QWE AI Academy You open Grindr, see 300 unread messages from the last two weeks, and have no idea which of the guys you were actually vibing with. That’s the specific problem Grindr AI Wingman was built to solve – not to write your flirty messages for you (yet), but to stop your best conversations from getting buried under the daily flood.
Here’s what most write-ups skip: AI Wingman isn’t one product. It’s a rollout of several features, only some of which exist in the app right now, and only for a slice of paying users. If you’ve read a dozen articles wondering when you can actually try it – this guide is the practical version.
What Grindr’s AI Wingman actually is (in 2025)
Wingman is Grindr’s umbrella name for an AI assistant that lives inside the app. Grindr’s 2025 Product Roadmap committed to six AI-powered features, with Wingman as the flagship. The first piece to actually reach users is called A-List – an auto-curated feed of your best past connections and high-potential matches, plus AI-generated summaries of long chat threads.
The model running it changed mid-2025 – and that matters more than the feature name. Claude Sonnet 3.7, Anthropic’s model, is the current engine, running on Amazon Web Services’ Bedrock infrastructure (Bloomberg, April 30, 2025). Before that, Grindr had partnered with a startup called Ex-human for its chatbot work. That partnership is over. If you tested an early Wingman prototype in 2024, you were talking to a completely different model than the one running today – which means those early Reddit reviews don’t describe what you’d actually get now.
Who can access it right now
Here’s where it gets annoying. Three access tiers, none of them obvious in the UI:
- A-List: Available to 25% of Grindr Unlimited subscribers as of end of April 2025 (per Bloomberg). Not all paying users. Not the free tier.
- Full Wingman chatbot: Still gated. The rollout plan per WSJ reporting: 1,000 beta users by end of 2024, 10,000 by end of 2025, full rollout to 14 million users by 2027.
- Opt-in only:Grindr’s own blog confirms they run internal lab testing before any controlled rollout with opt-in users.
No waitlist. No toggle to force it on. If you’re on Unlimited and A-List hasn’t appeared in your Inbox, you’re in the 75% that hasn’t been switched on yet – server-side, on Grindr’s schedule, not yours.
How to use A-List once you have it
One question it’s built to answer: which of my old chats are worth reviving? The actual flow:
- Open the Inbox tab. Users with access see an A-List section sitting above regular chats.
- The list surfaces what Grindr describes (in their roadmap) as your best connections and high-potential matches in one place.
- Tap any entry – you get an AI-generated summary of the thread. Long chats where you exchanged names, meetup plans, or specific interests get condensed so you can jump back in without re-reading 200 messages.
- You can filter connections by categories like “Met Up” and “Missed Connections.”
One thing to check first: Before you act on a summary, scroll back through the raw chat. Claude Sonnet 3.7 handles gist well but tends to flatten tone – sarcasm, inside jokes, and “enthusiastic but non-committal” messages can all read as genuine interest in a summary. If it says a guy “seemed keen to meet Saturday,” confirm he wasn’t being sarcastic before you double-text.
Think of A-List less like a matchmaking oracle and more like a highlighter over your inbox. It doesn’t know things you don’t – it just stops the good stuff from disappearing into the scroll.
The trade-off nobody spells out
Tinder, Bumble, and Hinge use AI too – but those features stay at the surface: pickup lines, profile rewrites, opener suggestions. They don’t go into your actual conversation history.
Grindr’s approach is different. The AI reads what you and another user actually said to each other – that’s how it generates the summaries. According to DatingNews, one stated goal is for the AI to analyze chats to determine compatibility. Platformer’s reporting added a sharper detail: Grindr has worked on a chatbot capable of explicit conversations, and early internal documents discussed the possibility of training the model on private chats between users, pending consent.
So: A-List on your account means Anthropic’s model is processing your DM text. Grindr’s public position is opt-in and consent-based. Whether you’re fine with that is a judgment call – worth making before you toggle the feature on, not after you’ve already used it for a week.
Common pitfalls when the feature goes live for you
1. Treating “high-potential match” as signal. Grindr has published zero benchmarks for how A-List ranks matches. No accuracy metric, no methodology note – nothing. It could weight recency, message length, mutual taps, or something else entirely. Treat it as a memory aid, not a recommendation engine.
2. Trusting summaries for logistics. Summaries compress. If you had a specific plan – day, time, place – read the actual message before you show up somewhere. AI summaries are good at gist, unreliable at specifics.
3. Expecting the full Wingman chatbot to be there. The conversational bot – the one that suggests venues, drafts openers, and eventually negotiates with other users’ AIs – is in limited beta. 10,000 users planned for 2025, full availability by 2027. A-List is what’s shipping now. They’re not the same thing.
4. Weighting old reviews. The Ex-human model is gone. Reddit threads and app store reviews from 2024 describe a product that no longer exists. For anything post-April 2025, Claude Sonnet 3.7 on AWS Bedrock is what’s running.
Grindr’s AI vs. everyone else’s
A quick comparison of how the big dating apps approach AI as of mid-2025:
| App | AI feature focus | Reads private chats? |
|---|---|---|
| Grindr (A-List / Wingman) | Match memory + chat summaries + future concierge | Yes, to generate summaries |
| Tinder | Photo selection + opener suggestions | No (profile-side only) |
| Bumble | Profile writing + AI concierge concept | Limited (profile-side) |
| Hinge | Prompt suggestions + prompt feedback | No (profile-side only) |
In Q2 2025, Grindr reported $104M in revenue – up 27% year-over-year, with $17M net income (Fast Company, October 2025). Their internal slide deck framed the strategy bluntly: “AI-native” means intelligence embedded at every layer, not bolted on as a feature. That’s the actual ambition behind A-List – it’s not a chatbot add-on, it’s a structural bet that reading your conversations produces better outcomes than staying at the profile layer.
The open question is whether Grindr’s userbase actually wants this. Other dating apps tested AI middlemen and faced immediate backlash. Grindr’s community skews more toward transactional interactions than romantic ones – which might mean less friction around AI involvement, or might mean users care even more about keeping those conversations private. Nobody knows yet. That experiment is still running.
FAQ
Do I need Grindr Unlimited to use AI Wingman?
Yes. A-List is Unlimited-only for now, and even then only 25% of Unlimited subscribers had it as of end of April 2025. Free tier users don’t have access.
Is my chat history being used to train Anthropic’s Claude model?
Using a model to summarize your chats is not the same as training the model on your chats – those are different operations with different data implications. Grindr’s public statements emphasize opt-in consent for anything training-related, and their blog describes an internal lab testing phase that precedes any user rollout. The document that actually governs what happens to your data is the in-app consent screen when you enable A-List – not a Bloomberg article, not this one. Read it. Specifically look for language about third-party model providers and training data use, because that’s the clause that matters if your comfort level is low.
Can I turn A-List off if I don’t like it?
Yes – it sits on top of your standard Inbox, not instead of it.
Your next move: Open Grindr, tap into your Inbox, and check whether an A-List section shows up above your regular chats. If it does, tap one summary and compare it against the actual thread. That side-by-side is the fastest way to learn what this AI handles well – and where you shouldn’t trust it yet.