Your AI Meeting Assistant Is Taking Notes. Who Is Doing the Work?
AI meeting notes are now a commodity. The real value is turning decisions into owned tasks and keeping context across meetings. Six questions to judge any tool.
Blog
Breakdowns, templates and practice: how to run meetings so that decisions are what remains afterwards, not a sense of wasted time.
Daily standup: how to run one that doesn't turn into a ritual
What a daily standup is and how it differs from a meeting: four formats with timing, a ready list of questions and what to capture so the time is not wasted.
How to run a meeting that ends in decisions, not in talk
How to run meetings that end in decisions: preparation, who to invite, roles and timing, handling disagreement, the record afterwards and four antipatterns.
System design interview: how the session actually works
What a system design interview actually measures: a seven-step answer framework with timing, questions to ask first, four classic problems and where candidates fail.
Internal knowledge base: how to build one people actually use
An internal knowledge base answers “how we do things here and why”. What belongs in it, why a structure by department falls apart, and how to launch one in a week.
Meeting summary: what belongs in it and how to get one
What a meeting summary is, how it differs from minutes, a ready template and the follow-up email — plus what to do when the built-in summary never arrives.
AI in interviews: where help ends and cheating begins
Real-time prompts, an overlay invisible on screen share, and the line where help turns into deception. How the mechanism actually works, what really gives a candidate away, and where AI belongs.
A live interpreter for Zoom, Google Meet and calls: speak your own language
Real-time conversation translation: why latency matters more than vocabulary, how to speak your own language on any call — and where the limits are.
An AI interview assistant: prompts during the interview and a broken hiring market
What AI can do before, during and after an interview, where help turns into substitution, and why interviews are a poor measuring instrument. No moralising.
An AI assistant for developer meetings: code, tasks and decisions that survive
Three things were decided in the meeting; two days later nobody remembers the second. Why a transcript isn't enough and how a decision reaches a task with its reason intact.
AI lecture notes: turning a recording of a class into notes you can study from
A transcript is not yet a set of notes. The seven blocks that make it study material, how a recording becomes structure, and what still remains your own work.
From a recording to minutes that work: decisions, tasks and commitments
A transcript, a summary, minutes and an outcome are four different things. How to get six working blocks out of a recording — and why owners and deadlines still need a human eye.
AI finds the tasks and commitments after a meeting — nothing gets lost
The path from «agreed» to «done» breaks in six places. What a machine can find in a conversation, what it must not decide for you, and why a recording alone is not enough.
What an AI meeting assistant is, how it works and why you need one
How it differs from a recorder and a transcriber, how the pipeline from audio to tasks works, where it fails, and why you need one when ChatGPT exists.
Why AI forgets, and how memory works in Whisperer
A context window is working memory, not storage. The engineering problem in full: entry kinds, half-lives, dedup, retrieval budget, and what we deliberately never record.
Meet Leo, the AI agent
Leo is the AI agent inside Whisperer: it sees your meetings, tasks and calendar, and acts in the same places. 107 tools, commitments extracted from meetings with a source, and boundaries no setting removes.
Meeting notes without a bot: three mechanisms and what each one cannot do
A browser extension, an official platform API and OS audio capture. How each works, what it fails to give you, and which to pick — including the price nobody states.
MCP server for meetings: what it is and how to connect Claude, ChatGPT and Cursor in five minutes
What an MCP server is, why meetings need one, and how to connect Claude, ChatGPT or Cursor to your calls in five minutes: what leaves, what never leaves, and how the permissions are built.
Agent skills explained: what SKILL.md is and how to install one without a terminal
SKILL.md is a plain-text instruction an AI assistant reads before a task. What is inside it, how it differs from MCP, how to install one without a terminal, and what to check first.
Meeting minutes: what they are, how to structure them, and how to take them
Meeting minutes are a short record of what was decided, who owns it and by when. Seven sections, how much detail to keep, a template and a builder.
Meeting agenda: what belongs on it, how items get submitted, six formats
A meeting agenda is a list of items, each with a verb, an expected outcome and a time box. Selection criteria, a skeleton agenda, six formats and a checklist.
Transcripts as a language-learning tool: reviewing your own calls
How to use a transcript of your own speech to spot recurring mistakes, poor constructions and filler words.
Structured interviews: assessing candidates from the transcript, not from impressions
Why memory fails you on the fourth interview in a row, and how to reduce assessment to comparable criteria instead of gut feel.
Customer discovery: pulling insights out of 20 interviews without drowning
A working scheme: from transcripts to coded quotes, from codes to patterns, from patterns to product decisions.