Research case R5
AI Operations · Browser Extension · Product Design
AI Chat Thread Manager
Tracking AI conversations across providers, with checkpoints and next actions stored locally.
Available nowChrome Web Store
AI chat thread manager / research prototype
Know where the work stopped and what needs to happen next.
Fictional threads demonstrate a unified work queue. Filters and checkpoints run locally in this reconstruction.
Workspace
Your work, across conversations.
Find the checkpoint. Review the next action. Continue with context.
In progress
2Waiting
1Needs review
2Done
1Status, checkpoint and next action remain visible across providers. Fictional records only.
The challenge
Complex AI threads become difficult to resume when status, checkpoints, next actions and context health are not explicitly managed.
Long AI conversations are difficult to resume when the last decision, the next task and the state of the context are scattered across providers.
The approach
The browser-extension prototype records thread identity, status, checkpoints and next actions in a local work queue. The focus is continuity and a clear hand-off, rather than copying full conversations into a central service.
System
Select a step to see its role in the work.
01Provider adapters
Read the supported providers through explicit, limited adapters.
02Thread identity
Keep a stable reference to each conversation across sessions.
03Status
Mark whether a thread is active, waiting or ready for review.
04Checkpoint
Save a checkpoint so work can resume without reconstructing the full conversation.
05Next action
Make the next action visible alongside the thread.
06Context health
Flag when the available context may be too weak for a reliable continuation.
07Handoff
Prepare a concise transfer of context and outstanding decisions.
08Unified work queue
Bring threads together in a local work queue without merging their content.
Current position
A working prototype is available as personal product research. The public examples use fictional conversations and do not represent an employer deployment.
Presented as personal research and product-development evidence, not as an employer project.
Next question
Continue testing whether checkpoints and context-health signals help people resume complex work accurately across providers.
Knowledge graph
Related case studies and research.
Follow the methods and information shared by this work and the cases below.