//
Bot-free AI meeting notepad that combines your typed notes with AI-enhanced transcripts
Granola is a London-based AI meeting notepad that captures meeting audio through system audio rather than a bot joining the call. Users type rough notes during meetings; Granola combines those notes with an AI-enhanced transcript summary afterwards. Founded in 2023 by Chris Pedregal (ex-Google PM, founder of Socratic) and Sam Stephenson, it reached a $1.5B valuation in March 2026 following a $125M Series C led by NFDG (Nat Friedman and Daniel Gross). Granola Labs Ltd is incorporated at Companies House (14703652) with a registered office in St. Albans, UK.
Headquarters
London, United Kingdom
Founded
2023
Pricing
EU Data Hosting
No
Employees
11-50
Free
$18/mo
$25/mo
Contact Sales
Billing: monthly, annual
The meeting bot is the worst part of AI meeting tools. It joins two minutes late, announces its presence to everyone in the call, and turns a business conversation into something that feels surveilled. Granola removes the bot entirely.
Granola Labs Ltd (Companies House 14703652, incorporated March 2023, registered St. Albans) is a London company building an AI meeting notepad for macOS, iOS, and Windows. The product captures audio through system audio rather than a call participant, meaning no bot joins, no recording notification fires, and no permissions need to be granted at the meeting platform level. During the call, the user types rough notes as usual. After the call ends, Granola merges those notes with an AI-enhanced transcript summary, producing structured output that reflects both what was said and what the user considered important enough to write down.
The commercial trajectory has been rapid. Chris Pedregal (CEO, formerly a product manager at Google and founder of Socratic, which was acquired by Google in 2018) and Sam Stephenson (CTO) founded the company in early 2023. By October 2024 the company had closed a $20 million Series A. A $43 million Series B followed in May 2025. In March 2026 NFDG (the investment vehicle of Nat Friedman and Daniel Gross) led a $125 million Series C at a $1.5 billion valuation, bringing total raised to $192 million.
The valuation places Granola among the most highly capitalised AI meeting tools globally, sitting above Fireflies.ai, Otter.ai, and at comparable territory to the upper range of tldv, the Aachen-based AI meeting platform that takes a different architectural approach. At this funding level, Granola's investors are betting on an enterprise expansion story that goes beyond individual meeting notes into AI memory across the full history of a team's meetings.
Granola captures audio through the operating system's audio layer, not through a call API or bot participant. On macOS, this uses Core Audio; on Windows (supported from June 2025), WASAPI. The user grants Granola microphone access once during setup, and the tool runs in the background across any meeting platform (Zoom, Google Meet, Microsoft Teams) without requiring integration with those platforms.
The practical consequences are significant. Meeting participants see no recording notification. The host does not need to admit a bot from the waiting room. There is no dependency on Zoom's or Google's API behaviour, which means Granola works identically on calls that other tools cannot record: calls where the host has disabled bots, calls on platforms without bot APIs, and calls on video conferencing software used primarily in regulated industries.
The limitation is that bot-free capture requires a local device. Granola cannot capture audio from calls the user is not actively on. There is no delegation model where a manager sends Granola to record a meeting they missed. If passive or delegated recording matters, a bot-based tool like Fireflies or tldv is structurally better suited.
Most AI meeting tools produce a full transcript and layer a summary on top. Granola takes a different approach: it treats the user's own typed notes as the primary signal and uses the AI transcript to enrich them.
During the meeting, the user types notes exactly as they normally would: partial sentences, action items, key decisions, anything worth capturing. After the meeting ends, Granola processes the audio transcript and produces a summary that incorporates both the transcript content and the user's notes. The output reflects what the user considered important, not just what was said most often. A 60-minute meeting where three people discussed a minor scheduling issue for 20 minutes will produce notes that reflect the actual decision taken, not a transcript proportional to speaking time.
The quality of the merge depends on note-taking density. Users who take detailed notes produce better Granola summaries than users who type nothing during the call. This is a genuine usage pattern requirement: the tool rewards engaged note-takers and is less valuable for users who take no notes at all.
Granola supports user-defined templates that structure the AI summary output. A sales call template might capture discovery questions, objections raised, next steps, and decision criteria. An engineering standup template might capture blockers, completed work, and dependencies. A board meeting template might capture motions, votes, and action owners.
Templates are configurable per meeting type and can be shared across a Business tier team, giving organisations a consistent output format across everyone using Granola. For teams that standardise on a meeting methodology, such as MEDDIC for sales or OKR check-ins for planning, the template layer converts individual note-taking into a structured knowledge output.
The Individual Pro and Business tiers include a chat interface that lets users query their full meeting history. A question such as "What did we decide about the pricing model in March?" surfaces the relevant meeting summaries and quotes the specific passage. For executives running multiple workstreams across dozens of meetings per month, this is meaningfully different from searching through a folder of documents. The AI understands the semantic content of meetings, not just keywords.
The enterprise roadmap extends this into proactive memory: before a meeting with a given customer, Granola would surface relevant prior conversation history, open action items, and relationship context. At Series C scale, this enterprise memory feature appears to be the primary value creation thesis for the next product phase.
Granola shipped iOS in April 2025 and Windows in June 2025, alongside the original macOS client. Cross-platform coverage matters for enterprise procurement: IT teams will not deploy a tool that requires all employees to use Mac. The Windows client uses WASAPI for system audio capture, maintaining the bot-free architecture on Windows without requiring platform-specific recording APIs.
The absence of a Linux client is a meaningful gap for engineering organisations where a portion of the team runs Linux. There is no indication that Linux support is on the near-term roadmap.
Granola's free tier allows 3 meetings per month, enough for a meaningful product test but not for professional evaluation at real meeting frequency. Moving to Individual Pro at $18 per month unlocks unlimited meetings, custom templates, the meeting history chat interface, folder organisation, and shareable summary links.
For context, Trint (another London transcription company) charges per transcription hour on its paid plans. Granola's per-month model is simpler for users with variable meeting loads.
Business at $25 per seat per month adds team folders, admin controls, SSO, and priority support. For a 10-person team, this is $250 per month, lower than comparable per-seat pricing from Fireflies.ai at its Business tier, and significantly lower than Otter.ai's enterprise pricing.
Enterprise pricing is custom, with the important additions of EU data residency, custom data retention policies, a dedicated customer success manager, and on-premises deployment on the roadmap. For regulated industries or organisations with strict data governance requirements, Enterprise is the only tier where Granola can credibly meet procurement requirements around data hosting location.
Granola holds SOC 2 Type II certification and provides a Data Processing Agreement. Granola Labs Ltd operates under UK GDPR. The company does not train AI models on customer audio without explicit consent and offers a zero data retention option that deletes audio and transcript data after processing.
The significant caveat for EU procurement teams is data residency. Granola's default infrastructure stores data outside the EU for most pricing tiers. EU data residency is available at the Enterprise tier through negotiation, not as a standard feature of Business or Individual Pro plans. Teams with GDPR Article 46 data transfer obligations should not assume EU residency without confirming it at contract stage.
The zero data retention option and on-device transcript processing (on roadmap) address privacy concerns about audio storage, but they are distinct from the question of where data is processed and stored. For most B2B SaaS customers operating under standard GDPR compliance, the UK adequacy decision covers the UK-to-EU transfer question, and Granola's SOC 2 Type II provides a credible security baseline. For regulated industries such as healthcare, financial services, and legal, the Enterprise tier with explicit EU residency is the appropriate path.
The ai-meeting-notes category is filling with European options. tldv (Germany) and Trint (London, Companies House 09225083) are both credible alternatives with different trade-offs: tldv for bot-based recording with CRM integration, Trint for transcription-focused workflows. For note-taking adjacent to meetings, Capacities (St. Wendel, Germany) offers a broader personal knowledge management environment that some users find more useful than meeting-specific tooling.
If you attend 10-20 meetings per week and the primary pain point is capturing decisions and action items accurately, Granola's note merge model produces higher-quality summaries than pure transcription tools because it incorporates your own judgment about what mattered.
If you run meetings where bot presence is unwelcome, such as investor calls, sensitive client conversations, or regulated industry discussions, the bot-free architecture is the only viable approach among AI meeting tools.
If you need full meeting recordings, delegated recording of meetings you did not attend, or a video clip library for sharing, Granola does not support those use cases. tldv or Fireflies handle them.
If your team requires EU data residency at a standard pricing tier, Granola is not currently the right choice. Enterprise negotiation is required to confirm data hosting location.
Granola's bot-free architecture is a genuine product insight, not a positioning choice. By removing the bot, it removes a friction point that every other AI meeting tool accepts as given. The note merge model produces more relevant output than transcript-only approaches. The commercial trajectory ($192M raised, $1.5B valuation) reflects real product-market fit. The limitations are real too: the free tier is too restrictive for proper evaluation, EU data residency is Enterprise-only, and the product rewards users who already take notes rather than those who hope AI will do it all. For professionals who take notes and want AI to make those notes better, Granola is the strongest option in its category.
No. Granola captures audio through system audio on macOS, iOS, or Windows; it never joins a call as a participant. Other meeting attendees see no recording notification and the meeting experience is uninterrupted. This is Granola's primary differentiator against Fireflies, Otter.ai, and most other AI meeting tools.
Yes. Granola Labs Ltd (Companies House 14703652) is a UK company subject to UK GDPR, which mirrors EU GDPR. Granola holds SOC 2 Type II certification, offers a zero data retention option, and does not train AI models on customer audio without consent. A Data Processing Agreement is available. EU data residency requires the Enterprise tier.
Both are European AI meeting note tools. tldv (German, tldx Solutions GmbH) uses a meeting bot and focuses on recording and clip-sharing for sales teams. Granola is bot-free, focuses on note enhancement rather than full recording, and is better suited to executives and product teams who want enriched notes without capturing full video. For sales teams who need shareable clips and CRM integration, tldv is likely the stronger fit.
Granola runs on macOS, iOS (April 2025), and Windows (June 2025). There is no Linux client. The tool captures system audio, so it works with any meeting platform (Zoom, Google Meet, Microsoft Teams) without requiring a bot integration or API access to those platforms.
Granola defaults to cloud storage outside the EU for most plans. EU data residency is available at the Enterprise tier. The zero data retention option deletes audio and transcripts after processing. On-device transcript processing is on the roadmap. UK entity domicile means UK adequacy applies for EU-to-UK data transfers.