When the next big AI model lands, who gets stronger and who gets replaced?
Same rules for every product. Higher strength means safer. Click a product to see why.
Check your own productThe names everyone talks about.
1
AbridgeGets strongerIt is built into regulated work and hard to rip out. A better AI just makes it better.Gets stronger72
What we told the checker: Ambient clinical documentation: listens to doctor-patient visits and writes the note into the electronic health record. Integrated inside Epic, sold to hospital systems, with clinician-verified outputs.
0 means it gets replaced, 100 means it gets stronger.
It is built into regulated work and hard to rip out. A better AI just makes it better.
They record what actually happened after each answer, at scale. No AI lab or rival can get that, and it grows with every use.
It's built into regulated or contract-bound work (audit trails, compliance, certified connections). Replacing it takes a formal procurement process.
They own a channel: a big loyal audience, a marketplace spot or a partner network that rivals can't buy quickly.
The AI does much of the work, but most of the value comes from the data and process built around it.
What they should do next
- 1Get the certifications the regulated part of your customers' work requires, before rivals do.
- 2Put your edge into the tests, rules and checking tools around the AI, so the AI itself is a small part of the value.
- 3Build and publish a test for quality in your field, so you decide how 'good' gets measured.
2
GrammarlyGets strongerThey own an audience others can't buy. A better AI just makes it better.Gets stronger67
What we told the checker: Writing assistant with a browser extension and desktop apps that works inside any text field. Tens of millions of daily users, enterprise plans, long-standing brand. Added generative AI features on top of its grammar engine.
0 means it gets replaced, 100 means it gets stronger.
They own an audience others can't buy. A better AI just makes it better.
They're building a growing pile of real examples from their field that makes the product better for everyone and is hard to get elsewhere.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They own a channel: a big loyal audience, a marketplace spot or a partner network that rivals can't buy quickly.
The AI does real work, but only inside a system of the company's own data, connections and checks that the AI can't replace.
What they should do next
- 1Put your edge into the tests, rules and checking tools around the AI, so the AI itself is a small part of the value.
- 2Add shared records and team features, so leaving means losing history, not just a tool.
- 3Become the place where the work lives, not a layer on top of someone else's tool.
3
HarveyGets strongerIt is built into regulated work and hard to rip out. A better AI just makes it better.Gets stronger59
What we told the checker: AI platform for law firms and legal departments: document review, drafting, research, and due diligence over the firm's own matters and documents. Enterprise contracts with large firms, strong security and compliance posture.
0 means it gets replaced, 100 means it gets stronger.
It is built into regulated work and hard to rip out. A better AI just makes it better.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's built into regulated or contract-bound work (audit trails, compliance, certified connections). Replacing it takes a formal procurement process.
They find customers the same way every startup does: ads, search, launch posts, cold emails.
The AI does much of the work, but most of the value comes from the data and process built around it.
What they should do next
- 1Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 2Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 3Get the certifications the regulated part of your customers' work requires, before rivals do.
4
GleanHolds its groundIt is part of the team's daily routine, but they reach customers like everyone else.Holds its ground54
What we told the checker: Enterprise search and work assistant that indexes a company's apps (Slack, Drive, Jira, Salesforce, Confluence) with permissions awareness, then answers questions and runs agents over that index. Sold to IT and knowledge-heavy enterprises.
0 means it gets replaced, 100 means it gets stronger.
It is part of the team's daily routine, but they reach customers like everyone else.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's part of a team's regular routine, with shared records, permissions and connections. Replacing it means retraining people and rewiring systems.
They find customers the same way every startup does: ads, search, launch posts, cold emails.
The AI does much of the work, but most of the value comes from the data and process built around it.
What they should do next
- 1Get the certifications the regulated part of your customers' work requires, before rivals do.
- 2Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 3Lock in a deal to be the default option inside a platform your customers already use.
5
SierraHolds its groundIt is part of the team's daily routine, but they have some word of mouth and nothing exclusive.Holds its ground53
What we told the checker: Customer service AI agents for enterprises, deployed over existing CRM and support systems. Outcome-based pricing, long enterprise implementations, used by large consumer brands.
0 means it gets replaced, 100 means it gets stronger.
It is part of the team's daily routine, but they have some word of mouth and nothing exclusive.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's part of a team's regular routine, with shared records, permissions and connections. Replacing it means retraining people and rewiring systems.
They have a brand or community that brings some word of mouth, but nothing a rival couldn't also do.
The AI does much of the work, but most of the value comes from the data and process built around it.
What they should do next
- 1Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 2Turn what you learn from each customer into something that helps every customer: patterns, prices, comparisons.
- 3Put your edge into the tests, rules and checking tools around the AI, so the AI itself is a small part of the value.
6ElevenLabsHolds its groundThey're building data nobody else has. A better AI keeps it on its toes.Holds its ground50
What we told the checker: Voice AI: text-to-speech, voice cloning, dubbing, and conversational voice agents, built on its own audio models. API for developers plus consumer and enterprise products, and a licensed voice marketplace.
0 means it gets replaced, 100 means it gets stronger.
They're building data nobody else has. A better AI keeps it on its toes.
They're building a growing pile of real examples from their field that makes the product better for everyone and is hard to get elsewhere.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They have a large user base or a known brand that makes winning customers noticeably cheaper.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 2Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 3Get the certifications the regulated part of your customers' work requires, before rivals do.
7
ReplitHolds its groundThe AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.Holds its ground50
What we told the checker: Browser IDE with an agent that builds, runs, and deploys full apps from a prompt. Hosting, databases, and auth bundled. Consumer and prosumer users, strong education and hobbyist base.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They have a large user base or a known brand that makes winning customers noticeably cheaper.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Make the AI swappable and keep a test suite, so every new AI model is an upgrade you ship the same week.
- 2Add shared records and team features, so leaving means losing history, not just a tool.
- 3Turn what you learn from each customer into something that helps every customer: patterns, prices, comparisons.
8
RunwayHolds its groundThe AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.Holds its ground48
What we told the checker: AI video generation and editing tools for filmmakers and creative teams, built on its own video models. Subscription and API, partnerships with studios, research lab publishing its own work.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They have a large user base or a known brand that makes winning customers noticeably cheaper.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Add shared records and team features, so leaving means losing history, not just a tool.
- 2Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 3Lock in a deal to be the default option inside a platform your customers already use.
9
JasperHolds its groundIt is part of the team's daily routine, but they reach customers like everyone else.Holds its ground47
What we told the checker: AI content platform for marketing teams: brand voice, templates, campaign workflows, and approvals. Sells to mid-market and enterprise marketing departments. Started as a GPT-3 copywriting tool.
0 means it gets replaced, 100 means it gets stronger.
It is part of the team's daily routine, but they reach customers like everyone else.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's part of a team's regular routine, with shared records, permissions and connections. Replacing it means retraining people and rewiring systems.
They find customers the same way every startup does: ads, search, launch posts, cold emails.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 2Add shared records and team features, so leaving means losing history, not just a tool.
- 3Become the place where the work lives, not a layer on top of someone else's tool.
10
Copy.aiHolds its groundThe AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.Holds its ground47
What we told the checker: Go-to-market AI platform with workflow automation for sales and marketing content: prospect research, outbound sequences, content repurposing. Integrates with CRMs. Sells to revenue teams.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They have a brand or community that brings some word of mouth, but nothing a rival couldn't also do.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Add shared records and team features, so leaving means losing history, not just a tool.
- 2Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 3Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
11
Otter.aiHolds its groundThe AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.Holds its ground47
What we told the checker: Meeting transcription and notes. Joins Zoom, Teams, and Meet calls, produces transcripts, summaries, and action items. Freemium with team plans.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's part of a team's regular routine, with shared records, permissions and connections. Replacing it means retraining people and rewiring systems.
They have a large user base or a known brand that makes winning customers noticeably cheaper.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Lock in a deal to be the default option inside a platform your customers already use.
- 2Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 3Add shared records and team features, so leaving means losing history, not just a tool.
12
SynthesiaHolds its groundThe AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.Holds its ground44
What we told the checker: AI video platform with avatars and voices for corporate training and communications. Enterprise customers, template library, collaboration and localization features.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and rivals could copy the rest. A better AI keeps it on its toes.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's part of a team's regular routine, with shared records, permissions and connections. Replacing it means retraining people and rewiring systems.
They have a brand or community that brings some word of mouth, but nothing a rival couldn't also do.
The AI does most of the work. The product adds real tools around it, but a rival with the same AI could rebuild it in months.
What they should do next
- 1Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 2Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 3Add shared records and team features, so leaving means losing history, not just a tool.
13
CursorHolds its groundIt is part of the team's daily routine, but they have some word of mouth and nothing exclusive.Holds its ground44
What we told the checker: AI code editor forked from VS Code with codebase-aware chat, tab autocomplete, and an agent mode that edits across files. Subscription pricing, used by individual developers and engineering teams. Routes to third-party frontier models plus its own smaller models.
0 means it gets replaced, 100 means it gets stronger.
It is part of the team's daily routine, but they have some word of mouth and nothing exclusive.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It's part of a team's regular routine, with shared records, permissions and connections. Replacing it means retraining people and rewiring systems.
They have a brand or community that brings some word of mouth, but nothing a rival couldn't also do.
The AI does most of the work. The product adds a nice interface and some tailoring that a better AI would mostly absorb.
What they should do next
- 1Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 2Add shared records and team features, so leaving means losing history, not just a tool.
- 3Make the AI swappable and keep a test suite, so every new AI model is an upgrade you ship the same week.
14
Character.AIGets replacedThe AI does most of it and the rest is a nice interface. A better AI can do the same job.Gets replaced42
What we told the checker: Consumer app for chatting with user-created AI characters. Tens of millions of users with long session times, mostly young. Subscription tier. Historically trained its own models; now licenses from a lab.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and the rest is a nice interface. A better AI can do the same job.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It saves history and settings, but switching to another tool loses little.
They have a large user base or a known brand that makes winning customers noticeably cheaper.
The AI does most of the work. The product adds a nice interface and some tailoring that a better AI would mostly absorb.
What they should do next
- 1Build and publish a test for quality in your field, so you decide how 'good' gets measured.
- 2Make the AI swappable and keep a test suite, so every new AI model is an upgrade you ship the same week.
- 3Lock in a deal to be the default option inside a platform your customers already use.
15
MidjourneyGets replacedThe AI does most of it and the rest is a nice interface. A better AI can do the same job.Gets replaced42
What we told the checker: Image generation service with its own models, originally via Discord, now also on the web. Subscription only, no outside funding, large creative community. Known for a distinctive aesthetic.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and the rest is a nice interface. A better AI can do the same job.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It saves history and settings, but switching to another tool loses little.
They own a channel: a big loyal audience, a marketplace spot or a partner network that rivals can't buy quickly.
The AI does most of the work. The product adds a nice interface and some tailoring that a better AI would mostly absorb.
What they should do next
- 1Become the place where the work lives, not a layer on top of someone else's tool.
- 2Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 3Lock in a deal to be the default option inside a platform your customers already use.
16
LovableGets replacedThe AI does most of it and the rest is a nice interface. A better AI can do the same job.Gets replaced38
What we told the checker: Prompt-to-web-app builder that produces React front ends wired to Supabase, with one-click deploy. Fast consumer growth, subscription pricing, mostly non-technical founders and solo builders.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and the rest is a nice interface. A better AI can do the same job.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They find customers the same way every startup does: ads, search, launch posts, cold emails.
The AI does most of the work. The product adds a nice interface and some tailoring that a better AI would mostly absorb.
What they should do next
- 1Turn what you learn from each customer into something that helps every customer: patterns, prices, comparisons.
- 2Build an audience that trusts your judgment in your field, so the brand carries the product when the AI changes.
- 3Become the place where the work lives, not a layer on top of someone else's tool.
17
HebbiaGets replacedThe AI does most of it and the rest is a nice interface. A better AI can do the same job.Gets replaced36
What we told the checker: AI document analysis for finance and legal: runs questions across thousands of documents in a spreadsheet-like matrix. Sold to asset managers, banks, and law firms.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and the rest is a nice interface. A better AI can do the same job.
They hold each customer's own data, which helps that customer, but it doesn't add up across customers.
It plugs into the tools people already use, so results land where the work is, but it could be swapped out in weeks.
They find customers the same way every startup does: ads, search, launch posts, cold emails.
The AI does most of the work. The product adds a nice interface and some tailoring that a better AI would mostly absorb.
What they should do next
- 1Add the checking, approval and record-keeping steps, so people trust the output without re-checking it.
- 2Connect deeply with the two or three tools your customers live in, and put results straight back into them.
- 3Add shared records and team features, so leaving means losing history, not just a tool.
18
PerplexityGets replacedThe AI does most of it and the rest is a nice interface. A better AI can do the same job.Gets replaced35
What we told the checker: Consumer AI answer engine that searches the web and returns cited answers. Free tier plus Pro subscription, mobile apps, a browser, and an enterprise offering. Uses a mix of third-party frontier models and its own fine-tunes.
0 means it gets replaced, 100 means it gets stronger.
The AI does most of it and the rest is a nice interface. A better AI can do the same job.
They keep what users type in, but a rival could collect the same thing within months.
It saves history and settings, but switching to another tool loses little.
They have a brand or community that brings some word of mouth, but nothing a rival couldn't also do.
The AI does most of the work. The product adds a nice interface and some tailoring that a better AI would mostly absorb.
What they should do next
- 1Build and publish a test for quality in your field, so you decide how 'good' gets measured.
- 2Lock in a deal to be the default option inside a platform your customers already use.
- 3Connect deeply with the two or three tools your customers live in, and put results straight back into them.