AI Group Call

What it is in one line

Type a goal and seconds later you are on a live voice call with six AI minds — each cast for that goal with a name, a role and a personality. They speak one at a time, build on and disagree with each other, and go quiet the moment you start talking.

Who built it

Tash Ahmed (tashthemaker at launch), an independent developer, solo project. Her launch post explains the motivation: she could ask a chatbot for advice, but what she actually wanted was the room — the meeting where a skeptic pokes a hole in your pitch, a strategist reframes it, and someone plays the customer who has to sign off. So she built the room.

Read: this is "the format is the product" in its pure form — she did not want a better answer, she wanted the shape of a meeting. The insight is honest and valuable: in many decision moments users do not want the single best answer; they want the process of positions colliding.

What it actually does

• Six AI roles on a live voice call → cast against the goal, each with name, role and personality; they speak one at a time, and build on or argue against each other
• Real barge-in → the room goes quiet the instant you speak. The hard part: over a phone speaker the mic also hears the agents, so the maker says heavy echo handling and raw-audio voice detection sit underneath, making interruption feel like a real call rather than a walkie-talkie
• Recast mid-call → tap any agent during the call and rewrite its name, role or personality
• Replay and capture → every call is transcribed, can be summarized into key points and action items, and can be rejoined later with the same cast
• Every participant is AI → there are no humans on the calls

What it deliberately is not: a one-on-one exchange with a single chatbot, or pre-recorded audio.

What old behavior it replaces

"Getting a few people to pressure-test my plan" used to mean three expensive routes:

Ask one chatbot — fast but single-sided: no skeptic, no opposing role, no rebuttal. You get "an answer" instead of "a discussion."

Actually hold a meeting — scheduling, availability, finding the right people. A good brainstorm costs hours plus social capital, and most of the time you cannot even assemble the skeptic-plus-strategist-plus-customer combination.

Pay a consultant — the most expensive option, and a consultant is still one person with one perspective.

AI Group Call replaces "organizing a multi-sided discussion": six positions become AI roles you cast on demand, and you are in the room in seconds without scheduling anyone. Its bet is that the most valuable parts of the old experience are the ritual of a meeting and the collision of multiple positions.

Business model

Minute bundles:

• New accounts get 1 free minute, no card
• Then monthly minute bundles starting at $4.99

Android is live; iOS is built and waiting on App Store review.

Read: per-minute pricing is sensible for a voice product — inference cost scales with time, so bundles keep cost and revenue aligned. But the pricing also exposes the ceiling: solo, time-based, no ads, a "good enough" small product trajectory, unlikely to grow into a big company.

Hard numbers

• Pricing: from $4.99/month; 1 free minute for new accounts (no card)
• Platforms: Android live; iOS awaiting review
• Users, call volume, retention: not disclosed
• Team: solo (Tash Ahmed)
• launch around 2026-08-09; no upvote data captured

Four-way read

The call

"Wrapping AI in the shape of a meeting" is a correct judgment, but this is a small business. It hits a real need precisely: people want positions to collide but cannot assemble the people or wait for schedules. Six AI roles with personalities, arguing with each other, interruptible at any moment — the product experience is complete, not a demo.

The technical bet is right. Barge-in done well feels like a real phone call; done poorly it feels like a walkie-talkie, and the maker names it as the bulk of the development time. That is the correct priority for a voice product.

The cold water: per-minute pricing plus solo development plus zero network effects means acquisition relies on launch windows like the launch platform, and retention relies on "someone genuinely wants to run this kind of meeting weekly." The need is real but low-frequency, and the ceiling is visible. The thing worth watching is not whether it gets big, but how much of the "AI multi-role discussion" format gets validated.

What to watch next

① First-month downloads after iOS ships — cross-platform is when real demand shows ② Bundle repurchase: how long between a user's first bundle and the second — the core metric for a voice product ③ Whether a high-frequency entry appears (custom meeting goals, industry-specific casts)

What you can take from it

Product logic: if discussion is your product shape, do not ship one AI role — ship a cast with conflicting positions, and let the user recast live. Multi-sided collision is worth more than a single "smarter" role, because the user wants the process, not the conclusion.

Pricing structure: one free minute with no card, letting people finish a call before any payment decision. And selling voice by the minute, aligning cost with revenue directly. Both are borrowable.

Verdict

Worth watching. The product is real, the experience is complete, and the pricing is honest, and it validates that the "AI multi-role meeting" format meets a real need. But it is naturally small — solo, per-minute, no network effects. Watch the post-iOS downloads and minute-bundle repurchase; those two numbers say whether it is a sustainable product at all.