Transcript
Formed with speaker separation while the meeting runs. Because it is searchable, “when did we discuss this?” is answered in seconds.
// NIJITECH PRODUCT · MEETING COACH
nijibot
An AI coach that listens to meetings, summarises them and turns them into action.
nijibot is an AI meeting assistant that listens to your meetings in real time: it transcribes what is said live, summarises it and splits it into clear action items. It does more than take notes — it coaches while the meeting runs. Every output stays on-premise or in the cloud, explainable and auditable.
Product site:nijibot.io
In short
nijibot is an AI meeting assistant and coach that listens to meetings in real time. It produces a live transcript with speaker separation, keeps a summary that grows as the meeting goes on, and splits what was said into action items with an owner and a date. Data is processed on-premise or in the cloud, and every summary item is tied to its source in the transcript.
Live meeting console — the demo where the product proves itself
Transcript
Ayşe
We finished the onboarding flow last sprint — let us look at the metrics this week.
Mehmet
Activation went from 18% to 24%, but there is a drop at step 2.
You
Shall we simplify step 2 and open an A/B test?
Ayşe
Sure — let us update the copy by Friday.
Live summary
- Onboarding shipped · activation 18% → 24%
- Drop detected at step 2
- Decision taken to run an A/B test
Action items
- Update the step 2 copy — MehmetFriday
- Set up the A/B test — Youthis week
Speaking balance: you 71% · team 29% — ask more questions.
3 action items extracted · 0 notes missed
01 — HOW IT WORKS
The niji core pipeline: listen, understand, produce, coach.
Every one of these steps runs on the same backbone: the niji core pipeline →
01Listen
Captures meeting audio in real time and separates the speakers (online & offline).
Every one of these steps runs on the same backbone: the niji core pipeline →
02Understand
niji core turns audio into text, extracts the context and marks the moments that matter.
Every one of these steps runs on the same backbone: the niji core pipeline →
03Produce
A live summary, a decision list and owned action items — ready as the meeting ends.
Every one of these steps runs on the same backbone: the niji core pipeline →
04Coach
Gently flags speaking balance, question quality and action items nobody followed up on.
02 — FEATURES
From meeting to output, in one flow.
Live transcription & speaker separation
Every sentence becomes text instantly, labelled with who said it.
A summary that grows in real time
The summary expands live as the conversation goes on and is ready at the end.
Automatic action extraction
Every action item is extracted automatically with an owner and a date.
Meeting coaching
Gentle live prompts on speaking balance and question quality.
Calendar & tool integration
Connected end to end with Google/Microsoft · Slack · Jira.
Explainable output
Every summary and action item is tied to its source in the transcript.
Data processing region
With an on-premise deployment the data stays inside your own infrastructure and does not leave it. In cloud use, processing takes place in the US and EU regions; KVKK Art. 9 safeguards apply to cross-border transfers. The full list of sub-processors is in Exhibit C of the Data Processing Addendum.
KVKK & GDPR
Data minimisation, retention control and an auditable record.
Human-approval step
Critical outputs pass human approval before they go out.
// OUTPUT
A recording is one thing; an output is another.
A tool that merely records leaves you an hour of audio — it postpones the problem rather than solving it. nijibot produces four separate outputs from a meeting, each usable on its own.
A growing summary
Ready when the meeting ends; it does not wait for post-processing. Every item is tied to its source in the transcript.
Owned action items
Who, what, by when. An action item without an owner cannot be followed up; nijibot extracts all three together.
Coaching note
Speaking balance and items nobody followed up on — a measurement of how the meeting went, separate from what was said in it.
// INTEGRATIONS
It starts where the meeting starts.
nijibot does not expect you to open another app. It connects to the calendar, joins the meeting on its own and drops the output into the tools the team already uses.
Calendar
Works with Google and Microsoft calendars. When it sees the invite it joins — nobody has to say “start recording”.
Slack
The summary lands in the relevant channel as soon as the meeting ends, so whoever missed it can catch up by reading.
Jira
Action items can flow into Jira with their owner and date, closing the gap between a decision and a work item.
// WHERE IT PAYS OFF
One meeting, different needs.
What a meeting record is worth depends on the role. nijibot serves three different needs from one recording.
Product and project teams
A decision made in a sprint meeting is forgotten by the end of the week. When action items come out with an owner, the follow-up becomes a progress check rather than a repeat.
Customer calls
What was promised on a call later becomes a matter of dispute. Because the summary is tied to its source in the transcript, the record settles it.
Managers
Instead of listening to a meeting you missed end to end, you read its summary. The speaking-balance measurement also shows who dominated the discussion.
// DATA
A meeting recording is sensitive data.
A meeting recording can contain customer names, prices, performance judgements and decisions not yet announced. That is why nijibot keeps its output verifiable: every summary item is tied to its source in the transcript, human approval is requested at critical steps, and processing happens in the data processing region.
Retention period, access rights and deletion policy are defined at setup — these are the organisation’s decision, not a product default.
03 — TRUST & COMPLIANCE
The enterprise trust layer, on by default.
on-premise or cloud processing preferenceExplainable AIHuman-approval stepKVKK & GDPR
Measurement summary
0M+
meeting minutes processed
%0
notes missed
Minutes
recovered per meeting
FREQUENTLY ASKED QUESTIONS
What people ask about nijibot.
How does nijibot transcribe meetings?
nijibot listens to the meeting in real time and turns speech into text as it happens. Because it separates speakers, the transcript shows who said each sentence. The transcript forms while the meeting is still running — there is no need to wait for it to end.
Which tools does nijibot integrate with?
It integrates with Google and Microsoft calendars, and with Slack and Jira. Thanks to the calendar integration it joins the meeting on its own; the resulting action items can go to Jira and the summaries to Slack channels.
Can you see where a meeting summary came from?
Yes. In nijibot every summary item is tied to its source in the transcript — you can see which part of the conversation it came from. This is the practical form of the explainable-output principle: you can challenge an output and ask for its reasoning.
Where is meeting data processed?
It depends on the deployment model. With an on-premise installation the data stays inside your own infrastructure and does not leave it; in cloud use, processing takes place in the US and EU regions and KVKK art. 9 safeguards apply to cross-border transfers. The full list of sub-processors is published in Exhibit C of the Data Processing Addendum. The product is designed with KVKK and GDPR requirements in mind.
What does “meeting coaching” mean?
nijibot does more than take notes. While the meeting runs it tracks speaking balance and reminds you of action items nobody followed up on. The intervention comes during the meeting, not after it.
Are action items extracted automatically?
Yes. nijibot extracts action items from what was said and assigns an owner and a date to each. Human approval is in place at critical steps; the resulting actions can be reviewed before they are applied.
How do people who missed the meeting see the summary?
The summary can land in the relevant Slack channel as soon as the meeting ends. Whoever missed it catches up by reading rather than listening from the start, and because action items appear with their owner, nobody has to ask who does what.
How long is a meeting recording retained?
Retention period, access rights and deletion policy are defined at setup — these are the organisation's decision rather than a fixed product default. Because a meeting recording can contain customer names, prices and performance judgements, these settings need to be agreed up front.
Can I correct an action item that was extracted wrongly?
Yes. Action items can be reviewed before they are applied, and because each one is tied to its source in the transcript you can see where it came from. Spotting a wrong inference is far harder when the reasoning is not visible.
For more, see the general FAQ · for terminology, theglossary
04 — GET STARTED
Let us bring this product into your work.
Let us connect nijibot to your team’s meeting flow and see your first value within days.
Talk to the team →