The meeting ends. Thirty seconds later, the summary lands in Slack - clean headers, action items, decisions bolded. Everyone reacts with the checkmark emoji. It feels like progress.

Nobody asks the obvious question: who’s turning this into the PRD, the tickets, the stakeholder update?

You are. This afternoon. Or tonight.

The receipts just came in

This week, Lenny Rachitsky and Noam Segal published their second annual tech worker sentiment survey - 5,332 working tech professionals, nearly half of them PMs. The headline numbers should stop you cold:

Burnout jumped from 44.7% to 55.7% in a single year. Career optimism fell below half. And this happened during the exact stretch when AI meeting tools went from novelty to default - when 82% of respondents say AI is measurably making them more productive.

More productive and more burned out. That’s not a contradiction. It’s a mechanism. And the respondents can describe it better than any analyst:

Researchers call it “smiling exhaustion” — output and burnout climbing together.

“I can do more, faster, but not better.”

“AI helps with the toil, but then it’s also an enabler to do even more toil.”

“We just set a new denominator for the job. And it moves higher and higher every month.”

Here’s the part almost everyone gets wrong. The dominant narrative says tech workers are afraid of AI taking their jobs. The survey says only 22% worry about that - near the bottom of the list. What 51% worry about, the number-one fear in the entire dataset, is being expected to do more for the same pay. Another 46% fear getting trapped in an unsustainable pace.

Nikhyl Singhal calls this “smiling exhaustion.” The survey calls it the squeeze. Either way: the fear isn’t replacement. It’s the treadmill speeding up while the readout says you’re winning.

The two-hour tax

So what do AI meeting summaries actually have to do with this? Let’s do the math on a real one.

Your summary says: “Aligned on Q3 scope - Sarah to update the PRD.” That’s 45 minutes of edits. “Ticket out the auth changes.” Forty minutes in Linear, because the summary captured the decision but not the acceptance criteria. “Circulate the decision to stakeholders.” Thirty minutes of carefully worded Slack diplomacy. “Follow up with design on the empty states.” Another thread, another context switch.

One meeting. One tidy summary. Roughly two hours of invisible follow-up.

Now run that across six meetings a day and you’ve found the missing hours the survey is describing. A summary is not a deliverable. A summary is a list of work you haven’t done yet - a work order, formatted to look like a receipt.

Why the gains disappeared

This is the survey’s sharpest finding, hiding in plain sight: productivity went up and burnout went up at the same time. How?

Because summary tools compressed the cheapest part of the work - capture - and left you the expensive part: completion. Nobody was drowning in the note-taking. They were drowning in the converting: notes into specs, decisions into tickets, discussions into updates.

And the capture time you “saved”? It didn’t come back to you. It got repriced. The bar moved. Every gain became the new baseline — and, as one respondent put it, it moves higher every month. The tools got credit for the savings. You absorbed the difference.

That’s the squeeze, itemized. Not a feeling. A workflow.

What if the meeting produced the work?

There’s a different way to close that gap, and it’s not a better summary.

It’s finishing the work while the conversation is still happening. Earmark listens to your meetings in real time and turns what’s said into the actual artifacts - the PRD, the Linear and Jira tickets with acceptance criteria, the stakeholder update - before the call ends. Not notes about the decisions. The output of them.

That changes what the productivity gain is. When the deliverable ships during the meeting, the saved time can’t be quietly repriced into a higher baseline - it shows up as capacity you can see and defend. The gain becomes relief instead of a new denominator.

The survey’s advice to leaders says it plainly: “The fastest way to end up with resentment on your team is to pocket the productivity and turn saved time into more work for them.” The advice to everyone else is simpler still - stop mistaking the summary for the work.

There’s a moment in every knowledge tool where you’re asked to make a decision you’re not ready to make.

Notion asks you to design a database schema before you’ve captured a single note. Your project tracker asks you to define phases and milestones before you’ve had the first real conversation. Your CRM asks you to categorize a relationship before you know what it is. These tools all share an assumption so deep it’s invisible: that structure comes first, and information gets poured into it.

We built Earmark on the opposite assumption. We call it the decide-late principle: capture everything now, structure nothing until the moment you actually need it.

The cost of deciding early

Every early decision is a bet that the future looks like the present. The database schema you design in week one encodes what you thought the project was in week one. The folder hierarchy reflects the org chart from before the reorg. The “project status doc” describes a project that has since become a different project.

And here’s the thing about early structure: it doesn’t just go stale, it actively resists correction. Once a schema exists, information gets contorted to fit it. Once a status doc exists, someone has to maintain it - which means reconciling every new development against an old summary, forever. The document becomes a second job. Most teams quietly stop doing that job, and the doc becomes the most confidently wrong artifact in the workspace.

The traditional answer was to try harder: better templates, stricter update rituals, a designated doc owner. But the problem was never discipline. The problem is that the structure was decided before the information existed.

What deciding late looks like

Take something concrete: meeting notes and project status.

The tempting design - the Notion-shaped design - is a canonical status doc that each meeting updates. Every meeting ends with someone editing the living document, merging new reality into old summary.

The decide-late design treats each meeting as an immutable snapshot in the project’s timeline. Nobody edits it. Nobody reconciles it. It’s simply what was true, and said, at that moment.

Then “current status” stops being a file and becomes a query. When you ask where the project stands, the answer is synthesized on demand from the sequence of snapshots - always fresh, never manually maintained, never stale by construction. The structure (a status summary) is created at the moment of need, from the full record, and then it can be thrown away, because you can always ask again.

This inversion - records are immutable, views are computed - is old news in software engineering. Event sourcing, append-only logs, materialized views: infrastructure people have known for decades that deriving state from an immutable history beats mutating state in place. What’s new is that the “view” no longer has to be a rigid aggregation someone programmed in advance. An agent can synthesize the view you need, phrased the way you need it, at the moment you ask. AI is what finally makes decide-late practical for messy human information, not just database rows.

The industry is already living this way

This isn’t just a data-modeling preference. It’s how work itself is shifting.

A product lead at OpenAI recently described how planning has changed inside the company: nothing gets planned more than a month or so out, because the ground moves too fast for longer horizons to mean anything. That sounds like chaos until you notice it’s the same principle. A twelve-month roadmap is a status doc for the future - an early decision that must be endlessly reconciled against a reality that refuses to cooperate. Teams that plan late aren’t failing to plan; they’re refusing to pay the reconciliation tax.

Honestly, the year-plan was absurd before AI, too. We just tolerated it because the tooling gave us no alternative - deciding early was the only way to coordinate. When your tools can synthesize a current picture on demand, the long-range plan loses its main job. You keep the direction; you drop the pretense that you know the route.

Deciding late is not deciding never

The objection writes itself: isn’t this just an excuse for having no structure and no plan?

No - and the distinction matters. Decide-late doesn’t eliminate decisions; it moves them to the point of maximum information. You still make the schema, the summary, the plan. You just make it when you need it, informed by everything that’s happened, instead of guessing up front and defending the guess. Deciding early feels responsible, but it’s mostly a way of feeling in control. Deciding late is what taking the information seriously actually looks like.

The prerequisite is trust in your capture. You can only defer structure if you’re confident the raw record is complete and durable. That’s the trade Earmark makes: we’re strict about capture - every meeting, immutable, in sequence - precisely so we can be relaxed about everything downstream. Rigid record, fluid structure.

Every document you write about work - the status update, the CRM entry, the project recap - is stale the moment you hit save. You know this. You’ve felt it. You spend Friday afternoon assembling a summary of the week, and by Monday’s standup half of it is wrong. So you update it. And then you update it again. Somewhere along the way, maintaining the record of the work became a second job on top of the work itself.

We think that second job is about to disappear.

Work about work

Most of the tools we use every day - CRMs, project trackers, status docs, weekly ceremonies — exist for one reason: coordination. How do we divide and conquer? How do we avoid colliding with each other? How does everyone stay accountable? Those are real problems, and for decades the answer was the same: make people write things down, then make other people read them.

But notice what all of those artifacts have in common. They’re snapshots. A CRM is obsessed with the outcome of activity - the deal stage, the pipeline number - while the actual machinery of the work lives somewhere else entirely: in the conversations. The meeting where the customer told you what they actually needed. The call where the blocker surfaced. The hallway conversation that changed the plan.

The snapshot was never the source of truth. The conversations were. We just didn’t have a way to ask them anything.

Ask, don’t maintain

Now we do. When every conversation your team has is captured and instantly retrievable, the whole equation flips. You don’t prepare a status update - you ask, “where are we with this customer?” and get a just-in-time answer built from everything that’s actually been said. You don’t maintain a document - you retrieve. A flat system with instant, intelligent recall beats any folder hierarchy you could ever hope to build.

This is the same reason we’ve stopped asking whether we even need a CRM internally. Not because tracking deals doesn’t matter, but because the answer to “remind me where we left off” shouldn’t live in a form somebody filled out from memory three weeks ago. It should come from the conversation itself.

The new primitive: live views, not documents

Here’s where it gets interesting. If retrieval replaces the document, what replaces the shared document - the thing a team looks at together?

Our answer: you don’t pin documents anymore. You pin questions.

Imagine a project view where the things you care about - latest blockers, open risks, where a decision landed - aren’t paragraphs someone wrote, but standing prompts that recompute every time a new meeting flows in. Nobody edits them. Nobody updates them. If the output isn’t what you need, you change the question, not the answer. It’s always live, it’s shared by default with everyone in the project, and if you want more depth you just ask a follow-up.

Don’t subscribe to summaries. Subscribe to answers.

The super-empowered individual

Zoom out and this is a bigger shift than a feature. The working model that’s emerging isn’t a bigger team with more ceremonies - it’s a single person with layers of automation underneath them and full context on demand. Status is ambient. Roll-ups are a prompt. The calibration moments that used to require a meeting now require a question.

Tools like Gong figured this out for one role: hop on a sales call and the system preps you - where you left off, the risks, the action items. We think that’s too narrow. Every role has conversations. Every role deserves that leverage. Why should sales be the only function that gets to show up to every interaction already knowing everything?

The people we talk to feel this pressure acutely. Everyone’s being asked to move faster with smaller teams. Everyone’s spinning up new workflows and quietly worrying whether the quality holds. The winners won’t be the ones who write better status updates. They’ll be the ones who stopped writing them.

Meeting you where you are

One honest caveat: not every organization is ready to burn the artifacts. That’s fine. This isn’t all-or-nothing. If your org runs on documents, generate them - from the real source of truth, in seconds instead of an afternoon - and ship them into whatever tools your team already uses. The foundation is the same either way: every conversation, captured and retrievable. How far you take it is up to you.

What’s coming in Earmark

This is the direction we’re building. Full-history retrieval - ask anything across every meeting you’ve ever had, not just the recent ones - is in internal testing now. Live project views built on pinned prompts are right behind it, along with deeper ways to pipe Earmark’s context into the rest of your workflow through MCP. Recording the meeting was where we started. It’s quickly becoming the least interesting thing we do.

The future of work isn’t a better document. It’s never having to ask someone “what’s the latest?” - because the answer is already there, already current, already yours.

Stay tuned.

Most meeting tools stop at recording, transcribing, and summarizing.

That was useful when the problem was:

“What happened in this meeting?”

But the harder, more valuable questions usually come later:

• What did I agree to last month?

• What did the customer keep pushing back on?

• Where did we discuss this project decision?

• What context do I need before this renewal call?

• How can I improve my 1:1s?

Those answers rarely live inside one meeting. They live across weeks of conversations, decisions, follow-ups, side comments, and shifting context. That is why we are building Earmark’s upcoming Agentic Search experience: a new way to search your meeting memory that behaves less like a keyword box and more like a research assistant.

From Meeting Notes to Organizational Memory

Earmark has always been about more than notes. In recent product discussions, we kept coming back to one idea: the real value is organizational memory. The ability to recover what your team knew, decided, promised, worried about, and planned across time - not just what was captured in a single call summary.

That is the foundation for Agentic Search.

Instead of asking you to remember the exact meeting, title, date, or phrase, Earmark will help retrieve relevant context across your meeting history, read through it, filter it, and synthesize an answer. You should not have to search like a database administrator to get value from your own conversations.

Search That Works More Like a Person

Traditional search is mostly matching. You type a word. It finds the word. But meeting memory is messy. People use different phrases. Decisions happen gradually. Follow-ups get implied. Important details show up across customer calls, internal syncs, project meetings, and prep conversations.

Our upcoming agentic retrieval system is designed to handle that complexity by searching in a more human-like way:

• understanding the question

• looking across meeting data

• filtering what matters

• reading the relevant context

• synthesizing a useful response

That means Earmark can move beyond “find the transcript where someone said X” toward higher-value questions like:

• What changed since the last customer check-in?

• What objections came up during this renewal?

• What did I commit to?

• What are the open threads on this project?

• How can I improve my recurring 1:1s?

This is the difference between searchable notes and useful memory.

Built for the Way Work Actually Happens

Meetings rarely exist in isolation. They belong to projects, accounts, initiatives, renewals, launches, and long-running relationships.That is why Projects are becoming a key organizing concept in Earmark. Projects help group meetings and shared context into recoverable workspaces, making it easier to build institutional knowledge around a customer, team, initiative, or ongoing workstream.

Agentic Search gets much more powerful when it understands that structure. If you are working on a renewal, you should be able to ask what the customer cared about across the last few calls.

If you are preparing for an employee review, you should be able to pull together feedback, wins, blockers, and commitments from prior meetings. If you are documenting a project, you should be able to recover the decisions and context that got you here, instead of piecing it together from memory.

The Next Step: From Recall to Action

Better search is only the beginning. The long-term opportunity is helping people act on meeting context:

• prepping for upcoming conversations

• drafting follow-ups

• surfacing reminders

• managing agendas

• turning past discussions into useful next steps

We have also been improving the underlying meeting data itself. For example, Earmark has been working on clearer “me vs. them” transcript separation, which lays the groundwork for more precise questions like:

• What did I agree to?

• Did I answer their questions?

• What did they ask for?

• What became my responsibility?

That matters because the best meeting assistant should not just know what was said. It should understand whose responsibility it became, what changed, and what should happen next.

Why We’re Building This Now

Work has changed. Teams are having more recorded conversations across more tools, with more context spread across more places. At the same time, expectations are rising. People do not want another dashboard to check.

They want fewer admin tasks, faster deliverable cycles, better follow-through, and less time spent reconstructing what already happened. That is the promise of Agentic Search in Earmark.

Not just:

Search your meetings.

But:

Search your memory.

Ask better questions. Recover context faster. Turn conversations into progress.

Join the Waitlist

We are opening early access to Earmark’s Agentic Search experience soon. If you want to be one of the first teams to try meeting memory that can search, reason, and synthesize across your conversations, sign up @ waitlist@tryearmark.com