Continuum.ai builds workflow software for mid-market finance teams. Their product design group — twelve people across three squads — runs twice-weekly design critiques on Zoom. For years, feedback lived in Figma comment threads, Slack threads, and someone's personal Notion page. By the time engineering picked up a ticket, the rationale behind a layout change was already gone.
The problem with comment archaeology
Lead designer Mara Okonkwo tried the usual fixes: a crit template, a dedicated Slack channel, a Loom after every session. None of it stuck. Designers spoke faster than anyone could type. Stakeholders who missed a crit asked for a recap that took longer to write than the meeting itself. When a PM asked why the settings panel moved from a sidebar to a tab bar, the team spent twenty minutes reconstructing a conversation from three weeks prior.
We weren't bad at crits. We were bad at preserving the reasoning. Figma shows you what changed — not why three senior designers agreed to kill a pattern we'd shipped twice before.
Recording every crit without changing the ritual
In March 2026, Continuum.ai started recording crits with Scriba from the menu bar. The habit was deliberately lightweight: one person hits record at the start, stops at the end. No one takes official notes during the call. After each session, the presenting designer spends ten minutes in Scriba's chat asking for a structured summary — open questions, agreed changes, dissenting opinions, and owners for follow-up.
- Real-time transcription lets absent squad members skim the transcript instead of requesting a recap.
- Meeting memory accumulates context across recurring crits on the same feature area.
- Search across past crits surfaces phrases like "tab bar" or "empty state" in seconds.
- Ghost mode keeps Scriba off screen shares when they walk through live prototypes.
What improved in six weeks
Okonkwo tracked three metrics: time from crit to Jira ticket, number of "why did we decide this?" Slack pings, and async viewership of session recaps. Ticket-writing time dropped from an average of forty minutes to under fifteen — designers paste the Scriba summary into the ticket description and link the meeting. Slack clarification pings on design decisions fell by roughly half. Crit attendance stayed flat, but transcript views from people who did not attend rose sharply.
The first week felt weird — like we were performing for a microphone. By week three it was invisible. Now I can't imagine shipping without a searchable record of what we argued about.
Advice for other design teams
Continuum.ai's playbook is simple: one recorder per crit, a consistent post-meeting prompt for summaries, and a team norm that Jira tickets link back to the source meeting. They did not replace Figma comments — those still track pixel-level feedback. Scriba holds the spoken reasoning that comments never capture. For a team that lives in critique sessions, that turns out to be the part that matters most.