NEMT Software for a Small Operator: A Checklist for Dispatch, Logs and AI

A small non-emergency medical transport operator should look for software that imports broker trips, dispatches them on a live board, captures trip verification in a driver app and exports clean billing. The AI features worth considering are route optimisation and, optionally, a call-answering assistant; most others are labels on automation you can test.

That order matters because the money in NEMT leaks in the back office: a trip that cannot be verified is a trip that may not be paid. A tool that routes cleverly but exports messy billing files has the priorities upside down. This checklist puts dispatch, records and billing first, then separates the AI claims you can verify from the ones you cannot.

Follow me on Instagram@sagnikteaches

The checklist, grouped by job

Score each line 0 (missing), 1 (clumsy) or 2 (works as we need) during a demo or trial. A zero in the first three groups ends the contest for that product.

Connect on LinkedInSagnik Bhattacharya

Dispatch and scheduling

• Brokers' trips arrive on your schedule without retyping. Ask which brokers have a live link (Bambi lists MTM, Modivcare, Verida "and more"; NEMT Platform lists Modivcare, MTM, Call the Car, Kaiser Permanente and VectorCare; NEMT Cloud Dispatch lists Modivcare, MTM, Alivi and others).
• A live board shows every driver, vehicle and trip status.
• Re-assigning a trip after a cancellation takes under a minute.
• Standing orders (dialysis on Monday, Wednesday, Friday) repeat without rebuilding.
• Wheelchair, stretcher and escort needs are fields that the scheduler respects.
• Shared rides can be grouped when your contracts allow it.

Trip verification and logs

• The driver app records pickup and drop-off times and location points.
• Notes, no-shows and wait times are recorded with a reason, not free text alone.
• Records cannot be edited silently; changes are logged with who and when.
• The record is searchable by rider, driver, vehicle and date.
• You can export a trip's record as a document if a payer or broker queries it.

Driver app

• Works with a weak signal. Bambi lists offline sync for dead zones.
• Turn-by-turn navigation inside the app, so drivers do not switch apps.
• Messages from dispatch arrive in the app, not by personal texts.
• Drivers can use it with one hand at a kerb. Test it on a real phone, not on your desk.

Billing and exports

• Invoices built per broker or programme from verified trips.
• Export formats your payers accept. NEMT Platform names EDI and CMS-1500 among its supported formats; confirm what your payers need.
• Payment tracking, so you can see unpaid and short-paid trips.
• A way to list trips that cannot be billed because a field is missing, before you submit.

Business terms

• Price per vehicle, whether it counts only active vehicles, and what is not included.
• Contract length and notice period.
• Onboarding, data migration and support hours.
• Data export if you leave.

What the vendors list, and the price shapes

I read these vendor pages on 4 November 2026. Prices are in dollars, as listed on each vendor's own site.

Subscribe on YouTube@sagnikteaches

Directory sites still repeat older figures for some of these products. One showed Tobi at $50 per vehicle plus $0.10 a trip, while Tobi's own pricing page now starts at $60 per vehicle with no per-trip fee listed. RouteGenie publishes no price at all and sends you to a demo. Budget from the vendor's own page or a written quote, never from a comparison site.

A worked comparison: a five-vehicle operator

Illustrative figures. A small operator runs five vehicles, but only three or four are out on a typical day, with 60 trips a day across two broker contracts and a few private-pay trips.

• Bambi at $69 per active vehicle: Bambi counts the most vehicles out on any single day last month, so your busiest day sets the bill. If all five went out together on even one day, it is 5 at $69, or $345. If four was the peak all month, it is $276.
• NEMT Cloud Dispatch Small Fleet: $149.99 a month flat for up to five vehicles, with no per-trip fee. A sixth van would move it to the Growth plan at $189.98.
• Tobi from $60 per vehicle: 5 at $60 is $300, assuming every vehicle is counted; ask whether idle vehicles are.
• A per-trip charge, if a vendor quotes one: 60 trips a day, 22 days a month is 1,320 trips. At a hypothetical $0.10 a trip that is $132 a month on top of the vehicle charge, which is why a per-trip price changes the comparison for a busy small fleet.

The gap of about $200 a month between the cheapest and dearest options is small next to two numbers: how many hours the dispatcher spends each day on scheduling, and how many trips a month are billed late or not at all. Say the dispatcher spends three hours a day building the next day's schedule by hand. If a tool cuts that to one hour, that is two hours across 22 days: 44 hours a month. Even at a modest $20 an hour that is $880, many times the software price. The comparison that matters is the reduction in your own hours, which you can measure in a trial with tomorrow's real trips.

Testing dispatch on tomorrow's real trips

During a trial, ask the vendor to load a day of your trips, with rider names replaced by codes if you prefer, and run it twice: once with the tool's scheduling, once by hand.

1. Import the day's trips from your broker feed or a spreadsheet.
2. Run the one-click scheduling or routing. Count how many trips it left unassigned and why.
3. Check three trips that you know are awkward: a stretcher trip, a rider with a tight appointment time, a pickup in a building with no easy access.
4. Cancel two trips and add a late one. See how long it takes to reshuffle.
5. Compare total miles and the number of late pickups against your hand-built plan.

Here is how a result might read (illustrative). The tool assigned 56 of 60 trips, left 4 unassigned because no vehicle had a lift free, and planned 5 per cent fewer miles than the hand-built plan. But it put a stretcher trip after a long wheelchair run, which would have made the pickup 25 minutes late. The fix is not to distrust the tool but to set the vehicle capability fields and to keep a dispatcher reviewing the plan. The tutorial on AI in a courier business: quotes, dispatch and proof shows a similar review habit for routes.

Trip verification: the record you are paid on

Your payer or broker contract defines what counts as proof of a trip. Software cannot change that; it can only capture the fields reliably. For each vendor, ask to see, on a real trip in the driver app:

• The pickup and drop-off time stamps, and whether the driver can change them.
• The location points saved at each stage.
• What happens when the signal drops, and when the record syncs.
• Where the rider's signature or confirmation sits, if your contract needs one.
• What the dispatcher sees when a record is incomplete.

A realistic failure: a driver completes a trip in a dead zone, the app does not sync, and the drop-off time is entered later from memory. The broker query a month later cannot be answered. Offline sync (Bambi lists it) and a rule that incomplete trips are flagged before billing close that gap. Test it by switching a phone to aeroplane mode during a trial trip.

Billing exports: where small operators lose money

Ask each vendor to produce an invoice from your test trips and check it line by line. Here is a filled-in example (illustrative) of what a good pre-billing exception list looks like:

Trips ready to bill: 1,214
Trips held (missing data): 38
  - 17: no drop-off time stamp
  - 9: no-show, reason code missing
  - 7: mileage outside broker's allowed range
  - 5: trip not matched to an authorisation
Action: fix the 38 before submission; do not submit held trips.

The list is the real feature. A system that submits everything and shows denials three weeks later makes you chase the same errors on a longer cycle. Ask whether denial and short-payment reasons come back into the software, so you can see which kinds of trips are paid late.

What unpaid trips cost: a sum to run on your own numbers

Illustrative figures. Suppose the operator above bills 1,320 trips a month at an average of $30 a trip, so $39,600 of billings. If 3 per cent of trips are held or denied for missing data and a third of those are never recovered, that is 1 per cent of trips, or about 13 trips and roughly $400 a month lost outright, with the other 26 or so paid weeks late and costing staff time to chase. Put 20 minutes of staff time on each held trip and the 40 held trips cost over 13 hours a month. Now compare that with a software price of $150 to $345 a month. A pre-billing exception list that halves the held trips would pay for the software on staff time alone. Replace every number with your own from last month's remittances; the exercise takes an hour and makes the demo questions sharper.

Three mistakes that show up after switching

• Duplicate trips from a broker feed. The same trip arrives twice because the broker re-sent a change, and two vehicles are sent. Check how the software matches a changed trip to the original, and ask for a duplicate warning on the dispatch board.
• Vehicle capability fields left blank. Every van is loaded as "ambulatory", so a wheelchair rider is assigned a vehicle without a lift. Fill in the capability for each vehicle and rider on day one, and spot-check the first week's assignments.
• Drivers on personal phones with notifications off. A message from dispatch is never seen. Make a one-page rule: app open, notifications on, battery above 20 per cent at the start of a shift.

A short script for the driver app test

Give two drivers the app for three days and ask the same questions each evening. Their answers, not the vendor's demo, show whether it works.

1. Did you ever have to ring dispatch because something in the app was unclear?
2. Did any trip fail to save, or save late?
3. How many taps does it take to log a no-show with a reason?
4. Did the navigation take you the way you would have gone, and could you override it?
5. Would you rather use this or the current method? Why?

If both drivers hesitate on question five, that is the answer, whatever the price. A tool the drivers avoid produces incomplete records, and incomplete records are what the billing checklist exists to prevent.

Sorting the AI claims into three piles

Vendors use "AI" for very different things. Sort each claim by asking what would happen if the feature were wrong.

Route and schedule optimisation

NEMT Platform describes AI route optimisation that proposes routes while allowing manual adjustments. Bambi's visible copy describes its scheduling run without calling it AI. Either way, the test is the one above: run your own trips and compare. The risk is moderate and visible, because a dispatcher reviews the plan.

Call answering

NEMT Platform lists an AI receptionist that answers calls, books trips, handles cancellations and gives ride updates. Test with care: riders calling about a missed pickup are often anxious, and many are elderly or have limited English. Listen to recorded test calls, check that the assistant hands over to a person on request, and check that it cannot promise a pickup time the schedule does not support. The tutorial on what to do when an AI receptionist gets a booking wrong sets out the fixes.

Driver monitoring

NEMT Platform lists AI in-cab coaching, and NEMT Software lists an AI dashcam that monitors driving behaviour and offers in-cab education. These raise questions beyond software: what is recorded, where the footage is stored, who sees it, and what your drivers have been told. Cameras inside a vehicle carrying patients capture riders too. Agree the policy with drivers before switching anything on, and ask the vendor how long footage is kept.

Claims you cannot test

Marketing numbers such as "save 4 to 5 hours every day" or "reduces manual workloads by up to 66 percent" come from vendors' own pages and from their chosen customers. They may be true for those customers. Put them in the "ask for a reference" pile and test your own day instead.

Data questions for rider information

NEMT software holds names, addresses, appointment times and sometimes medical needs. The vendor should answer these in writing:

1. Where is the data stored, and who can access it?
2. Do call recordings, dashcam footage or trip notes go to outside AI providers?
3. Is any of it used to train models?
4. How long are recordings and location history kept, and can you shorten that?
5. What happens to your data if you cancel?

If your contracts include health information rules, ask your broker or an adviser what they require of your software supplier. The general vendor questions in questions to ask an AI vendor before you sign anything and the audio-specific ones in AI scribe security: audio, retention and vendor questions apply here too.

A two-week switch plan

• Days 1 to 3: list your brokers, the formats each wants, and your standing trips. Ask the vendor to confirm each broker link in writing.
• Days 4 to 7: load vehicles, drivers and standing orders. Train two drivers on the app and have them run real trips.
• Days 8 to 10: run the new system alongside the old for one full week of trips. Compare trips and times each evening.
• Days 11 to 12: produce a trial invoice for that week and compare it to the old system's invoice.
• Days 13 to 14: move all drivers, keep the old system read-only for a month, and keep exporting data from it.

Do not switch in the final week before a billing close. Do it right after one, so any gaps are caught in a fresh cycle.

A quick scorecard to finish the demo

Once you have run the tests, write one line per vendor in this form (illustrative): "Vendor X: imports both brokers' trips, scheduled 56 of 60 on the test day, flagged 38 incomplete trips before billing, drivers preferred the app, $276 a month for four active vehicles, no annual contract, call-answering assistant not tested." A line like that is easy to compare a week later, and it keeps marketing language out of the decision. If a field is blank because you did not test it, say so, and do not fill it from the vendor's brochure.

Deciding between two shortlisted products

If both pass the first three groups, decide on the last two: the exceptions list in billing and the driver app. Ask the two drivers who tested the apps which one they would choose; they will tell you within a day. After that, price, contract terms and the AI extras can break the tie. If you would like someone to go through the shortlist with you, that is what my AI implementation consultation is for. If not, start by timing your own scheduling hours for a week, because that number tells you what any tool is worth to you.

Questions small NEMT operators ask about software and AI

Do I need AI in NEMT software?

Not to run a small fleet well. Dispatch, trip import from brokers, a driver app and billing exports matter far more. AI shows up in route planning, call answering and dashcam coaching. Treat each as an optional extra and test it on your own trips before paying for it.

How is NEMT software usually priced?

Most vendors price per vehicle per month or by fleet band. Bambi publishes $69 per active vehicle a month, NEMT Cloud Dispatch $149.99 a month flat for up to five vehicles, and Tobi from $60 per vehicle a month. NEMT Platform says its pricing depends on fleet size and features. Third-party directory figures are often out of date, so confirm any number on the vendor's own page.

What does trip verification mean in software?

It is the record that a trip happened as billed: times, location points, and sometimes a signature or note. Your broker or payer contract decides what they accept. Ask each vendor which of those fields their driver app captures and whether they appear in the billing export.

Can I switch software mid-contract with a broker?

Often yes, but check the integration. If trips arrive from a broker through a live link, your new vendor must support that same broker before you move. Run both systems for a short overlap, and compare a week of trips and invoices before switching off the old one.

Further reads

• HVAC Dispatch Software Compared: Which Has AI Routing for Small Fleets? — How AI routing claims compare in another small-fleet trade.
• AI Job Scheduling and Dispatch for Trades and Field Service Teams — AI scheduling and dispatch for teams out on the road.
• Process Server Software: Dispatch, Proof of Service and Where AI Fits — Dispatch and proof records compared in a different trade.
• AI Dispatcher for a Towing Company: Calls, Locations and Quotes — How an AI dispatcher handles calls and locations.
• Can an AI Dispatcher Replace Your Plumbing Office Dispatcher? — Whether an AI dispatcher can replace an office dispatcher.
• How to Measure an AI Receptionist's Return in the First 90 Days — Measure an AI receptionist before you keep paying for it.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Bambi home and pricing pages; NEMT Platform home page and FAQ; NEMT Software home page; NEMT Cloud Dispatch home and pricing pages; Tobi Cloud pricing page; RouteGenie home page (all read 4 November 2026). Check every price with the vendor.