Fellow for iOS: AI Meeting Notes for In-Person Meetings

Fellow’s iOS app recorded 29 Product Hunt upvotes on its mobile launch listing, a modest number that understates what the product actually does differently from every other AI meeting tool competing for calendar real estate in 2026.

The core claim is worth stating plainly: most AI meeting assistants don’t work when you leave your computer. The Product Hunt listing describes the iOS app as turning “your phone into a powerful AI note taker” for meetings that don’t happen on a screen. That’s a narrow but real problem, and it’s one the category has mostly ignored.

The Infrastructure Assumption Baked Into Every Competitor

Zoom, Google Meet, and Teams built a transcription pipeline that assumes a specific context. You’re at a desk. A browser window is open. A bot joins the call, captures audio, and hands you a summary. Otter, Fireflies, Notion integrations, the whole stack of tools that has accumulated around enterprise calendars: they all share this architecture. The bot is the product, more or less.

That works for the portion of professional life that runs through a scheduler link. It doesn’t work for a lot of the conversations that actually matter. The coffee chat where a colleague mentions they’re being recruited. The hallway exchange that becomes a product decision. The client lunch where pricing gets agreed to in rough terms before anyone writes a contract. None of those meetings have a calendar invite with a Meet link, so there’s nowhere to inject a bot even if you wanted one.

The workaround most people land on is pulling out their phone and hitting record. That produces an audio file. The audio file sits in your voice memos app. Nobody transcribes it. According to productivity research literature, unstructured notes from high-stakes informal conversations get referenced at dramatically lower rates than structured summaries, which is a polite way of saying the recording was pointless.

Fellow’s iOS app automates that second path. Hit record, have the meeting, get a transcript and action items when it’s over. The pitch from the app’s listing is direct: “just hit record and we’ll give you a transcript.” No bot. No computer. No requirement that the other person be on the same platform or even know what software you’re using.

Aydin Mirzaee and the Credibility Problem

Aydin Mirzaee, who co-founded Fellow and previously co-founded CanvasPop, was running the launch push himself on the day the Product Hunt listing went live. That’s a tactical choice worth reading into. In this product category, founder visibility at launch isn’t just about generating upvotes. It’s about the privacy question, which is the first thing any reasonable person asks when they hear that an app is recording meetings without a visible bot.

The notification problem is well-documented. Every existing tool has a version of the same moment: the bot joins a Zoom or Meet call, a banner appears that says “FellowAI is recording this meeting,” and then someone makes a joke and everyone’s slightly on edge for the first five minutes. That’s the best-case scenario, where the notification works as intended and everyone knows they’re being recorded. The compliance layer around AI recording is still developing, and IAPP’s guidance on AI in meetings reflects genuine unsettled territory around consent obligations, especially for in-person conversations that cross jurisdictions.

Fellow’s approach, which the listing describes as working “with or without visible bots, including support for in-person” meetings, doesn’t eliminate the legal question. It does change it. Whether it changes it in a direction that enterprise legal teams will find more or less comfortable depends on the specific context, and that’s a real consideration.

The “Most secure AI meeting assistant” claim on the Product Hunt listing is the kind of superlative that typically requires a footnote. Fellow doesn’t provide one there, which is normal for a product listing and slightly less acceptable as a feature your legal team would rely on. That said, the security angle is clearly load-bearing for the product’s enterprise positioning, and Mirzaee’s direct involvement in the launch suggests the team knows they need to own that claim personally, not just list it.

What the App Actually Does, Step by Step

Before a meeting, the app surfaces an “Ask Fellow” prompt for pre-meeting context. You can edit agendas from the same interface, set recording preferences from your calendar, and decide upfront what level of notes you want generated. The calendar integration means you’re not starting from a blank screen every time you open the app.

During a meeting you’ve started recording, the flow is simple. You’re not managing anything. The app is doing the capture.

After the meeting is where the product delivers its primary value. You can replay audio and video, read through full transcripts, review the AI-generated summary notes, and handle action items by either accepting what Fellow suggests or writing your own. That full loop, from pre-meeting prep through post-meeting action item management, is what separates this from a voice recorder with a transcription API bolted on.

The app works across meeting types: Zoom calls, Google Meet, Teams sessions, and in-person conversations with no digital infrastructure at all. The breadth there is genuinely wider than most competitors in 2026, where tools have generally specialized rather than expanded.

The Market It’s Entering

The AI meeting assistant space isn’t underpopulated. Otter and Fireflies have years of user data and established enterprise relationships. Notion’s meeting notes integration is deepening. Every major calendar provider is adding some version of AI-generated summaries natively. The pressure from platform-level features is real: when Google Meet builds transcription directly into Meet, the case for a standalone tool has to be stronger than “we also do transcription.”

Fellow’s differentiator is the mobile-first, no-bot architecture. That’s genuinely hard to replicate for tools built around the browser-bot pipeline, because the infrastructure assumptions run deep. Otter has a mobile app, but it was designed as an extension of the desktop product, not as a replacement for it. Fireflies is similar. Fellow’s iOS app, at least in its current positioning, is treating the phone as the primary recording device rather than a backup option, which changes what the product can do and where it can go.

The 215-word Product Hunt description the listing uses to make its case is doing a lot of work. It’s short for a product with this much surface area, and some of the claims, “Most secure AI meeting assistant” especially, deserve more substantiation than a landing page can provide. Whether that’s a concern depends on what kind of buyer you are. Individual users who want to stop forgetting what happened at lunch meetings don’t need an audit trail for the security claim. Enterprise buyers with 29 or more seats going through a procurement review absolutely do.

What’s Not Answered Yet

The in-person recording use case is the product’s strongest pitch and also its most legally ambiguous feature. Recording someone in an in-person conversation without their knowledge or clear consent creates a different compliance exposure than recording a video call where everyone sees a bot join. This isn’t a Fellow-specific problem; it’s a problem for any tool that records in-person audio. But it’s Fellow that’s marketing this capability as a feature, which means they own the category question more than their competitors do.

The IAPP’s guidance on AI in meetings is worth reading for anyone deploying this across a team. The short version is that consent rules vary significantly by jurisdiction, and “I told my phone to record this conversation” doesn’t map cleanly onto most existing consent frameworks. Fellow will need a clearer answer to this as the product scales beyond individual users who’ve made peace with the ambiguity.

None of that makes the product less useful. It makes it more complicated to deploy at scale without a legal review, which is true of most AI recording tools in 2026 and not unique to Fellow.

The Product Hunt listing’s straightforward directive, “just hit record and we’ll give you a transcript,” is accurate for what the app does. It’s also accurate about where the simplicity ends: the transcript is the easy part.