Automate the part that leaks revenue and eats time, the writing up and the pricing, and keep a person on the part that should not be automated, the approval. Mentat is field service revenue recovery software built on exactly that split. A technician's field observation becomes a drafted, priced estimate on its own. It reaches a customer only after someone on your team reviews and approves it.

What gets automated, and what stays human?

Draw a line through the work. Drudgery on one side, judgment on the other.

The drudgery is everything between the observation and the drafted estimate. Someone has to remember what the technician saw, reconstruct it, price it, and type it up. That is where the revenue leaks, because most of the time no one gets to it. So that is the side you automate. Mentat captures the observation from a short voice note and structures it into priced recommendations, with no reconstruction by hand. Voice Notes to Estimates walks that capture step by step.

The judgment is the send. A drafted estimate carries a price and your company's name on it. Approving it is a decision, not paperwork. So a person stays there. An operator on your team reviews each drafted estimate, adjusts what needs it, and approves what is worth sending.

Automate the write-up. Keep the human on the price.

Why not automate the whole thing?

Because what you would be automating is a customer-facing quote, and that is the one place you do not want a machine acting alone.

This is the quiet fear under any "AI estimates" pitch, that software starts firing prices at your customers with nobody looking. Mentat does not work that way, by design. Nothing it drafts reaches a customer on its own. The approval gate is not a missing feature it will grow out of later. It is the shape of the product.

So you get speed on the part that was slow and control on the part that matters. The drafting is automatic. The deciding is yours.

What do you have to change to run this?

Close to nothing, and no new tools.

The technician records a voice note, near what they already do. The estimate drafts into the same field service software your office already runs and gets sent like any other. There is no new system to log into and nothing to migrate. Mentat is the layer on top, never the system of record.

Today the loop runs against Jobber, a real voice note becoming a real draft estimate in a real account. That is the honest proof point. Mentat is built to work alongside the systems these businesses run, ServiceTitan included, though a live ServiceTitan integration is the direction and not something that exists yet.

Every estimate that gets approved and sold is tracked, and the credit goes to the technician who spotted the work. That is the recovery in field service revenue recovery.

See also

Further reading

FAQ

Does the customer ever see an estimate Mentat wrote without a person checking it? No. Nothing Mentat drafts reaches a customer on its own. Every drafted estimate stops at an operator on your team who reviews it, adjusts it, and approves it. The approval gate is the shape of the product, not a setting you can switch off.

What happens if the pricing on a drafted estimate is wrong? You fix it at review, the same way you would fix any draft. The operator sees the priced recommendation before anyone else does and can change the price, change the scope, or drop the item entirely. The draft is a starting point for a person, not a finished quote.

Do we have to move off our current field service software? No. Mentat rides on top of the system you already run and is never the system of record. Approved estimates write back into that system as drafts. There is nothing to migrate and no second place to log in.

Which field service systems does this run against today? Jobber is live today, where a real voice note becomes a real draft estimate in a real account. Mentat is built to work alongside the other systems these businesses run, ServiceTitan included, but a live ServiceTitan integration is the direction and not something that exists yet.