EthosPowerAIReference design · Intelligence
Project Memory for Site Works
- Macro
- Macro MCP server
- Routines
- Team memory
- LiveKit calls
- EU hosting
The problem
The record of a battery-room retrofit lives in client emails, site calls, the crew channel, the drawing register and the RAMS. Every Friday someone writes the client's status report from memory, and the hold point nobody wrote down goes missing from it.
What I would build
Each site gets a Macro project, the client a CRM company, the crew a channel; the engineer files client threads into the project and Macro records and transcribes site calls. A Friday routine reads the week's emails, calls and documents from source through Macro's MCP tools, searches a firm-owned list of literal site terms, and drafts the status report as a Macro document with an @-mention on every line. Where the nightly memory disagrees with the latest source, the draft flags it. The routine opens a send-email review card; the engineer corrects and approves, and nothing reaches the client before that.
The logic flow
Click any node for what it does, what it hands on, and where a person is required.
What it returns
This is a reference design on Macro's open-source workspace, not a system I run, so there is no measured saving to quote. The tool names are Macro's own and the run shown is illustrative. The number I would watch is how many lines in each approved report the engineer had to correct, and how many of those the memory had wrong.
Why it fits
A utility's asset manager can now ask her own assistant to summarise a week of emails from three contractors and get a tidy page back before any of their Friday reports arrive. The weekly status report used to be something a contractor sold as diligence, and now the client can assemble a version of it herself. A contractor that answers by having a chatbot write that report faster keeps the habit that made it unreliable in the first place; somebody rebuilds a week of hold points, outage bookings and variations from memory, and now a model does the rebuilding with more confidence and the same gaps.
What the client's assistant cannot see is the site record itself: the crew channel, the call where the electrician said the isolation was not yet signed off, the drawing register, the revision of the method statement actually on site. So I would put each job in one workspace, with a project folder per site, the client as a company, the crew in a channel, the client threads filed beside them and the site calls recorded and transcribed. Every Friday a routine reads the week from the source items rather than from the nightly memory. It searches for the literal words a site uses, 'hold point', 'isolation', 'outage', 'variation', 'defect', and drafts the report with a link to every item it cites. Where memory and the latest source disagree, the draft says so. The engineer reads it, corrects it and approves the send; calls about price stay out of the crew's shared memory.
One five-year strategy it could serve, offered as a sketch rather than a plan: sell the record behind the report rather than the report. A contractor whose every status line points back to the email, the call or the drawing it came from can hand a client the trail behind a claim instead of a summary of it, and after a few dozen jobs that trail answers the questions clients ask at practical completion and in a dispute, where a tidy page from memory answers none of them. The client's assistant can write a report. It cannot write the evidence.
See it run
Want one of these?
Every part of it is open source and self-hostable. What I bring is the build, the checks around the project's gaps, and the judgement about where your gates should sit.