AI made every product look the same, and the positioning followed
Most of what used to take months of domain-specific engineering can now be assembled in a weekend using the same foundation models everyone else has access to. Foundation Capital found that a legal ops system now looks like a claims processor, and a RevOps copilot looks like an underwriting assistant. Products have converged, and the language has followed.
At the same time, competitors started showing up from directions nobody planned for. A startup with 18 months of history enters your category. A buyer prompts ChatGPT and gets 80% of what your product does. A product manager builds an internal tool with Lovable over a weekend.
On the buyer side, the room got bigger. Forrester's 2026 survey of nearly 18,000 business buyers found that when a purchase includes AI features, the buying group roughly doubles, from about seven people to fourteen.
April Dunford makes a similar observation: product, competitors, and market are all shifting at the same time. She notes it's the first time since 2020 she's seen positioning break this broadly.
Companies replacing manual work don't mention AI
When the buyer's real alternative is a manual process, they're not comparing AI products. They're weighing your product against what they already do: the freelancer they hire for each project, the three-sprint wait for engineering bandwidth, or skipping the task entirely.
Naming AI adds nothing to that comparison.
Lovable turns a text prompt into a working full-stack application. The thing it replaces isn't another dev tool. It's the decision not to build because it would take too long. The homepage describes what you get ("Build Software in Minutes") and doesn't mention how. The buyer cares that the app exists, not what model built it.
This pattern shows up when the product does something the buyer couldn't do before without serious cost or time. In those cases, explaining the mechanism gets in the way of the result.
If you're entering a mature category, AI is the selling point
When you're entering a mature category with an established leader, you need two things in the same breath: the category name so the buyer can place you, and a reason to switch. Right now, AI is often that reason.
Attio positions as "the CRM for agentic revenue." CRM tells the buyer which budget line and which evaluation to put Attio in. "Agentic" tells them why it's worth considering alongside Salesforce. Take out either word and the sentence stops working.
Cursor does something similar in dev tools: it forked VS Code, inherited the category, and used AI-native architecture as the reason to switch editors. Lemlist leads with "the AI Outbound Platform," entering the ground Outreach and Salesloft defined and claiming the next generation of it.
In all three cases, AI is the answer to "why now?" It tells the buyer why a decision they already made is worth revisiting.
The problem is shelf life. The incumbent can make the same claim within a year, and usually does. HubSpot, Salesforce, and Outreach have all added AI branding in the past 18 months. Once they do, the challenger needs a new differentiator.
The biggest players use the AI label to defend the category they own
For a large established company, the positioning threat right now often isn't a better competitor. It's the buyer deciding the category no longer matters. If CRM becomes a data layer that agents query, Salesforce's #1 position in CRM stops being relevant.
That's why these companies lead with AI even though AI isn't what differentiates them from each other.
Salesforce saying "AI CRM" is Salesforce insisting that CRM is still the right frame for the buyer's evaluation. If buyers keep shopping for CRMs, the incumbent keeps winning. Dreamforce 2026's theme was "Becoming an Agentic Enterprise," which is an argument addressed to investors and analysts as much as to customers.
Each incumbent also tells a specific story about where AI is headed, and every story points back to what they already own. Microsoft argues that models are commoditizing and context is what matters (Microsoft owns the documents and inboxes where context lives). ServiceNow says workflows and governance are the ballgame (ServiceNow owns the workflows). HubSpot introduced "Growth Context," arguing that AI needs their customer data to be useful, which makes HubSpot the prerequisite rather than the thing that gets replaced.
Dunford observed the same pattern: each company's point of view on the future is rooted in its existing strengths. That's strategic positioning, not just messaging.
There's also a simpler factor at work. Saying AI costs an incumbent almost nothing, because buyers already expect it. Not saying it sends a signal that you've fallen behind. That asymmetry on its own explains a fair share of what we see.
When the buyer's alternative is to DIY
This is the competitive alternative that grew fastest and that almost nobody talks about publicly. For a growing number of AI products, the real competition is a general-purpose model plus the buyer's own engineers.
The objection sounds like: why would I pay for your platform when three engineers with Claude or ChatGPT could build something close enough in a few weeks?
Against this alternative, leading with AI backfires. It reminds the buyer that the underlying technology is available to everyone, including them.
Harvey is a useful case here. "AI software for legal and professional services" sounds like a vertical label, but it describes the company's delivery model. Dozens of legal AI vendors can extract entities from contracts. Harvey sends legal engineers into Am Law 100 firms for weeks. They learn how that specific firm handles redlines, how it structures clauses, how decisions escalate internally. A general-purpose model can't acquire that knowledge, and Harvey's positioning points at exactly that gap.
What calling your product AI actually costs you
Forrester's 2026 survey found that when a purchase includes AI features, the buying group roughly doubles, from seven to about fourteen. More than 60% of buyers insist on a trial before committing, and only about a third convert with the same vendor afterward.
On the vendor side, the pattern is the same: buyers expect to test with their own data before signing, and when a product claims to do a job that a human currently handles, scrutiny goes up.
Labeling a product as AI changes who shows up to evaluate it. For a challenger entering a mature category (like Attio), that cost comes with the territory because AI is the reason the buyer is reconsidering. For a company replacing a manual workflow (like Lovable), the same label brings more people into the room without adding anything to the argument.
Every position here is basically a bet
Each of these four patterns rests on an assumption about what stays true.
Lovable is betting the job persists. Attio is betting the CRM category still organizes how buyers shop. Salesforce is betting its category survives as the dominant frame. Harvey is betting that a general-purpose model won't close the gap between raw capability and the embedded, firm-specific work its engineers do.
Some of those assumptions hold for years. Others break with the next AI model release. The difference usually comes down to whether the positioning is anchored in what the product can do today or in the value the buyer gets out of it. Features change every quarter. Value tends to move more slowly.
These four patterns aren't clean boxes, and a company can sit in more than one depending on which buyer segment it's talking to. But they're a useful starting point for the questions most teams are working through right now: what role should AI play in how we position our product? When is category creation worth considering? Should the pricing model match the positioning claim? And how do you build a position that holds when the next model update changes what's possible?
The answer to all of them depends less on how much AI is in the product and more on who you'd lose the deal to. That's what the next piece picks up.