Earkeep for UX designers

The job, as designers describe it, is mostly non-design work wrapped around the craft: "I am a designer, a researcher, a reporter, a strategist, a presenter, and a slave to meetings that give me 3/8 hours to design." A senior IC on Blind reported spending 65 to 75 percent of the week in meetings, and the thread's consensus was blunt: a designer "needs time to think and design." The more senior you get, the more meetings eat that time, not fewer.

None of that meeting load goes away because a tool exists. What Earkeep changes is what happens to everything said in those meetings once they're over: it stops being something you have to remember, reconstruct, or chase down later, so the hours you do get back go to design, not to detective work.

Feedback arrives in volume, from everywhere, and contradicts itself. After a PM pulled more stakeholders into reviews, one designer described the result: "now every design decision requires 10 opinions and nothing ships. Every person has feedback, most of it contradicts and we end up with mediocre compromises nobody likes." When decision six contradicts decision two, reconstructing who actually said what, and in which meeting, falls on the designer, from memory, days later.

Critique is high-stakes, and it's hard to capture in the moment. One designer described a director tearing into a flow after weeks of work: "when the director started criticizing the user flow i just shut down and went blank… i walked away feeling like they think i'm a mediocre designer." When you freeze in a review, you also stop taking notes, and the exact feedback you most need to act on is the feedback you're least likely to have written down.

Context is fragmented and decisions get relitigated. Designers describe being unable to focus "for 10 minutes" because every meeting spawns new edge cases and dependencies, and stakeholders arriving "attached to poor solutions" that have to be argued down again and again, sometimes with no record that the argument already happened once.

Earkeep runs continuously in the background, capturing mic and system audio and transcribing it locally the whole time you're at your desk. You never hit record before a crit or a stakeholder sync, so there's nothing to forget to start.

• Retroactive spans. After the meeting, whether it went well or you froze halfway through, you select the span on the timeline and it's kept as a real meeting, transcript included. You don't have to know a review is worth capturing until after it's over.
• Calendar auto-spans. Connect Google Calendar, Apple Calendar, Outlook, or any ICS calendar URL, and design reviews, crits, and stakeholder syncs on your calendar are captured and closed off automatically. No selecting spans by hand at all, if you'd rather not.
• Every word is greppable afterward. The contradictory feedback, the director's actual critique, the stakeholder's verbal sign-off two sprints ago: all of it exists as a plain-text transcript on your disk, not a memory you're reconstructing under pressure in the next meeting.

The mechanics behind all three are on always-on capture and calendar connection.

Earkeep runs in-app agents (Claude Code, Codex, Cursor Agent, Gemini CLI, Hermes, Kimi CLI, OpenClaw, opencode, Pi, Qwen Code, or any ACP-compatible agent) directly against your local meeting history, and ships a built-in MCP server so external tools can read it too. For a designer whose real workspace is Figma and FigJam, with decisions scattered across Notion or Confluence docs, Jira or Linear tickets, and Slack threads, that means:

• Ask the agent what the PM actually agreed to about a specific flow, and get an answer sourced from the real conversation in this week's review, not a half-remembered Slack thread.
• Have the agent draft the Notion or Confluence decision log entry from the actual critique transcript, instead of writing it from memory after the meeting has faded.
• Point the agent at a stakeholder review to pull out what changed since last time, so you walk into the next session already knowing where the disagreement actually is.
• Because transcripts are plain files, nothing stops you feeding a critique transcript straight into whatever tool sits next to your Figma file, with no MCP integration at all.

This is agent access on your terms: Earkeep does not judge your designs or generate feedback about your work. It captures what people actually said, in meetings that already happened, and hands that record to whichever tool you choose to point at it.

• Figma and FigJam for the work itself, with comments as an informal feedback trail. Earkeep doesn't replace this, it captures what happens around it.
• Notion, Confluence, Google Docs for meeting notes and decision logs, usually written from memory after the fact.
• Jira and Linear tickets carrying partial context, missing the discussion that led to the decision.
• Slack threads, where designers say half the real feedback actually lands.
• AI meeting bots via their PM, most often Otter, Fireflies, or Copilot in Teams: designers report exposure to these through someone else's account, rarely their own, and rarely covering meetings the PM wasn't in. See Earkeep vs Otter and Earkeep vs Fireflies.

€39, once, for one device. No subscription, no per-seat pricing, no tiers. Try the full app free for 14 days, no email, no account required to start. If you buy and it doesn't work out, email [email protected] within 14 days for a full refund. See full pricing details →