Every meeting becomes a connected note.
Otter records. Granola pretty-prints. Knovya is what comes after — the structured note that lives in your knowledge base, links to the decision it produced, and shows up where Claude can read it. Bring transcripts from any meeting tool or capture them by voice; if you want an empty scaffold instead of an AI workflow, see Note Templates.
Pick a source. Watch the meeting build.
Four input sources — voice, an Otter transcript, a ChatGPT chat, a manually pasted board summary. Six lenses that turn raw words into the artifacts you actually use. Toggle anything; the note rebuilds.
Same meeting, different surface. 6 of 6 lenses on
Decisions
block · callout
Action items
transform · extract_actions
Attendees
mention · linked people
Linked notes
graph · noterank surfaced
Follow-ups
smart-archive · deferred
MCP visibility
routing · agent-aware
Available to Claude · Cursor · ChatGPT tool: knovya_search
What just happened. A voice recording became a structured note with six surfaces. Knovya read the words once — every block is sourced from the same transcript, structured by a different capability.
Free includes paste-and-structure across all four sources. Pro unlocks voice transcription and MCP routing → See Pro
Ten capabilities, four stages.
Capture turns words into text. Structure turns text into a note. Connect turns the note into knowledge. Route makes that knowledge findable — by you and by every AI you reach for.
Stage I · Capture Catch the words
3 capabilities 01
Voice transcription Hit record on mobile or desktop. Knovya transcribes live, end-of-call gives you a clean text — ready to be structured. The microphone goes straight into the note.
02
Paste & ingest Drop in a transcript from Otter, Granola, Fireflies, Zoom, Teams, Google Meet, or any other meeting tool. Knovya does not care where the words came from; it cares what to do with them.
03
File attach PDF transcripts, audio files, video recordings — attached to a note and processed the same way. The meeting that already happened is not lost to you.
Stage II · Structure Make it readable
3 capabilities 04
Conversation → Note A multi-turn AI session that decides the headings, the order, and what to keep. Long transcripts collapse into a navigable shape — readable in a minute, not an hour.
05
Action item extraction An AI Transform that finds the verbs. Decisions become callouts. Owners become assignees. "Let’s circle back next Friday" becomes a follow-up with a date attached.
06
Outline & summary A one-paragraph overview at the top, navigable headings beneath. The shape of the meeting, before the words of the meeting — for the colleague who could not attend.
Stage III · Connect Wire it into the graph
2 capabilities 07
Knowledge graph linking The meeting note finds related notes by itself — projects mentioned, prior decisions, people in your workspace, the document that was being reviewed. The graph builds, you do not have to.
08
NoteRank surfacing Important meetings rise to the top of your home feed. The board call from last week beats the standup from yesterday — because ranking is not the same as recency.
Stage IV · Route Make it agent-aware
2 capabilities 09
MCP routing Claude, Cursor, ChatGPT — every agent that asks "what did we decide about pricing in March?" finds the meeting note. The decision returns, sourced, with the room it was made in.
10
Templates Standup, 1:1, retro, board meeting, kickoff — each meeting type carries its own structure. Pick one and the note arrives shaped; skip it and Knovya picks the closest.
The transcript was the easy part. Everything after it is where meetings break.
Otter records. Granola pretty-prints. Fireflies emails you a summary. Then the meeting closes — and the transcript is sealed in a SaaS you only open when you need to relive the call.
The decision is forgotten by Friday. The action item lives in Notion if someone remembered to copy it. The "we'll talk about that next sprint" disappears with the calendar invite. The meeting tool's job ends where the work begins.
- The cost
- Knowledge workers attend more meetings every year than the year before. The ratio of decisions retrieved to decisions made keeps falling.
- The fix
- Stop treating the transcript as the deliverable. Start treating it as the input.
From recognition to routing.
Speech recognition has been getting better for seventy-four years. The gap was never the words — it was where the words went after they were captured.
- 1952Bell Labs Audrey The first speech recognition system. Could understand spoken digits — one speaker, ten words. Speech became something a machine could read. Bell Labs · signal processing
- 1971DARPA SUR — CMU Harpy Continuous speech, around a thousand-word vocabulary. The recognition problem became tractable; the next problem became what to do with the transcript. DARPA · CMU · pattern recognition
- 2009Google Voice transcription Voicemail became searchable text. Speech-to-text moved from device to cloud — and from research to product. The first time most people read a machine reading them. Google · cloud ML · mobile UX
- 2022OpenAI Whisper Multilingual, open-source, Transformer-based. The transcript became cheap and abundant — every meeting could be transcribed. The bottleneck moved one step downstream. OpenAI · deep learning
- 2026Knovya AI Meeting Notes The transcript stops being the output. It becomes the input — to a structured, linked, MCP-readable note that lives in your knowledge base for as long as the decision matters. ★ Knovya · production
First of its kind
Everyone makes a transcript. Nobody connects it to your knowledge.
Otter records. Granola pretty-prints. Fireflies sends a summary email. Read.ai measures sentiment. tl;dv saves you watching the recording. Fathom highlights the moments. Zoom puts AI inside Zoom. Every one of them ends where the call ends. None of them gives the meeting a place to live afterwards — a place where it can link to the decision it produced, surface in your home feed when it matters, or be readable by Claude when you ask what did we decide about pricing in March.
- Otter meeting bot · transcript
- Granola local capture · markdown
- Fireflies meeting bot · summary email
- Read.ai meeting bot · analytics
- Fathom meeting bot · highlights
- tl;dv video clip · transcript
- Zoom AI Companion host-tied · in-Zoom only
- ★ Knovya any source · linked note · agent-readable
A meeting note is a living surface.
Once Knovya structures it, the meeting shows up in four places — none of which you have to visit on purpose.
Title, attendees, decision callouts, action items with owners, linked notes, follow-ups — rendered as a Knovya note you can edit, share, and version.
Important meeting notes rise to the top — the board call from last week beats the standup from yesterday. NoteRank decides which meetings still matter.
When Claude or Cursor calls knovya_search with
"what did we decide about pricing", the meeting note
returns — decision callout, action items, the room it was made
in.
Every meeting that produced a decision shows up retroactively in your decision log — sourced, dated, and linked back to the room it came from.
AI Meeting Notes composes with the rest of Group I.
07
Voice
The microphone path. Speak it, Knovya transcribes, the meeting
note begins.
09
Conv \u2192 Note
For the meeting that happened in a ChatGPT chat. Same
structure, different room.
04
Transforms
Extract actions. Summarize. Translate. The 10 actions a meeting
note needs.
05
Skills
Build a standup workflow that runs the same way every Monday.
A few honest answers.
What is an AI meeting note taker?
An AI meeting note taker is a tool that turns a meeting — voice, video, or transcript — into structured notes you can read, search, and act on. Most tools focus on the transcript itself. Knovya focuses on what happens after: the transcript becomes a connected note that links to decisions, action items, and the rest of your knowledge.
How is Knovya different from Otter, Granola, Fireflies, or Read.ai?
Otter, Granola, Fireflies, and Read.ai compete on the meeting itself — bots that join your call, transcribe, and email you a summary. Knovya does not run a meeting bot. It is the layer above: bring a transcript from any of those tools (or your own voice recording), and Knovya turns it into a structured note that lives in your knowledge base, links to decisions, and is readable by Claude or Cursor through MCP.
Does Knovya have a meeting bot that joins my Zoom?
No, and that is intentional. Bot-in-call is a crowded space — Otter, Granola, Fireflies, Read.ai, Fathom, Zoom AI Companion all do it well. Knovya is what they all should connect to: the place the meeting goes after the call ends. You keep your bot of choice. Knovya gives the output a permanent home.
What sources can I bring transcripts from?
Voice recordings from inside Knovya (mobile or desktop, see Voice Notes), pasted transcripts from Otter, Granola, Fireflies, Zoom, Google Meet, Teams, or any other meeting tool, ChatGPT or Claude conversations where the meeting was effectively held (see Conversation → Note), and uploaded files — audio, video, PDF transcripts. Whatever has the words, Knovya can structure it.
Can Claude or Cursor read my meeting notes?
Yes, through MCP — the Model Context Protocol. When Claude is asked something like "what did we decide about pricing in March?", it can call Knovya's search tool and retrieve the relevant meeting notes. The decision and its context come back to the conversation. This is included on Pro.
Does it work without a meeting bot?
Yes — that is the default flow. Record voice, paste a transcript, attach a file. No bot has to be invited to the call, no calendar permissions are required, and meetings that happened before you signed up can still be ingested by pasting their transcripts.
Is it on the Free plan?
Paste-and-structure is on Free — bring any transcript and Knovya turns it into a meeting note with decisions, actions, and links. Voice transcription, MCP routing for agents, and Skills-driven workflows are on Pro. Looking for an empty meeting template instead of an AI workflow? See Note Templates — Knovya ships standup, 1:1, retro, PRD and decision-log scaffolds you can fill in by hand.
Bring your next meeting to a place that remembers.
Free for paste-and-structure across every meeting tool you already use. Pro adds voice transcription and MCP routing for Claude, Cursor, and ChatGPT.
Mn element 10 · Group I — AI