Dots imports six AI handover records
Moriya reports that a Dots assistant imported six Drive-inbox records into a shared handover document, verified they were saved, then marked them imported.
Original prompt
Shared by the source author. See source notes for context.
書く直前に最新版を読み直し、他の記録を消さずに末尾へ追記してください。
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Ryohei Asada Moriya
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Moriya reports that a Dots assistant imported six Drive-inbox records into a shared handover document, verified they were saved, then marked them imported.
- Human involvement
- Moriya inspects the saved document. The first six-record import works, but reliable future handoffs and reduced re-explanation effort remain unproven.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Create a shared handover document and a Drive inbox for pending records.
- Keep completed actions, decisions and unresolved checks separate.
- Have Dots append the six initial pending records on its schedule.
Ryohei Asada Moriya creates a shared Google document so his Dots assistant, Cham, and Claude Cowork can record work in one place. A handover record distinguishes completed actions, decisions, unresolved checks and the next step, with the responsible environment and artifact links.
The actual first scheduled import processes six records from a Google Drive inbox. Cham appends them to the handover document, checks that the saved content exists, and only then changes their status from unimported to imported. The schedule is three in the morning in Japan. This ordering matters because marking a record too early could hide work lost during an interrupted write.
Moriya also verifies the document updates himself. Claude can read through the Drive connection in his tested environment, but needs the in-app browser to append content. The account proves the initial record collection and document updates, not that later cross-agent work reliably resumes or saves time. The sales-material example in the article is explicitly fictional. The useful implementation detail is to reread before appending, preserve other records, verify after saving and keep missing URLs and unfinished checks explicit instead of inventing completeness.
Prompt
Original prompt
Source: Blog · Ryohei Asada Moriya
Frequently asked questions
Why verify before changing status?
A record marked imported before its document write succeeds could disappear from the pending queue without reaching the destination.
What did the first run accomplish?
It imported six pending records, checked the saved document and then marked those source records imported.
Did this prove smooth work handoffs?
No. The author distinguishes reading, writing and successfully continuing work, and says future handoff effectiveness still needs checking.
More details
Remaining steps
- Read the saved document before marking source records imported.
- Open the destination and verify the appended content yourself.
About this account
The Dots assistant collects pending Drive records, appends them to the shared document, verifies persistence and updates their import status.
Firsthand account by Ryohei Asada Moriya; original publication date verified against Primary Japanese note byline 2026年10月5日 10:39. Outcomes are the author’s reports, not independently reproduced. Original Japanese handover-instruction excerpt; the surrounding instructions also require post-save verification. The article’s sales-material example is fictional.
Credit: Ryohei Asada Moriya
Source published
Added to Agents In Use
Source checked
Tools and topics
Dots · Cham assistant · Google Drive · Google Docs · Claude Cowork · In-app browser
AI handover records · Verified append · Import-state ordering
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.