Ask Your Fleet a Question in Plain Words and Get a Checkable Answer
The number you need is almost never the one a dashboard was built to show. Conversational Insights lets you type the question the way you would ask a colleague — which vehicles ran below 4 km/l on the Trichy lane last week, what TN-45 cost you in June — and answers it from the trip, fuel, cost, maintenance and utilisation data KO Fleetz already holds. It turns the question into a query, runs it, and shows the figures and filters underneath so you can see exactly where the answer came from.
The answer exists in the data. Getting it out takes a day.
The question is simple and it is Tuesday. Which vehicles on the Trichy lane dropped below four kilometres to the litre last week? Everyone agrees the answer is in the system somewhere. It is also nowhere you can reach it: the efficiency report groups by depot not by lane, the fuel export is per vehicle not per corridor, and joining the two means someone with the right access, a spreadsheet and the rest of the afternoon.
So the real questions — the specific, one-off, this-week questions that actually run a fleet — mostly go unasked. Not because the data is missing, but because the cost of extracting it is higher than the value of knowing, and by the time an analyst could turn it around the load has already been dispatched to the same underperforming trucks.
The fallback is to ask the one person who knows where everything lives, and that person becomes a bottleneck and a single point of failure. When they are on leave the ad-hoc questions stop entirely, and the fleet goes back to running on the two standing reports that happen to exist, which answer last quarter's questions rather than this morning's.
Turn the question into a query, then show the working
You type the question in plain English. Conversational Insights reads it, works out which vehicles, which lane, which date range and which measure you mean, and builds a query against the trip, fuel, cost, maintenance and utilisation records the platform already stores. What comes back is the answer to the question you actually asked — the seven vehicles under four kilometres per litre on that lane last week, named and listed — not the nearest pre-built chart.
Crucially, it shows its working. Under every answer are the figures it was computed from and the exact filters it applied: the date window it read 'last week' as, the lane it matched, the vehicles it included and the ones it excluded. You can open any number down to the trips and fuel events behind it, and you can adjust a filter and re-run without retyping the whole question. The answer is a starting point you can check and refine, not a verdict you have to trust.
It does not invent numbers. Every figure it reports is drawn from records the platform holds; where the data to answer a question does not exist, it says so rather than producing a plausible-looking total. And it answers strictly within what you are already allowed to see — it obeys the same role-based access as the rest of KO Fleetz, so a depot supervisor asking about the whole region gets only the depots that are theirs to see.
Capabilities
What KO Fleetz conversational insights gives your team
Plain-language questions
Type the question the way you would say it — no query builder, no report menu, no remembering which export holds cost per lane. The sentence is the interface.
Intent and entity resolution
It works out that 'TN-45' is a registration, that 'the Trichy lane' is a corridor and that 'thirsty' means high consumption, mapping your words onto the fleet's actual vehicles, routes and measures.
Queries over your own records
Each question becomes a filtered read across the trip, fuel, cost, maintenance and utilisation data already in the platform, so the answer is your operation's real figures rather than a canned example.
Shown figures and filters
Every answer carries the numbers it was built from and the exact filters applied — the date window, the lane match, the vehicles in and out — so you can confirm it asked what you meant.
Time expressions understood
'Last week', 'in June', 'since the monsoon started' and 'the last ten trips' all resolve to concrete date ranges, and the range it settled on is printed with the answer so an ambiguous phrase is never silently guessed.
Trend answers over time
Ask how a vehicle's fuel economy has moved across the quarter and it returns the series, not a single figure, so a question about direction gets a shape rather than a snapshot.
Comparison and ranking answers
Questions that pit vehicles, drivers, lanes or months against each other come back as a ranked breakdown — the ten costliest vehicles in June, worst first — with the basis of the ranking stated.
Save a question as a watch
A question you ask often can be pinned or set to re-run and alert you — tell me each Monday which vehicles fell below their target on this lane — turning a one-off query into a standing check without building a report.
How it works
How KO Fleetz does it
Step 1: Ask in plain words
Type the question as you would put it to a colleague. There is no syntax to learn and no need to know which module the figure lives in — the whole point is that the sentence is enough.
Step 2: It reads and scopes the question
Conversational Insights identifies the vehicles, lane, period and measure you mean, resolves them against your fleet's real records, and quietly clamps the whole thing to the data your role is permitted to see.
Step 3: It answers and shows the working
The answer comes back with the figures it was computed from and the filters it applied laid out beneath it, each number opening down to the trips, fuel events or work orders it came from.
Step 4: Check, adjust and re-run
If it read a filter differently from how you meant it, change the date range or the lane and re-run in place. Once the question is right, ask a follow-up or save it as a recurring watch.
Outcomes
What changes
- The question is the query — no report to build
- Plain English
- Figures and filters printed under every answer
- Shown working
- No invented numbers, answered from real records
- Your data only
- Answers respect the roles you already have
- Same access
Frequently asked questions
A dashboard is a pre-built view. Someone decided in advance which measures matter and how to group them, and it answers those questions well every day. Conversational Insights is for the question nobody built a tile for — the specific, one-off, this-lane-this-week question. The dashboard shows you the eleven vehicles outside tolerance on the KPIs it tracks; Conversational Insights answers what a vehicle cost you in June, or which trucks dropped below four kilometres per litre on the Trichy lane last week, because you asked. They are complementary: the dashboard is your standing watch, this is your ad-hoc enquiry.
No, and this is the line the feature is built around. Every figure it returns is read from records the platform actually holds, and it shows those figures and the filters it used beneath each answer so you can check them. When the data needed to answer a question does not exist — a measure you do not capture, a vehicle with no fuel sensor — it tells you the answer is not available rather than producing a confident-looking total. It reports from your data; it does not fill gaps with plausible invention.
From the same operational records the rest of KO Fleetz runs on: your trips, fuel events, cost entries, maintenance work orders and utilisation history. The question is translated into a filtered query over those records, run, and returned with the working shown. There is no separate, second copy of your numbers and nothing pre-computed behind the scenes — which is why an answer here never needs reconciling against the module it came from, because it came from that module.
No. Conversational Insights obeys the same role-based access as everything else on the platform. A depot supervisor who asks about the whole region gets only the depots that are theirs; a driver querying costs sees only what their role permits. The question is scoped to the asker's permissions before it runs, so plain-language access does not become a back door around the access controls that govern the reports and dashboards.
It is only as good as the records underneath it, and it is honest about that rather than hiding it. If your maintenance is not on the platform, it cannot answer maintenance questions and it says so. What it will not do is quietly answer from a thin slice of the fleet and present it as the whole — where coverage is partial, the shown figures and filters make that visible, so you can see the answer rests on forty vehicles and not four hundred before you act on it.
Yes — the most likely error is misreading the question, not miscomputing the data. It might read 'last week' as the calendar week when you meant the last seven days, or match a lane more broadly than you intended. That is precisely why it prints the date range, the filters and the figures it used with every answer: you can see how it interpreted you and, if it read a filter wrong, adjust it and re-run in place. The working is shown not as decoration but so a misread question is caught by you rather than trusted blindly.
Ask your fleet the question you could never get answered
Bring one question that usually takes an analyst a day — a lane, a vehicle, a month — and KO Fleetz will answer it live and show you the figures underneath.