Agent timeline UI design: re-rank instead of collapse, let events speak with icons and plain titles
_pilliin_ · x · 2026-09-27
@pilliin and zeeg debate how to present event streams (deploys, checks, merges) in an agent chat UI:
- The view is for humans; for an agent you'd render it differently, but the content is the same typed events, just a different projection.
- Events shouldn't be hidden — they are turns. The real issue is inverted hierarchy: bot rows read louder than the human's own voice, so re-rank rather than collapse.
- Since the timeline is chronological narration, you can't lead with the merged card — it hasn't happened yet at the top. What you can change is how inbound events are presented.
- The author's solution: one rail, with an icon per event type (deploy, checks, merge) and a plain title like "checks passed"; bot names go secondary. Today's "received a comment from github-actions[bot]" still tells you nothing about what happened.
- He also tried grouping events into a "3 events" node (trade-offs discussed in the thread).
Related event: Sentry CEO Shows Off Agent Transcript View Design(5 posts)→
More from coding & agent
- Quail project on building with agents: stand on battle-tested community work — charles_irl · 2026-09-27
- Using formal methods to verify agent-generated code instead of reviewing slop — sh_reya · 2026-09-27
- Apodex Open-Sources FrontierAgent Runtime Alongside 35B Mini Model — ChrisGPT · 2026-09-27
- Mid-run prompts add reproducibility and leakage-risk columns to Apodex research table — ChrisGPT · 2026-09-27
- Apodex agent turns 2024-2026 continual learning research into timeline and evidence table — ChrisGPT · 2026-09-27
- Opus 5.5 paired with Astra subagents is the new meta, says Yacine — yacineMTB · 2026-09-27