THE OPERATOR STACK
Thirty to forty minutes.
That is how long most coaches spend writing after every client call. Session recap. Action items. Follow-up email. Notes for next time. Repeat, eight times a week. That is nearly five hours weekly that isn't being billed, isn't resting, and isn't growing a practice.
The promise of AI note-takers is simple: record the session, get a usable summary, move on. The reality is more complicated. After three months of testing across real coaching workflows — not staged demos, not vendor trials — here is what the testing found.
Otter.ai — Deleted after week 3
Otter is where most coaches start because it is the name they have heard. The problem: the free plan caps at 300 minutes a month and 30 minutes per session. A 60-minute coaching call cuts off halfway through. That gets discovered after it has already happened.
Beyond the limits, the summaries are generic. 'The team discussed goals and next steps' is the kind of output Otter produces — technically accurate, completely useless. It only transcribes in English, French, and Spanish, has no video recording on any plan, and no keyword tracking. For a coach running bilingual sessions or wanting to search past conversations, it falls short immediately.
Verdict: The name recognition isn't matched by the product. Skip it
Read.ai — Deleted after 12 days
Read.ai has a clever pitch: it analyses talk time, engagement score, and 'meeting quality.' In practice, coaching sessions don't need a meeting quality score. They need a clean summary a client can read.
The larger problem: Read.ai has a documented practice of appearing in meetings for participants who didn't agree to it — a non-starter in a coaching context where client trust is foundational. When a client sees 'Read.ai has joined the meeting' without prior agreement, that is a conversation no coach wants to have mid-session.
Verdict: Fine for corporate analytics, wrong for a coaching relationship.
Notta — Deleted after 3 weeks
Notta is a clean, well-designed tool with solid transcription accuracy. The issue is what it does with the transcript: it renders the words in a marginally better format than a wall of text, and not much more. Output structure matters as much as accuracy — a wall of transcript text, however accurate, is less useful than a summary that cleanly separates action items, decisions, and follow-ups.
Notta's AI summaries aren't structured in a way that maps to how coaching sessions actually work. The editing required exceeded what a prompt-based alternative would take from scratch.
Verdict: Good transcription engine, weak coaching application.
tl;dv — Tested for 3 weeks, passed on
tl;dv is genuinely impressive for what it's designed for: sales teams and customer research. The timestamp clipping, the moment-sharing, the cross-meeting trends — all excellent. But that is not coaching.
The summary templates aren't structured around coaching frameworks. There is no way to configure outputs around session recaps, client action plans, or between-session commitments without significant customisation work. It joins calls as a visible bot, which some clients still find unsettling.
Verdict: Brilliant if you're in sales. Not built for coaching. Strong tool in the wrong context.
Fireflies — Still running as a backup
Fireflies is the most capable tool on this list in raw feature terms. It achieves consistent accuracy across diverse meeting types and supports over 60 languages. The keyword search across your entire archive is genuinely useful — when a client references something from six weeks ago, you can find it in seconds.
Why it isn't the primary recommendation: the Fireflies bot joining a call can shift the energy in a session where trust and openness matter. Two clients have mentioned it unprompted. One asked for it to be turned off mid-session. There is also no tailored coaching output — it produces a strong transcript and a generic summary, not something structured around how coaching actually works.
Verdict: Best in class for search and multilingual teams. The bot presence is a genuine friction point for coaching specifically. Use it as a backup or archive tool.
Fathom — The one worth keeping
Fathom is the best free option for individuals, and in a coaching context specifically, it isn't close.
Why it works:
— It is genuinely free with no catches. Unlike competitors that cap at 300–600 minutes per month, Fathom allows recording every call without hitting a paywall. Eight coaching calls a week, every week, zero cost. That alone eliminates Otter.
— The summaries are structured. Fathom produces clean outputs with action items, decisions, and highlights separated and labelled — not a wall of text. Within 30 seconds of a session ending, there is a draft recap that is 70–80% of what a coach would write themselves. Five to ten minutes of editing, not thirty minutes of writing.
— The privacy and compliance story is solid. Fathom maintains HIPAA, GDPR, and SOC2 Type II compliance — which matters when clients are sharing personal or professional vulnerabilities on a recorded call. Coaching is not HR surveillance, and data needs to be treated seriously.
— Setup takes about 8 minutes. Connect to Zoom or Google Meet, enable calendar sync, and it joins automatically. No management during the session. When the call ends, the summary arrives.
One important thing to know:
Fathom joins meetings as a visible participant named 'Fathom Notetaker.' This isn't a flaw — transparency is the right call in a coaching context. But clients need to know about it before the first session. One sentence in the onboarding email handles it: 'Our sessions are supported by an AI note-taker called Fathom, which helps create accurate session recaps. You can ask to turn it off at any time.' In practice, no client declines.
Verdict: Set it up once. Leave it running. The only note-taker worth recommending for coaching businesses in 2026.
OPERATOR MOVE
The session recap prompt — Fathom output to client-ready document in 3 minutes
Fathom's summary is good raw material. It is not the finished product. Here is the prompt to paste into Claude immediately after every session, with Fathom's summary copied in:
You are a professional [life/business/executive] coach. Below is an AI-generated summary of a coaching session with my client [Name], who is [brief context: role, what they're working on]. Using this summary, write a polished session recap (200–250 words) structured as:
1. THIS SESSION'S KEY INSIGHT — the most important thing they realised or articulated today
2. THEIR COMMITTED NEXT ACTION — specific, with a deadline if one was mentioned
3. WHAT TO CARRY INTO NEXT SESSION — the thread to pick up
4. A CLOSING LINE — warm, direct, not therapy-speak Match this voice: clear and direct, never clinical. Here is the session summary: [paste Fathom output]
Total time from call ending to client-ready recap: under 5 minutes. Send it the same day. Coaches consistently report this is one of the most valued things they receive — it costs almost nothing to produce once the system is running.
SIGNALS
Three things moving in the coaching + AI world this week
1. Apple pulled back on its AI health coaching project. Apple has wound down development of its AI-powered health coach designed to analyse data from Apple devices and provide tailored recommendations on exercise, diet and sleep, citing accuracy concerns and regulatory hurdles. The interpretation most people got wrong: AI is losing to coaches. The real signal: AI is struggling with the empathy gap in health contexts — exactly where human coaches have an irreplaceable advantage. File this one.
2. The coaching market hits $24 billion. The health coaching market alone is rising from $22 billion in 2025 to $24 billion in 2026, with a compound annual growth rate of 9.3%, driven by AI-integrated coaching tools and demand for mental health support. The market is growing and AI is a growth driver, not a threat. The coaches losing clients aren't losing them to AI — they're losing them to coaches who use AI more effectively.
3. Google Meet is flagging AI bots as security risks. Google now flags third-party notetaker bots as "potential risk" by default in Google Meet — which means tools like Fireflies and Otter may fail to join your calls entirely, depending on your client's organisation settings. If you're running sessions on Google Meet and your note-taker keeps getting blocked, this is why. For now, Fathom has smoother Google Meet integration, and desktop app avoids the issue entirely by recording locally. Worth knowing before it surprises you mid-session.
FROM THE OPERATOR
Welcome to Issue #1.
The Coaching Operator exists because most AI content written for coaches isn't written from a coaching workflow perspective. Every roundup found during the research for this newsletter was written for sales teams or enterprise software buyers. The nuances that matter to a coach — client confidentiality, session energy, the difference between a transcript and a recap — were always missing.
The format here is fixed: one tool reviewed properly, one move you can implement the same day, three signals from the space. Under 10 minutes to read, every Tuesday.
One ask: hit reply and tell me which part of your coaching practice you most want AI to help with. Every reply gets read, and the answers shape what gets covered next.
See you next Tuesday.
— The Coaching Operator