Connect your AI assistant straight to your Payaca account and ask it to do real work: pull stuck jobs, bulk-update projects, draft follow-ups, build reports. Here's why we built it and the practical things install teams use it for.
Matt Franklin
We shipped an MCP server for Payaca. It lets you connect your own AI assistant to your account and ask it to do real work: find the jobs stuck waiting on a grid connection, tag every project from last month's import, draft a follow-up note, pull a conversion report. You ask in plain English. It does it in your account.
If you have used Agent Dave, this is the same engine. Dave runs the operational work that happens on a schedule or a trigger. The MCP server hands that same ability to whatever AI client you already use, so you can do things on demand, in a conversation, whenever a question comes up.
See it on your own data
The fastest way to understand this is to watch it run against a real account. Book a 20-minute demo and we'll connect it live and answer the questions you actually ask your ops team every week.
Every growing install business runs on a steady stream of small questions. Which jobs are waiting on the DNO. Who hasn't been invoiced. What converted last quarter and what didn't. Which customers are due a service reminder.
Most of those questions have an answer sitting in your system. The problem is getting it out. Today that usually means someone exports a CSV, opens a spreadsheet, and spends twenty minutes turning data into the thing you actually asked for. Or it becomes a request to your software provider to build a report, and you wait.
We did not want to build a report for every question an installer could ever ask. That road never ends, and most of the questions are one-offs anyway. So we did the other thing: we made the account itself something you can talk to.
MCP is an open standard for connecting AI assistants to real systems. We built Payaca's server on it deliberately. It means your account is not a walled garden. You bring the AI client you already use, point it at Payaca, and it can work with your data directly. No export, no waiting on us.
The point is not the technology. It is the everyday operational jobs it takes off your plate. Here are the ones install teams reach for most.
Instead of building a view and filtering it by hand, you ask:
You get the list back in seconds, with the project links, ready to act on.
The jobs that used to mean a spreadsheet round-trip become a single instruction:
It makes the change in your account. You review it. No CSV leaves the building.
Ask the question once, get the number. Ask it again next month without rebuilding anything.
Data hygiene is the work everyone knows they should do and nobody schedules. This makes it a five-minute conversation.
Because it can read the project, it can write from it:
It is working from what is actually in the record, not a blank page.
This is a powerful tool, so it is worth being clear about what it is and is not.
It acts as you. It uses your login and your permissions. It cannot see or do anything in Payaca that you could not see or do yourself.
It is only as good as your data. If your pipeline stages are inconsistent or half your jobs are missing a value, the answers will reflect that. The flip side is that using it tends to surface exactly those gaps, which is no bad thing.
Review before you commit a bulk change. When you ask it to update fifty projects, look at what it is about to do before you say yes. It is fast, and fast is worth a glance.
It is a tool for your team, not a replacement for knowing your business. It gets the data work out of the way so the people who understand the work can spend their time on it.
For years, getting an answer out of your business software meant either knowing where to click or asking someone who did. Software vendors decided which questions were worth a button and which were not.
That model is ending. When your account is something you can talk to, the set of questions you can ask is no longer fixed by what we thought to build. You ask the question you have, today, and you get the answer.
Agent Dave was the first step: AI doing the operational work inside Payaca on its own. The MCP server is the second: that same ability, in your hands, on demand. We are building Payaca to be AI-native from the data up, not an old system with a chatbot bolted on the side.
It's live now
If you're already on Payaca, you can add the MCP server as a connector in your AI assistant, authorise it with your login, and start using it today. If you're not, book a demo and we'll show you what it does with the questions your team asks every week.
See how Payaca helps clean tech installers save time and grow their business.
Book a demo
What an MCS-certified solar installer has to produce by commissioning sign-off - MIS 3002 evidence, IEC 62446 tests, the EIC, DNO notification, MID registration within 10 working days, and a handover pack that unlocks the customer's SEG application.
G99 rejections delay solar installations by weeks. These are the most common causes of failed applications - and what to fix before you submit.
Mid-sized UK installers spend two engineers a year maintaining a bespoke CRM that handles maybe 60% of the workflow. The other 40% is on spreadsheets. Here's what's actually being paid for, and why the build never catches up to MCS, DNO and BUS V5.