline one more time.

Then ask who else could sign it.

---

For a complete overview of B2B SaaS positioning strategy, including frameworks, common failure patterns, and how positioning connects to messaging, ICP, and GTM, see the B2B SaaS Positioning Guide.

What to do next

If three competitors could sign your headline without lying, the repair starts with customers rather than a copywriter.

That's what a messaging sprint is for. Three customer conversations, the outcome language they already use, and a headline nobody else can put their name on.

If that's where you are, start here. The first conversation is free.