FiBOT connects the systems your company already uses — the ERP, the spreadsheet, the GPS — and answers over WhatsApp what someone would otherwise have to go dig up. It also flags what nobody gets around to checking.
It's not a dashboard. A dashboard waits for you to log in. Six months in, nobody does — we know, because that happened to us before we built this.
Example conversation · illustrative numbers and names
Almost all business software dies the same way: bought with enthusiasm, half-loaded with data, and six months later only one person still opens it. The data exists. Nobody sees it.
We flip that around: the information goes looking for the person, through the one channel they actually keep open — nothing to install, nothing to learn.
That comes with a cost we'd rather state upfront: not everything should be automated. If what people ask changes every day and never repeats, FiBOT won't help you — and we'll say so in the first conversation.
What does work is the boring stuff: the question someone asks fourteen times a week, and the document that expires while nobody's watching.
Where everything is, how much fuel it used, which permits are expiring, and which unit to send. It connects to the systems you already have — no switching providers, no migrating anything.
In productionThere's a computer at the counter, and it works fine — until the power goes out, or there's a line and nobody can step away from the register. FiBOT rides in your pocket: prices, coverage, equivalents, the full record.
In productionEvery unit gets its own QR code, traceable from intake to delivery. Nothing gets edited or deleted — status builds itself from each scan. Works with no signal too — it queues up and syncs the moment there's a connection.
In productionEverything that's about to expire, in one place, alerting whoever needs to act on it. Permits, insurance, contracts, inspections. Starts from a spreadsheet you already have.
NewIf there's a piece of data living in a spreadsheet that someone asks about every day, that's the candidate. We start with one case, not everything at once.
Custom-builtBefore — “Where's unit 42?”
Open the system → Log in → Search for the unit → Apply filter → Check the position → Copy the data → Send it over WhatsApp
After — a seven-second voice note.
"Where's unit 42?"
9:12 ✓✓No single source tells the whole truth: the system doesn't always know where everything is, the spreadsheet doesn't say what's missing, and your team knows what never gets written down anywhere. Cross-checking them is the actual work.
The spiral opens: from scattered data to the answer
The system you already have, the usual spreadsheet, and what your team reports. Nothing gets migrated.
Webhooks where the platform offers them, periodic polling where it doesn't. Eleven GPS platforms already mapped, among other sources.
Different formats go in; a single table comes out, in a database that belongs to your company alone, separate from everyone else's.
What gets asked gets answered; what's about to expire flags itself, without anyone having to ask.
The biggest square in the spiral, and the little one it comes out through: one line, on the one channel your team already keeps open.
SubjectRE: RE: RE: do we do this manually?
From: Operations · 3 attachments
Forwarding the updated spreadsheet. Heads up, yesterday's version is outdated — work off this one. When you're done, let the team know so they can see it before the meeting.
Minutes lost per question, how many times a day it repeats, business days per month. You run that math yourselves, with your own numbers — we don't have them, not yet. But we already know how it's done: we've been doing it elsewhere.
8 min × 15 questions × 22 days =
44
hours a month — more than a full workweek, spent on questions that already had an answer somewhere.
Reference numbers, not data from any specific client
That's 528 hours a year. The question that follows isn't whether automating is worth it — it's how much it costs to pay someone to do that, twelve months straight.
The math above is illustrative — the pattern isn't. There are already 90+ functions running like this, in production, on top of 300+ units monitored over WhatsApp around the clock.
You tell us what your team gets asked every day and how they handle it now. That's enough to tell whether this fits. If it doesn't, it ends right there and costs you nothing.
We pick the question that repeats the most and connect it to what you already have — the usual system, the spreadsheet, whatever it is. Within two weeks there's something running with a small group of real people.
The second case gets picked by the people using the first one, not by a planning meeting. What nobody asks for doesn't get built — and that's exactly where most projects blow their budget.
You write to it like you would to a person. No exact commands, no menu of options: you ask the way you'd ask a coworker, and the system figures out what's needed.
By API, not scraping. It connects to the systems you already use —ERP, spreadsheets, your own databases— through the API or a direct read. No shared credentials, no touching the original source.
It reads, never writes. Read-only access to whatever gets authorized. It doesn't modify an existing system unless explicitly asked to, and that comes with its own separate permission.
Role-based access, everything logged. Credentials and tokens never pass through the chat — they stay out of it, like in any serious integration. Every access gets logged: who, what, when.
Each of these four points goes into more detail the moment it's needed. The more you ask, the better.
No. We connect to whatever you already have — ERP, spreadsheets, GPS, anything. If yours isn't mapped yet, we look into it, but it isn't always possible — and we find that out in the first week, not six months in.
That's the right question to ask, and the reason this lives on WhatsApp instead of an app. Either way, we measure usage from day one: how many questions, from how many different people. If after four weeks only one person is writing in, the problem wasn't the tool, and we'll tell you that.
We don't publish a price list, because the first case defines the scope and the differences are big. What we can promise: the first conversation ends with a firm number, not a range or an "it depends."
In a database that belongs only to your company, separate from any other client's. No query from one ever touches another's data — and that's not a sales promise, they're simply separate projects.
You take your data, exported, and it shuts off. No lock-in, no long contract. We'd rather you stay because you're using it — which is also the only way this actually works out for us.
Tell us what your team gets asked every day. If it's always the same thing, FiBOT can probably answer it. If not, we'll tell you — selling something that won't work is bad business for both of us.