Field notes.
On voice AI, ambient teammates, and the end of typing.
One post a day, by Hyunyoung Kim.
-
Comparison
Google Meet AI notes vs a real meeting teammate
Google Meet's AI notes summarize the call after it ends. A meeting teammate works during the call: live research, answers, and follow-ups pushed to your tools before you hang up.
-
Analysis
Meeting AI ROI: how to measure the hours you actually save
Meeting AI ROI is the time your team stops spending on notes, follow-ups, and re-explaining decisions, minus the cost of the tool. Here's how to measure it honestly, including the costs most people miss.
-
Adoption
How to get your team to actually use meeting AI
Teams adopt meeting AI when it does the work without being asked, not when it adds another tool to learn. Here's the rollout pattern that makes it stick, one painful meeting at a time.
-
Comparison
Microsoft Teams Copilot for meetings: what it does and doesn't do
Teams Copilot recaps your meeting and answers questions about the transcript, all after the conversation moved on. Here's where it helps, where it stops, and when you need a live teammate instead.
-
Buyer's guide
AI meeting software buyer's guide (2026 edition)
The right AI meeting software depends on one question: do you want a transcript, a recap, or a teammate? Here's how to evaluate, compare, and buy in 2026.
-
Integrations
Slack, Notion, Linear: where meeting AI output should land
Meeting AI output should land where the work already lives: decisions in Notion, action items in Linear, the heads-up in Slack. Here's how to route each one.
-
Examples
Beyond transcripts: 10 things meeting AI should actually do
A transcript is the floor, not the ceiling. Beyond it, meeting AI should research live, capture decisions, draft follow-ups, and route the work to your tools.
-
Framing
AI teammate vs AI assistant: why the distinction matters
An AI assistant works for one person and waits for prompts. An AI teammate works with a team and acts on shared context. The distinction decides what your tool can actually do.
-
Persona
The 5-meeting startup: how founders actually use AI teammates
The 5-meeting startup runs the whole week on five recurring calls and lets an AI teammate carry the rest. Here's the actual cadence founders use in 2026.
-
Thought leadership
Prompt fatigue is real: what it is, and how voice AI ends it
Prompt fatigue is the slow burnout from typing, re-typing, and babysitting AI all day. Here's what it is, why it's spreading in 2026, and how voice AI ends it.
-
Workflows
The AI meeting follow-up workflow that saves 3 hours a week
The AI meeting follow-up workflow that saves 3 hours a week runs while the call is still happening, not after. Here's the five-step pattern teams use to make it stick.
-
Examples
Real-time research in meetings: 12 examples from real teams
Real-time research in meetings is when AI pulls the answer your team needs while the conversation is still happening. Here are 12 examples from real teams.
-
Trust & legal
Meeting recording consent: what's legal where in 2026
Meeting recording consent depends on where every participant sits, not where you sit. Here's the 2026 map of one-party, all-party, and GDPR rules for team calls.
-
Operating practice
Remote vs hybrid meetings: where AI teammates actually help
Remote meetings need a teammate that closes the silence. Hybrid meetings need one that levels the room. Here's where AI actually helps, and where it gets in the way.
-
Operating practice
AI decision logs: how to never lose a meeting outcome again
An AI decision log is the short record of what your team decided, who owns it, and when. Here's how to build one that survives the week, not just the meeting.
-
Design
When AI should stay quiet in a meeting (and how we decide)
The best meeting AI talks less than a junior teammate. Here are the five moments voice AI should stay quiet, the three it should speak, and the rule we use to draw the line.
-
Comparison
relly vs Fathom: what's different, what's not
Fathom records, summarizes, and recaps your meetings after they end. relly joins live and does the work during the call. Here's what's actually different.
-
Comparison
relly vs Granola: when each is the right pick
Granola polishes your notes after a meeting. relly joins the meeting and works during it. Here's when each is the right pick for your team's workflow.
-
Trust
Voice AI privacy: what actually happens to your meeting data
Voice AI privacy comes down to three things: who hears the audio, where the transcript lives, and how long it sticks around. Here's the honest map.
-
Templates
The perfect AI meeting minutes template (and why most are wrong)
The best AI meeting minutes template captures decisions, owners, and dates in that order. Most templates copy old paper formats that buried the work.
-
Thought leadership
Deep work vs meetings: the false trade-off in the AI era
Deep work vs meetings is a false trade-off. The real cost isn't the meeting block, it's the cleanup shift on either side. Voice AI removes that shift.
-
Templates
How to brief an AI teammate: 5 templates that work
Brief an AI teammate the way you'd brief a new hire: goal, context, constraints, output. Five copy-paste templates that turn vague prompts into useful work.
-
Thought leadership
Typing is the new bottleneck: why voice-first AI wins
Typing is the new bottleneck because every prompt costs a context switch, a wait, and a re-entry. Voice-first AI takes the keyboard out of the loop.
-
Concepts
What "agentic AI" actually means for knowledge workers
Agentic AI is software that decides, acts, and follows up without being prompted at every step. For knowledge workers, it's the line between a tool and a teammate.
-
Concepts
Continuous team memory: the missing layer in team AI
Continuous team memory is the shared layer that lets AI remember what your team decided last week, last month, and last quarter. It's the missing piece that turns prompt tools into teammates.
-
Thought leadership
The two-shifts problem: why AI slows teams down today
The two-shifts problem is the new tax AI quietly added to knowledge work: a meeting shift, then a cleanup shift to feed, prompt, and edit the AI. Here's how to spot it and end it.
-
Operating practice
Meeting hygiene: 10 rules that save hours every week
Meeting hygiene is the small set of rules that keeps a recurring meeting from rotting. Ten that take ten minutes to adopt and save hours every week.
-
Design
How AI should behave in a real meeting: a design guide
AI in a meeting should listen by default, speak only when it has something the room actually needs, and act without asking for a prompt. Here's the design guide we use.
-
Comparison
Why Zoom's AI Companion isn't enough for real team meetings
Zoom's AI Companion writes a tidy recap after the call, but the meeting that needed help is already over. Real teams need an AI that participates while the room is still talking.
-
Comparison
AI meeting notetakers compared: Granola, Fathom, Circleback, Otter, and relly
The best AI notetaker for your team depends on whether you want a transcript, a recap, or a teammate. Here's how the five most-asked-about tools actually differ in real meetings.
-
Analysis
The hidden cost of prompt-based AI in team workflows
Prompt-based AI looks free, but every prompt costs your team a context switch, a wait, and a re-entry. Here's what that actually adds up to in a week of meetings.
-
Operating practice
How to run meetings that actually decide things
A meeting decides something when one person owns the call, the question is sharp, and the team leaves with an action. Here's how to run that meeting every time.
-
Concepts
What is ambient AI, and why it matters in 2026
Ambient AI is software that listens, watches, and acts in the background of your real work without being prompted. 2026 is the year it stops being a demo and starts shipping.
-
Product & workflow
Voice AI vs chatbot: which is right for your team meetings
Chatbots answer when you type. Voice AI answers when you talk. In real meetings, one keeps the room moving and the other stops it.