Kuulo vs Fathom: when unlimited free isn't enough

Fathom built one of the cleverest go-to-market strategies in the AI notetaker category: unlimited free recording. Not 300 minutes per month. Not a 5-recording trial. Unlimited — every meeting, no cap, no per-minute charge, basic CRM sync included. For a product in a market where every competitor charges by the minute or caps by the month, this is a genuine differentiator.

The free tier converts users effectively because it removes the friction of cost during evaluation. You don't run out of minutes three weeks in and face a paywall. You just use it.

The question is what you're actually getting — and what "unlimited free" means for your audio.

What Fathom does well

The unlimited free tier is real. Fathom's free plan provides genuinely unlimited recording and transcription. This is not a marketing claim that collapses under scrutiny — users report recording hundreds of meetings without hitting a cap. For teams evaluating AI notetakers, the ability to trial indefinitely without cost is genuinely unusual.

Transcript accuracy is strong. Independent benchmarks place Fathom's transcription accuracy in the same range as Otter and Fireflies — around 85–90% for clear speech in a standard video meeting environment. For its target use case, this is sufficient.

Basic CRM sync on free. Fathom includes basic Salesforce and HubSpot sync on its free tier, making it viable for sales teams without requiring a paid upgrade to access integrations.

Clean, simple UX. Fathom's interface is less cluttered than Fireflies or Otter. Meeting summaries are produced quickly and the format is readable without configuration.

Paid plan at $19/user/month (Pro) adds AI summaries, advanced search, and additional integrations at a price point below many competitors.

The structural constraints

Video calls only — with a meeting platform requirement

Fathom joins your meeting as a bot on Zoom, Google Meet, or Microsoft Teams. This is the only way it captures audio.

This means Fathom cannot record:

• In-person team meetings in a conference room
• Client site visits
• University lectures
• Hospital ward rounds
• Field research interviews
• Informal one-on-ones (unless deliberately moved onto a video call)
• Any conversation happening in a physical space

The most important business conversations increasingly don't happen on scheduled Zoom calls. Walk-and-talks where decisions are made. Hallway conversations that change the direction of a project. Client office visits where the real requirements emerge. Team lunches where the unspoken dynamics of a situation become clear.

Fathom was designed for the scheduled video meeting. It cannot capture anything else.

Stable internet required — always

Fathom sends audio to cloud servers for transcription. No internet, no notes. For users in stable office environments with reliable broadband, this is irrelevant. For users in hospital buildings, rural client sites, underground meeting rooms, or international locations with variable data connectivity, it's a genuine failure mode.

The visible bot

Fathom joins your meeting as a named participant. Every person on the call sees "Fathom is recording this call" — a disclosure that is required by Fathom's platform agreements and by legal recording consent requirements in many jurisdictions.

In most standard business meeting contexts, this is unremarkable. In some, it changes what happens:

• A sensitive performance conversation
• A mental health support check-in
• A negotiation where having a third-party record is unwelcome
• A client meeting where the client hasn't consented to their audio being processed by a US AI company
• Any clinical, legal, or therapeutic context

The visible bot is not a flaw — it's how Fathom functions. But it limits the contexts where it can appropriately operate.

"Unlimited free" and what that means for your audio

The economics of unlimited free cloud transcription are worth thinking through carefully.

Transcription through a cloud provider costs money — GPU compute, storage, bandwidth. Fathom's free tier processes unlimited audio through these systems. The business model that makes this sustainable is not public, but the standard industry mechanisms are: data used for AI model improvement, upselling to paid teams, and building a dataset of business meeting content at scale.

Fathom's privacy policy — like all cloud tools — describes data retention, processing purposes, and user rights. The relevant fact is that your audio is transmitted to and processed on Fathom's cloud infrastructure. The privacy guarantee is contractual.

For users whose meetings contain commercially sensitive information, client data, or personally identifiable information, the question of what "unlimited free cloud transcription" means for data handling is worth reading the privacy policy directly to understand.

The comparison

Two scenarios

Scenario A: product manager at a SaaS company, working from a city centre office with reliable broadband, attending 6–8 Zoom calls per day. Fathom is an excellent fit. Unlimited recording, good transcription accuracy, no cost, CRM sync if needed. There is no meaningful reason to switch from this for this context.

Scenario B: community physiotherapist, 8 home visit patients per day across a rural area, documenting treatment sessions on an NHS-supplied iPhone, patient data under UK GDPR Article 9 restrictions. Fathom cannot function in this context. No Zoom call. No reliable internet in rural areas. Patient audio cannot be sent to a US cloud server without a trust-level DPIA and DPA. And even if all those problems were solved, the visible bot cannot join a home visit.

Kuulo serves Scenario B. It cannot serve Scenario A as well as Fathom does — there is no CRM sync and no team collaboration layer.

The suggested workflow

These two products do not actually compete for the same use cases most of the time.

If your primary meeting context is scheduled Zoom/Meet/Teams calls from a stable internet connection with no sensitivity requirements: use Fathom's free tier. It is an excellent product for that context and the price is right.

For everything else — in-person meetings, offline recording, clinical or legal sensitivity, sensitive conversations where a bot is inappropriate, or any environment where internet access is variable — Kuulo is the tool that can actually capture the conversation.

The gap Fathom leaves is not a failure of the product. It's a consequence of an architecture that was designed for one specific context. For users whose most important conversations happen outside that context, that gap is the whole problem.

When the call drops and the notes don't arrive

Fathom is a video-call recorder. Its capture is entirely dependent on the video call itself — meaning every instability in the call platform is simultaneously an instability in the recording. If Zoom drops mid-meeting and reconnects, Fathom's capture is interrupted at that moment. If the call host ends the meeting prematurely, the recording ends with it. If Fathom's bot is removed from the session at any point, everything after that point is unrecorded.

Like all cloud-based recorders, Fathom depends on its downstream transcription pipeline. Processing delays or service disruptions mean notes arrive late or not at all — discovered when the meeting is over and the expected summary hasn't appeared.

The failure is always silent during the meeting itself. There is no on-screen indicator when Fathom's cloud pipeline experiences an issue. The user experience during the recording looks identical whether everything is working or nothing is.

For US teams operating under HIPAA, or internationally under GDPR or equivalent frameworks, silent failure compounds the cloud privacy exposure: the audio has already been transmitted for processing regardless of whether usable notes were produced. The data left the device; whether it was successfully transcribed is a separate question.

Kuulo's recording is independent of any external platform, call service, or cloud pipeline. It records from the device microphone, processes on the Neural Engine, and produces notes that are available on the device before you leave the room. For meetings where the notes are too important to risk on a cloud pipeline, the architecture is the guarantee.