Best AI note taker for Australian teams: what to compare
A documentation-based comparison of Otter, Fireflies, Fathom, Teams Copilot and Zoom AI, with the Australian recording law most guides skip.
Short answer: There is no single best AI note taker. The right choice depends on your meeting platform, your admin requirements and whether Australian recording consent rules apply to your calls. For a team on Microsoft Teams, evaluate Copilot first. For Zoom, try the built-in AI Companion. For mixed platforms or Google Meet, compare Otter, Fireflies and Fathom on the axes below.
This comparison is for Australian team leads, ops managers and IT decision-makers evaluating AI note-taking tools for a business with ten to two hundred staff. It is not a hands-on review. We have not tested these tools in a controlled environment and this is not a ranked list. It is a procurement comparison built from each vendor's own documentation, verified on , with unknowns marked where official pages did not confirm a claim. The Australian recording and privacy section is the part most international guides leave out.
Bot versus bot-free
The first decision is whether you want a bot joining your meetings.
A bot is a named participant (often called something like "Otter.ai Notetaker" or "Fireflies.ai Notetaker") that joins the video call and records from inside it. Everyone on the call sees it in the participant list and usually gets a notification. This is transparent, but some clients and external contacts find it off-putting.
Bot-free recording captures audio from your desktop or device without adding a visible participant. Otter offers this on Mac and Windows through its desktop app1. Fathom offers it on all platforms, labelled beta3. Fireflies offers it for Google Meet only, via a Chrome extension2. Teams Copilot and Zoom AI Companion are both bot-free by default because they are built into the platform46.
Bot-free does not mean invisible. Even when no bot joins, the tool is still recording. Australian surveillance devices legislation still applies, and the consent question does not change.
Online meetings versus in-person
All five options compared here are built primarily for video calls. If your team also runs in-person meetings that need notes, the field narrows.
Otter has a mobile app that records in-person audio from your phone's microphone1. Accuracy depends on room size, background noise and how far speakers sit from the device. Dedicated hardware like the Plaud Note Pro (roughly A$250 to A$310) records in-room meetings with better range, but that is a separate product category outside this comparison.
Fireflies, Fathom, Teams Copilot and Zoom AI Companion are designed for scheduled video calls. They do not have a meaningful in-person recording mode. If your meetings split between online and in-person, you may need two tools or a hardware recorder alongside the software.
Which platforms each tool supports
This matters more than features. A tool that does not work with your meeting platform is not an option.
| Zoom | Google Meet | Microsoft Teams | |
|---|---|---|---|
| Otter | Yes1 | Yes1 | Yes1 |
| Fireflies | Yes2 | Yes2 | Yes2 |
| Fathom | Yes3 | Yes3 | Yes3 |
| Teams Copilot | No | No | Yes (only)4 |
| Zoom AI | Yes6 | Claimed6 | Claimed6 |
Otter, Fireflies and Fathom work across all three major platforms. Teams Copilot is locked to Microsoft Teams meetings and will not even work if the meeting was created by an external participant in a different tenant4. Zoom's AI note-taking feature claims to work with third-party meeting platforms6, but we could not confirm the specifics from official documentation.
If your team uses a mix of platforms, the standalone tools are the practical choice. If you are already paying for Microsoft 365 Copilot, the built-in Teams meeting summary handles the note-taking without adding another subscription.
The comparison, documentation-based
Every cell below comes from the vendor's own website, fetched on 5 October 2026. Source numbers match the list at the end. Where we could not confirm a fact from an official page, the cell says "not confirmed".
| Otter | Fireflies | Fathom | Teams Copilot | Zoom AI | |
|---|---|---|---|---|---|
| Entry paid plan (annual) | Pro: US$8.33/seat/mo1 | Pro: US$10/seat/mo2 | Team: US$15/seat/mo3 | US$18/seat/mo add-on4 | Included in Pro: US$14.16/seat/mo5 |
| Free tier | 300 min/mo, 30 min cap1 | Limited; 400 min storage2 | Unlimited recordings3 | No free tier | 3 summaries/mo5 |
| Bot-free option | Desktop app1 | Chrome ext, Meet only2 | Yes (beta)3 | Built-in4 | Built-in6 |
| Languages | Not confirmed7 | 100+ (beta)2 | 38 for transcription11 | Not confirmed | Not confirmed |
| AI training default | De-identified data; enterprise opt-out7 | Does not train8 | Opt-out in settings9 | Does not train10 | Not confirmed |
| SSO | Enterprise only1 | Enterprise only2 | Team and above3 | Via Microsoft 365 | Via Zoom admin14 |
| Custom retention | Enterprise7 | Enterprise2 | Enterprise3 | Purview policies10 | Zero-retention option14 |
| Transcript export | txt, pdf, docx, srt1 | Download available2 | No native file export3 | Via Content Search10 | Not confirmed |
| CRM integration | Salesforce, HubSpot1 | Salesforce, HubSpot2 | Salesforce, HubSpot (Business+)3 | Via M365 ecosystem | Via Zoom integrations |
Training defaults: the row that matters most
Fireflies is the clearest: its privacy policy states it does not use customer data to train AI models and contractually prohibits its vendors from doing so8. Microsoft makes the same commitment for Copilot: prompts, responses and data accessed through Microsoft Graph are not used to train foundation models10.
Fathom sits in the middle. It may use de-identified meeting data to train its own in-house models, but you can opt out in your account settings. It does not allow third-party AI vendors (OpenAI, Anthropic, Google) to train on your data9.
Otter uses de-identified audio recordings and transcriptions for training its own AI. The enterprise plan offers an opt-out, but you need to contact an account manager to request it. It is not a self-serve toggle on lower tiers7.
Zoom's trust page was not accessible when we checked, so we could not confirm its current AI training policy from an official page. Check Zoom's Trust Centre directly before signing.
Admin controls and SSO
If your business has more than ten people, SSO is not a nice-to-have. It is how you enforce who can access meeting recordings and how you deactivate accounts when someone leaves.
Fathom stands out here: SSO is available from the Team plan at US$15 per seat per month3. Otter and Fireflies gate SSO to the Enterprise tier, where pricing is custom and you are negotiating12. Teams Copilot and Zoom inherit SSO from your existing Microsoft 365 or Zoom admin setup, so there is no additional SSO cost if your identity provider is already configured.
Seat costs at three team sizes
All prices are in US dollars, annual billing. Teams Copilot is the add-on cost only: you also need a qualifying Microsoft 365 licence underneath (see our Copilot checklist for the full calculation). Zoom AI is included in Zoom's paid plans, so the cost below is the Zoom Pro subscription itself, not a separate AI add-on.
| Team size | Otter Business | Fireflies Business | Fathom Team | Teams Copilot | Zoom Pro |
|---|---|---|---|---|---|
| 5 seats | US$100/mo | US$95/mo | US$75/mo | US$90/mo | US$71/mo |
| 10 seats | US$200/mo | US$190/mo | US$150/mo | US$180/mo | US$142/mo |
| 25 seats | US$500/mo | US$475/mo | US$375/mo | US$450/mo | US$354/mo |
Zoom Pro is the cheapest option because you are paying for the meeting platform and the AI together. If your team already uses Zoom on a paid plan, the AI note-taking costs nothing extra. The same logic applies to Teams Copilot if your team already has Copilot licences for other Office apps. The standalone tools make financial sense when your team uses a platform whose native AI is not good enough, or when you need cross-platform coverage.
Recording and privacy in Australia
This is the section that separates a procurement decision from a compliance incident. Most AI note-taker guides are written for a US audience where the legal picture is different. In Australia, there is no single nationwide consent rule for recording meetings. The law varies by state, and getting it wrong carries criminal penalties in some jurisdictions15.
State surveillance devices legislation
Australia has eight separate state and territory regimes governing when you can record a private conversation. They split into three groups15:
| Consent rule | States and territories | What it means |
|---|---|---|
| One-party consent | VIC, QLD, NT | A person who is a party to the conversation can record it without telling the others |
| All-party with exceptions | NSW, TAS, ACT | All parties must consent, with limited exceptions for lawful interests or personal use |
| Strict all-party | WA, SA | All parties must consent, no personal-use exception. Penalties up to $5,000/12 months (WA) or $15,000/3 years (SA) |
NSW and the ACT also have separate workplace surveillance legislation that requires employers to give written notice at least 14 days before starting any ongoing workplace surveillance, including the use of AI recording tools15.
When your team runs a meeting with participants in multiple states, the strictest rule applies. A call with someone in Western Australia means all-party consent, regardless of where the host sits.
The Privacy Act and OAIC guidance
The Privacy Act 1988 does not specifically cover workplace surveillance13. But the Australian Privacy Principles apply to what you do with the recording afterwards, particularly if the meeting includes anyone outside your organisation: clients, contractors, candidates or suppliers.
Entering meeting content into a third-party AI tool is a disclosure under APP 6. If the tool stores, processes or trains on that content, your organisation bears responsibility for that disclosure12.
The OAIC's guidance on commercially available AI products is direct: "Given the uncertainty of what may be discussed in a meeting, seeking consent should be pursued as a matter of best practice"12. That is the regulator's recommendation, not a suggestion from a vendor.
From 10 December 2026, a new transparency obligation requires organisations to disclose in their privacy policies any use of computer programs (including AI) that substantially support decisions affecting individuals. If your AI note-taker's output feeds into performance reviews, hiring decisions or client assessments, this applies12.
Australia also introduced a statutory tort for serious invasions of privacy in June 202516. Recording someone without their knowledge in a context where they had a reasonable expectation of privacy is a plausible, though still untested, basis for a civil claim. This is a new risk that did not exist two years ago.
What your team needs
Regardless of which tool you choose, your team needs three things before the first AI-recorded meeting:
- An AI acceptable-use policy that names the approved recording tools, what data staff can share with them, and what happens to recordings after the meeting
- A consent process: every meeting with an AI note taker starts with a clear announcement that the meeting is being recorded and transcribed by an AI tool, and participants who object can leave or request the tool be turned off
- A retention rule: how long recordings and transcripts are kept, and who can access them. This should align with the tool's built-in retention controls (see the table above)
Law firms and financial services businesses face additional obligations under professional conduct rules and APRA CPS 234 respectively. The general framework above is a starting point, not the ceiling.
How to run a team pilot

Do not roll any of these tools out to every seat on day one. A two-week pilot with five to ten people tells you more than any feature comparison.
- Pick one tool. Connect it to your primary meeting platform. If you are already evaluating Microsoft Copilot, start there.
- Run the consent process. Announce the pilot to the team and set the recording notification. If anyone objects, that tells you something about adoption, not about the tool.
- Measure what matters. At the end of each meeting, check: did the summary capture the key decisions? Were the action items accurate? Did anyone on the call ask the bot to leave?
- Check the admin controls. Can you see who accessed each recording? Can you delete a recording that should not have been kept? Can you export what you need for compliance?
- Decide. If more than half the pilot users find it useful and nobody raised a privacy concern that the policy does not cover, you have a case for a wider rollout. If the tool missed decisions, garbled names or made participants uncomfortable, try a different one.
If you are starting from scratch with AI, the pilot checklist above works for any tool, not just note takers. The Bulletproof approach is to measure the workflow before changing it.
Frequently asked questions
What is the most accurate AI note taker?
Accuracy depends on your audio setup more than the tool. Native platform options like Teams Copilot and Zoom AI Companion often handle audio better because they access the stream directly. Among standalone tools, Otter, Fireflies and Fathom all perform well in English. Accents, background noise and overlapping speakers affect every tool equally. Run a two-week pilot with your actual meetings before committing to a paid plan123.
Are AI note takers legal in Australia?
Yes, but the rules vary by state. Victoria, Queensland and the Northern Territory allow one-party consent for recording. New South Wales, Tasmania and the ACT require all parties to consent, with limited exceptions. Western Australia and South Australia require strict all-party consent. There is no single nationwide rule. The practical answer for a national team: inform every participant the meeting is being recorded and get consent1512.
Do AI note takers work for in-person meetings?
Most are built for video calls. Otter has a mobile app that records in-person audio from your phone's microphone1. Accuracy drops with distance and overlapping speakers. Dedicated hardware like Plaud Note Pro offers better in-room range but is a separate product category. Fireflies, Fathom, Teams Copilot and Zoom AI Companion are designed primarily for scheduled video conferences.
Can I use an AI note taker with Microsoft Teams?
Yes. Otter, Fireflies and Fathom all integrate with Microsoft Teams, typically by sending a bot to join the meeting123. Teams also has its own built-in option: Copilot summarises meetings and captures action items, but it requires a paid Copilot licence on top of your Microsoft 365 subscription4. If your team already pays for Copilot, evaluate that before adding another tool.
Do AI note takers train on my meeting data?
It depends on the vendor. Fireflies states it does not use customer data for AI training8. Fathom allows you to opt out in account settings9. Otter uses de-identified data for training on lower tiers, with an enterprise opt-out available on request7. Microsoft states it does not train models on meeting content10. Check the privacy policy before sharing sensitive meeting data.
Is a free AI note taker good enough for a team?
Free tiers work for trying a tool, not for a team rollout. Fathom offers unlimited recordings on its free plan but lacks action items and CRM integration3. Otter caps the free tier at 300 minutes a month and 30 minutes per meeting1. Fireflies limits storage to 400 minutes2. None of the free plans include SSO, audit logs or custom data retention, which most Australian businesses with more than ten staff will need.