Best AI Notetaker for Consultants (2026): Fathom vs Otter vs Granola vs Basil for Client-Confidential Work

Published November 17, 2026

Key takeaways

Quick answer: The best AI notetaker for consultants depends on your client contracts. For solo consultants bound by NDAs that restrict third-party processors, on-device tools like Basil AI keep audio on your Mac or iPhone. Fathom suits budget-conscious solos on internal calls, Granola fits bot-averse relationship work, and Otter covers mature virtual-meeting workflows — but each sends audio to the cloud.

Published November 17, 2026 · 11 min read

The best AI notetaker for consultants is not a single product — it is whichever tool survives the NDA you signed before the kickoff call. Independent consultants concentrate risk the way few other professionals do: one laptop, one notetaker account, and audio from pricing talks for Client A, acquisition rumors for Client B, and sometimes two clients who compete with each other. For NDA-bound client calls, on-device tools like Basil AI keep audio on your Mac or iPhone and never hand it to a third-party processor. For free internal-call capture, Fathom is the strongest option. For bot-averse relationship work on video, Granola leads. For mature virtual-meeting workflows, Otter covers more integrations — but with the biggest legal cloud over it.

This guide compares the four options buyers actually ask about, with the policy language and pricing you need to defend your choice if a client asks.

Why "best AI notetaker for consultants" is a different question

Most "best of" lists rank tools on features: summary quality, speaker labels, CRM integrations. For a consultant, the ranking inverts. A tool can have the slickest summaries in the world and still be unusable on a client call because the confidentiality clause you signed decides which notetakers you can run, and for most cloud tools the answer is already no.

The reason: Eaton Smith's analysis of AI notetaker contracts points out that if an NDA, MSA or engagement contract restricts disclosure of confidential information to third parties, or if a DPA requires prior approval for sub-processors, then using a cloud AI tool may amount to a contractual breach. Independent consultants almost never get that prior approval — because they never ask.

The other wrinkle, from Lilach Bullock's writeup on NDA risk: if an AI notetaker like Otter, Fireflies, Fathom or Read AI is joining your client calls, it is recording, transcribing and storing that conversation on a third-party server, and most small business owners have never checked whether that breaks an NDA, a client contract, or GDPR.

The quick comparison table

Here is the shortlist a buyer actually needs, scored on the things that matter for solo consulting work.

ToolCapture styleWhere audio is processedTrains on your data?Entry price (verified Nov 2026)Best for
Basil AIOn-device, no botYour Mac / iPhone (RAM)No — audio never leaves deviceOne-time purchase, no per-seat feeNDA-bound client calls, in-person meetings, regulated industries
FathomBot joins the callFathom's US serversYes, de-identified, opt-out in settingsFree tier; Premium ~$15/user/moSolo consultants on internal calls, budget-first
GranolaBot-free, listens via mic + system audioAWS US (notes + transcripts)Yes, anonymized, opt-out (Enterprise-gated for full controls)Basic free (30-day history); Business $14/user/mo; Enterprise $35Bot-averse relationship work on video
Otter.aiOtterPilot bot joinsOtter's US serversOtter says no (per its AI providers); active class-action pending on past practices~$8.33/user/mo (Pro, annual); Business ~$20Mature live transcription on virtual calls

The rest of this guide walks each option through the four questions a consultant actually has to answer: does it work with my NDAs, does it train on my client's words, does it add a bot to the call, and what will it cost me as a solo operator.

Fathom: the free default, with an asterisk on training

Fathom is the tool most lists name as the best free pick for solo consultants, and that remains true as of this writing. Fireflies' 2026 consulting roundup calls Fathom the fit for independent consultants who need reliable transcription without paying anything, and notes it holds SOC 2 Type II, GDPR and HIPAA compliance with a generous free tier.

The policy detail buyers miss is that Fathom distinguishes between its third-party AI vendors and its own models. Per a ScreenApp review of Fathom's trust center, Fathom says its AI subprocessors cannot train on customer data, but separately describes using de-identified customer data to improve Fathom's own proprietary models, with an opt-out. Saying "no customer data is ever used for training" would omit that distinction.

The Jazz Security analysis of Fathom's policy (dated August 16, 2026) is more specific: Fathom may use de-identified data from meeting content for "training, improving, and customizing our in-house artificial intelligence models," and users can opt out in account settings. The same policy says Fathom does not authorize OpenAI, Anthropic or Google to train on meeting content.

A note on corporate control: Fathom was acquired by Superhuman in September 2026, and the same analysis flags that Superhuman has said meeting data "can become available to AI agents." Nothing in Fathom's terms has changed yet, but it is worth re-reading on each renewal.

Where Fathom fits for consultants

Internal calls and low-sensitivity prospect discovery, on a zero-dollar budget. The bot participant, the de-identified-training default, and the cloud storage make Fathom a harder sell on client calls covered by an NDA that bars third-party sub-processors.

Granola: bot-free notes, but still cloud-hosted

Granola is the tool consultants pick when they want AI-enhanced notes without a line labeled "Notetaker" appearing in the participant list. The Meetergo review of Granola describes it correctly: you type rough bullets, Granola listens through your mic and system audio, and it rewrites your bullets into full notes. Nothing appears in the participant list, which matters on a client workshop where an unexplained bot reads as unprofessional.

Granola's pricing re-verified in November 2026 by Scribbl's Granola pricing breakdown: Basic at $0, Business at $14 per user per month, Enterprise at $35 per user per month, all per workspace seat. The Basic tier gives unlimited meeting notes but only shows the last 30 days of history.

The confidentiality posture is mixed. On the privacy win side, the SmartCreatorHub Granola review notes that third-party AI providers such as OpenAI and Anthropic are contractually prohibited from training their models on Granola customer data, and Granola itself says it may use anonymized data to improve its own product. On the caveat side, the Convo review of Granola flags that a real privacy opt-out and org-wide privacy controls sit only on the Enterprise plan at $35/user/month, and HIPAA support is Enterprise-only. Notes and transcripts live on AWS servers in the United States.

A detail worth watching: AIBirdview's review notes that in 2026 Granola also raised a $125M Series C and picked up a proposed class action over consent — so the "no bot" positioning does not fully insulate it from the all-party-consent questions testing the rest of the category.

Where Granola fits for consultants

Relationship-heavy video work with sophisticated clients who tolerate AI-enhanced notes but would balk at a bot. A reasonable choice when your NDAs do not specifically bar cloud sub-processors — and when you are willing to budget for Enterprise if you need the full privacy controls.

Otter.ai: mature, but living under an active class action

Otter remains the most feature-mature virtual-meeting notetaker, and TicNote's consulting roundup calls it the right pick for solo consultants who live in virtual calls and want a mature, well-integrated bot. Pro pricing is around $8.33/user/month billed annually, with Business around $20.

The legal picture changes the calculus. The FindLaw docket for In re Otter.AI Privacy Litigation describes the core allegation: plaintiffs claim Otter's AI-powered meeting assistant joined meetings as a "silent participant," recording and transcribing conversations in real time without obtaining consent from all participants. Otter also allegedly enriched itself by saving the costs it reasonably should have spent by obtaining users' consent to record and use their private communications.

On August 13, 2026, Judge Eumi K. Lee issued an order granting Otter's motion to dismiss only in part. ISBA Mutual's summary for Illinois lawyers explains that voiceprints were found to be biometric identifiers under BIPA, and plaintiffs plausibly alleged that Otter collected them without written notice or a signed release. The judge also refused to assume that a visible notetaker means participants agreed to be recorded.

As the Settlement Insight tracker notes, as of November 2026 there is no certified class, no settlement, and no claim form, but Otter filed an amended answer on October 29, 2026 and the case is in discovery.

For a consultant, the operative fact is not the eventual verdict — it is that the legal theory survived. Running Otter on a client call without explicit, logged, all-party consent is now a known-risk posture. For deeper background, see our guide to the Otter.ai lawsuit and recording-consent exposure.

Where Otter fits for consultants

Established consultants who already use Otter internally, who have explicit written AI-recording clauses in their client contracts, and who announce the tool verbally on every call. For anyone still in procurement, the open litigation is a reason to look at tools that never put a bot in the room.

Basil AI: the on-device option built for NDA-bound work

Basil AI takes a different approach from every tool above: audio never leaves your Mac or iPhone. The recording, transcription (using Apple's on-device Speech framework), speaker diarization, and summary generation all happen locally on the Apple Neural Engine. There is no cloud upload, no vendor server, no AI subprocessor to disclose.

That design maps cleanly onto the confidentiality rules consultants actually operate under. If a client NDA bars third-party sub-processors (as most do), and if a cloud notetaker transmits call audio to at least one third party and usually several — the on-device path is the cleanest answer to a client-security review. For in-person client site visits and workshops where Wi-Fi is unreliable or banned, on-device capture is also the only option that works offline end-to-end; see our comparison of iPhone apps for recording in-person meetings.

Where Basil fits for consultants

Solo and small-firm consultants who handle client-confidential conversations, especially in regulated industries, cross-border engagements subject to GDPR data-minimization principles under Article 5 GDPR, or in-person client work. Also the right choice when you need one answer to the "who else hears this call?" question that survives procurement.

How Basil AI solves this: on-device vs cloud for consulting work

The core difference between Basil and the cloud options above is architectural, not a toggle. Cloud notetakers — Otter, Fathom, Granola, Fireflies — all route your client's audio to vendor infrastructure, whether or not a bot joins the call. Humla's November 2026 storage roundup confirms this directly from each vendor's policy pages: Granola and Fathom both train in-house models on de-identified data and both let you opt out in settings; Otter, Fireflies, and Granola default to US storage.

Basil, by contrast, uses Apple's on-device Speech Recognition and runs inference on the Apple Neural Engine. The recording file, the transcript, and the summary stay on your device unless you deliberately export them (e.g., to Apple Notes via iCloud, under Apple's privacy framework). For consultants, that eliminates the sub-processor problem outright: there is no third party to list on a DPA because there is no third party in the data flow.

It also eliminates the training question. No de-identification debate, no opt-out setting to maintain across every account, no Superhuman-style acquisition surprise that quietly changes what "available to AI agents" means.

For a deeper technical comparison, see how privacy-conscious teams handle AI meeting notes and our full AI meeting notetaker comparison guide.

When a cloud tool is actually fine (who should skip Basil)

To be fair: on-device is not the right answer for everyone. You should skip Basil and pick a cloud tool when:

The point is not that cloud notetakers are bad. The point is that "best for consultants" has to be answered against the contract on the table, not the feature grid.

A pragmatic decision rubric for solo consultants

  1. Pull the NDA. As HintMint's consulting guide puts it, reviewing your specific client agreements is a required step before using any AI notetaker, not an optional one.
  2. Check the sub-processor clause. If it bars disclosure to third parties without consent, a cloud notetaker is a contractual problem — not a technical one you can patch with settings.
  3. Disclose verbally on every call. Even for tools you are allowed to use. The Traq Collective field notes recommend naming the tool out loud on every call, not just in two-party-consent states.
  4. Turn off training by default. On Fathom and Granola, this is a settings toggle. Do it once per account and document when.
  5. Keep privileged and board calls off cloud tools entirely. This is the one place where the on-device path is not just preferable but necessary.

FAQ: what consultants actually ask before buying

Below is the short version — the full FAQ at the bottom of this page covers the long-tail questions buyers type into Google.

A couple more practical points consultants raise: for teams that record discovery calls and need to replay specific moments, see our writeup on iPhone on-device call recording and transcription. For the broader category question of what counts as an "AI notetaker" in the first place, we have a plain-English explainer.

The bottom line

There is no universal "best AI notetaker for consultants" because consulting engagements are not universal. The correct answer is the tool whose data-flow matches the confidentiality terms you signed.

For free internal-call capture with the clearest policy language in the cloud category: Fathom. For bot-free video on sophisticated clients: Granola. For mature live transcription when you have explicit written AI consent: Otter — with the caveat that In re Otter.AI Privacy Litigation is active and unresolved. For NDA-bound client calls, in-person workshops, and regulated-industry work where the data flow itself has to be defensible: Basil AI, because the audio never leaves your device.

Try Basil AI free

On-device AI meeting notes for iPhone, iPad, and Mac. No bot. No cloud. No sub-processor to list on a DPA.

Download on the App Store   Download on the Mac App Store

Frequently Asked Questions

What is the best AI notetaker for consultants overall?

There is no single winner — it depends on your engagement. For NDA-bound client calls, on-device tools like Basil AI avoid third-party processors entirely. For free internal-call capture, Fathom is the strongest option. For bot-free note-taking on video calls, Granola leads. For teams already standardized on virtual meetings, Otter's OtterPilot has the most mature integrations.

Can consultants legally use AI notetakers on client calls?

Legality varies by jurisdiction, but legality is only half the issue. Standard confidentiality clauses in consulting NDAs bar disclosing client information to third parties without consent, and a cloud notetaker transmits call audio to at least one third-party processor. Before using any AI notetaker on a client call, review your specific engagement's confidentiality terms and get written consent, or use an on-device tool that avoids third-party transmission.

Does Fathom train AI on my meeting data?

Yes, by default, on de-identified data. Fathom's privacy policy (dated August 16, 2026) says it may use de-identified meeting content to train its in-house AI models, with an opt-out in account settings. Fathom's AI subprocessors — Anthropic, OpenAI, and Google — are contractually prohibited from training on your data. Admins on Team plans can opt out an entire organization.

Is Granola safe for confidential consulting work?

Granola captures audio locally and does not place a visible bot in meetings, which are privacy wins. However, notes and transcripts are stored on AWS servers in the United States, and Granola trains its own models on anonymized data unless you opt out. HIPAA support and org-wide privacy controls sit only on the Enterprise plan at $35/user/month.

What is the Otter.ai lawsuit about and does it affect consultants?

In re Otter.AI Privacy Litigation (5:25-cv-06911, N.D. Cal.) is a consolidated class action alleging Otter's notetaker joined Zoom, Teams, and Meet calls without all-party consent and used audio to train its AI. On August 13, 2026, Judge Eumi K. Lee allowed core claims to proceed. For consultants, the case highlights real wiretap and BIPA exposure when bots join client calls without explicit consent.

What's the cheapest AI notetaker for solo consultants?

Fathom offers a free tier with unlimited recording and transcription, making it the lowest-cost entry point. Granola's Basic plan is also free but caps visible history to 30 days. Basil AI is a one-time or subscription iOS/Mac purchase with no per-user monthly fees. Otter's Pro plan is roughly $8.33/user/month billed annually, and Granola Business is $14/user/month.