The short answer: A fleet AI agent should never spend money or change records without a person's sign-off. In HoneyRuns, repair estimates always go to the fleet manager to approve before work goes ahead, and the fleet chooses which other actions Riggs may take on its own, which need approval, and which it may only draft. Writing a completed service into Fleetio always requires approval. Riggs reads telematics data rather than changing it, and it does not need access to your ERP or accounting system.
What fleets worry about
When fleets consider an AI agent for maintenance, the same concerns come up:
- Will it book work or spend money without asking?
- Will it make mistakes in our maintenance records?
- What systems does it need access to, and will IT approve that?
- If something goes wrong, can we see what it did?
These are the right questions. An agent that saves a coordinator's time but creates surprise invoices is not worth it.
How Riggs handles approvals
Estimates need your approval. When a shop sends an estimate, Riggs sends it to the fleet manager, who approves it before work goes ahead.
You set what Riggs can do alone. Each type of action Riggs takes can be set to one of three levels: draft only (Riggs prepares it, a person sends it), require approval (the default), or run automatically for low-risk actions such as reminders.
Record changes need sign-off. When a repair is done, logging the completed service into Fleetio always requires a person to approve it first.
What Riggs connects to
- Telematics: read-only for Samsara, Geotab, Verizon Connect and Motive. Riggs reads mileage, engine hours, fault codes and vehicle health and does not change settings in those systems. With Azuga, logging a completed service also resets the matching service reminder.
- Fleet records: Fleetio, reading open issues and writing completed service entries after approval.
- Communication: Slack, text message and email, to reach the fleet manager, drivers and shops.
Riggs does not need access to your ERP or accounting system.
Why not build this with Copilot?
General AI assistants are good at drafting and summarizing. Fleet coordination needs more: live telematics data, a record of every open repair, the ability to reach drivers and shops by text, and approval rules tied to each action. Building and maintaining that around a general assistant is a project. Riggs comes with it.
Starting small
A sensible way to start is with a small group of vehicles or one location, every action set to require approval, and settings loosened only after you have seen what Riggs does.
Frequently Asked Questions
Q: Can an AI agent approve fleet repairs on its own? A: It should not approve spending without a person. In HoneyRuns, repair estimates go to the fleet manager for approval before work goes ahead.
Q: What can Riggs do without asking? A: Only what you allow. Each action type can be set to draft only, require approval, or run automatically if it is low risk, such as sending a service reminder.
Q: Does a fleet AI agent need access to our ERP? A: Riggs does not. It reads telematics data, works with Fleetio if you use it, and communicates over Slack, text and email.
Q: Is HoneyRuns' telematics access read-only? A: For Samsara, Geotab, Verizon Connect and Motive, yes. Riggs reads vehicle data such as mileage, engine hours and fault codes. With Azuga, logging a completed service also resets the matching service reminder.
Q: Why use a fleet AI agent instead of Microsoft Copilot? A: Copilot is a general assistant. A fleet agent needs live vehicle data, a record of every open repair, direct contact with drivers and shops, and approval rules per action. Riggs includes those; with Copilot you would have to build them.
Get Started with HoneyRuns
Start with a few vehicles and every action on approval. Get started.
HoneyRuns gets fleet maintenance done. It watches mileage, intervals, and telematics, books the repair with the shops and vendors you already use, stays on the driver and the shop until the vehicle is back, and logs the completed service in the system you already run. Your team approves repair estimates before work goes ahead. Founded by fleet operators who manage 70 vehicles across three fleets today.