Works on live portfolio data
Does it read current data from every site and brand, or only an export from last month?
Platform comparison
Not a ranking. AI tools for solar portfolios fall into four kinds, each good at different jobs. Here is what each kind does, and five criteria to judge any of them.
An AI coworker that works on live portfolio data every day, such as Ray (AMMP AI). Ray checks every site each morning, groups repeat alarms and opens tickets. People decide; Ray doesn't control hardware.
Fits best: Teams that want daily checks and prepared work, with the decision kept by people.
Meet RayAI functions built into a monitoring platform, working on that platform's own data.
Fits best: Teams whose whole portfolio already runs on one platform.
General AI assistants such as Claude and ChatGPT, connected to portfolio data through an MCP server.
Fits best: Analysts who want to ask their own questions in the assistant they already use.
About the MCP serverSeparate tools for analysis, modelling or forecasting, fed with exported data.
Fits best: Specialist analysis that sits outside daily operations.
Does it read current data from every site and brand, or only an export from last month?
Does every answer show its sources, so you can check the working before you act?
Do people approve tickets, reports and messages before anything goes out?
Does it only see what the person asking is allowed to see?
Does it work where your team already works: Microsoft Teams, Slack or the monitoring platform?
Ray sits on AMMP's vendor-agnostic data layer, with telemetry from 10,000+ sites in 30+ countries.
Every answer shows its sources, so you can check the working before you act on it.
Ray investigates, prioritises and prepares the work. Your team stays in charge of the decisions.
Ray only sees what the person asking is allowed to see.
Ray works in Microsoft Teams, Slack and AMMP OS. Claude and ChatGPT can connect through the AMMP MCP server.
Meet RayMCP server and Data APICan Claude or ChatGPT analyse solar monitoring data?
It depends on the job. AI agents such as Ray handle daily checks and prepared work, general assistants such as Claude and ChatGPT answer ad-hoc questions through an MCP server, and analytics tools cover specialist modelling. Check each against live data, evidence, human approval, permissions and where it runs.
Yes, through an MCP server. The AMMP MCP server gives assistants such as Claude and ChatGPT permissioned access to normalised portfolio data.
No. Ray doesn't control hardware. He checks sites, groups repeat alarms and opens tickets, and people decide what happens next.