Skip to content
Titan Intelligence · Powered by Bonisa AI

Stop the Loss
Before the Deal Is Done.

Titan already keeps vehicles where they belong. Titan Intelligence adds a risk score at the point of transaction, so your team can spot the deals likely to end in theft, fraud or damage while there's still time to change them.
No credit dataReason codes on every scoreFree proof of concept
Transaction checkHigh risk

Rental booking 48213 · 2025 Chevrolet Tahoe

362Score 300–700 · higher is safer
Reason codes
  • Booked 35 minutes before pickup
  • One-way to a border-state branch
  • Device linked to 3 other bookings

Your policy: manager review and extra ID

Illustrative screen. Customer shown as a hashed ID; no credit data.

Prevention, now with intelligence

Titan stops the theft. Intelligence flags the bad deal.

Many losses are set before the vehicle leaves: a straw buyer, a booking ring, a deal that was never meant to be paid. The warning signs are often already in your data. Titan Intelligence reads them at the moment you can still act.

On the vehicle

Titan prevention

Immobilization, live location and tamper alerts, managed in the Partner Service Dashboard (PSD) and Halo.

At the transaction

Titan Intelligence

A risk score with plain-English reason codes on every rental, lease or deal, before it closes. Your policy decides what happens at each score.

How the score works

Built from your data, not credit data.

Modelled by Bonisa AI's data scientists on your own past transactions. Delivered and supported by Titan.

The transaction

Timing, channel, branch or rooftop, term, one-way or local, and how the deal was put together.

The vehicle

Make, model, class and value, so high-target SUVs and trucks stand out.

Customer and payment, as hashed IDs

History with you, license state, distance between billing, delivery and branch ZIPs, card type, and device or IP hashes that link applications across locations.

Your outcomes

Which past deals ended in theft, fraud, damage or loss, so the model learns your patterns, not someone else's.

Act on the right deal, at the right time.

Stop losses before handover

Flag high-risk transactions before a vehicle ever leaves the lot.

Fewer false alarms

Tell a late return or a slow payment from a theft, so good customers aren't treated as suspects.

Keep good customers

Flag only the riskiest deals, so legitimate customers move faster, not slower.

Decisions you can defend

Plain-English reason codes behind every score your team acts on.

Prevention where it counts

Put Titan on the vehicles the score says are most at risk, and arm them before they leave.

Proof before you pay

A free blind test on your own past transactions before you commit.
Use case · Rental operators

Know which rentals will go wrong, before they do.

Score every booking before the keys are handed over with Counter Check, then keep scoring on the road with Live Score once Titan units are fitted.

Straw renters

Renters with clean records collect vehicles for someone else. Credit checks miss them.

Booking rings

One phone, device or batch of cards books high-value vehicles at several branches.

Fast exits

SUVs and trucks head for border corridors and staging stops, where trackers are swept or jammed.

Blunt rules

“Overdue” and “near the border” alerts fire on good customers, and too late on real thieves.

Counter Check: review, steer, watch or fast-track

You set the cut-offs and what happens at each one. The middle 15% keep your standard process.

Riskiest 1%

Review

Manager review and extra ID before any keys change hands. Your policy decides whether to decline.

Next 2.5%

Steer

A standard car instead of a high-target SUV or truck, or a larger hold.

Next 6.5%

Watch

Rent normally. Staff get an alert two hours before the return time if the car isn't back.

Safest 75%

Fast-track

Skip the counter, lower holds and loyalty perks.

80%
Of airport renters

still visit the counter.

8 min
Faster pickup

for those who skip it.

+42
Points in satisfaction

when they skip it.

Source: J.D. Power 2025 rental study, via Auto Rental News.

Live Score: know when an overrun isn't an overrun

A later phase. With Titan telemetry and device alerts (jamming, battery disconnect, OBD tamper), the score keeps updating until the vehicle is back, where state law allows. An overdue rule only reacts once the rental is late.

Late, booking score only
Score stays highA safe-score renter two hours late gets grace and a polite reminder.
Score dropsA risky-score walk-in is flagged two hours before the return time.
Late, with Live Score
Score stays highStill driving normally, in usual areas, or heading back toward a branch.
Score dropsHeading away from every branch after a risky booking, or the unit stops reporting.
Near the border, with Live Score
Score stays highA low-risk renter at normal speed, passing through without lingering.
Score dropsA high-target SUV or truck that stops at a staging zone, or reports jamming or power loss.
What your team does
Score stays highSend a reminder. No alarm, no police report.
Score dropsGet an alert while the vehicle can still be recovered.
Illustrative · one example branch

Everyday rental losses

Bookings a month
10,000
Bad outcomes a month
92
Average write-off
$2,500
$230,000
a month in everyday losses

Unpaid overruns of 7+ days, domestic conversions, write-off accidents and chargebacks. Border thefts are left out: Titan prevention is built to stop those vehicles.

If the score helps you avoid…

Losses avoidedA monthA year
1 in 10 losses$23,000$276,000
1 in 5 losses$46,000$552,000
1 in 3 losses$76,700$920,000

Plus paid checks: send credit, license and ID checks to flagged bookings only. A driving record alone averaged over $10 a report, so every 1,000 checks you skip saves about $10,000.

Illustrative assumptions from Bonisa AI's branch model; your proof of concept replaces them with your own figures. Driving-record fee: Verisk (2020 average state fee).

Use case · Dealer leases

Score the lease before the vehicle is delivered.

A lease hands a new vehicle to a customer for years. When the application is a straw lessee, a stolen identity or a vehicle bound for export, the loss is set the day it's delivered.

What it flags

  • Applications that share a phone, device or address with other recent applications
  • High-target vehicles on rushed deals, or delivery far from the customer's stated home
  • Patterns your own past lease losses share: skips, conversions and vehicles never seen again

What your team does

  • Extra identity and stipulation checks on flagged applications only
  • Fit and arm Titan before delivery on the riskiest deals
  • Move the clean majority through with no added friction

Through the term

With Titan fitted, Live Score can keep watching for tamper alerts, unusual movement and units that stop reporting, where state law and the lease allow, so a missing vehicle is found while it can still be recovered.

Titan Intelligence adds fraud and loss signals alongside your lender's credit decision. It doesn't replace it and uses no credit bureau data.

Use case · The F&I desk

A second opinion while the deal is still on the desk.

F&I is the last stop before the keys change hands. A score there gives your manager reason codes to slow the risky deal down, and the confidence to move the clean one through.

Catch fraud before funding

Flag deals with linked devices, mismatched addresses or rushed timing for extra verification, before the contract goes to the lender.

Match protection to risk

See where theft risk is highest by vehicle and deal, so the Titan recommendation fits the customer's real exposure.

Move clean deals faster

Most deals score safe. Spend desk time on the few that need it.

Consistent and on record

The same reason codes for every customer, with cut-offs and actions your store sets and records.
Your team stays in control

The score recommends. Your team decides.

At the transaction

A score and reason codes for each rental, lease or deal, sent by API into the system your team already uses.

On the road

Alerts in PSD on the desktop and Halo on the lot, where staff check the vehicle and act. Risk alerts are a planned addition to both.

On record

Every vehicle command and its confirmation is kept in PSD's Activity log.

Guardrails

  • No automatic immobilization. Staff always confirm it.
  • You set the cut-offs and what happens at each one.
  • Live Score only where state law allows tracking.
  • No credit data. Customers are hashed IDs, never names.
  • Success and false-alarm limits agreed before go-live.
Free proof of concept

We prove it on your data, in a blind test.

Month 1 trains the model, month 2 checks it, and month 3 is scored blind with its outcomes held back. No cost, no integration, no credit data and no obligation to continue.

  1. 01

    Agree

    Sign an NDA and data agreement. Agree how success is measured.

  2. 02

    Share

    Send 3 months of transactions, good and bad. Hold back month 3's outcomes.

  3. 03

    Build

    We train on month 1 and check the model on month 2.

  4. 04

    Score blind

    We score every transaction in month 3 and send you the scores first.

  5. 05

    Reveal

    You release the outcomes. We show which losses we flagged.

Every transaction gets a score band. You choose where to act.

Your readout shows the losses in each band, how many good customers each cut-off would flag, the savings, and the reason codes behind each flagged deal.

In Bonisa AI's Titan South Africa model, the riskiest 10% of applicants held 30% of the bad accounts.

Score bandShare of applicantsBad rate
Band 1 · safest12.69%0.00%
Band 223.02%1.25%
Band 310.73%3.76%
Band 411.94%4.35%
Band 515.12%13.74%
Band 616.16%25.36%
Band 7 · riskiest10.33%30.17%

Example of the format: Bonisa AI's payment-risk model for Titan South Africa's tracker contracts, in seven bands from safest to riskiest. Your readout uses your transactions and your losses.

Case study · Consumer finance

The same blind test, proven in lending.

A UK consumer finance provider wanted better approval decisions than its credit bureau score gave. A model was built on its past applications, then tested on 990 later applications it had never seen: the same kind of blind test as your proof of concept.

18%
Stronger separation

than the bureau score, between good and bad customers.

+21.2%
More loans approved
+23.75%
More lent

An extra £778,145.

−18.55%
Lower charge-off rate

All at a similar or lower bad rate.

Estimates before reject inference. Source: case study on known performance, August 2024. Client name withheld.

Commercial lending

Credit approval with thin data

Bad rate down from 15% to 13.4%. Lending up 6%. Revenue up 8.1%.

For you: works even when a customer has little history.

Contact centres

Ranking the risk of every lead

Higher sales hit rate. Fewer man-hours.

For you: your team works the deals that matter.

Mining

A model that updates itself

A model of over 60 parameters, automated to update as new data arrived. No manual re-fitting.

For you: the same approach keeps the Live Score current.

Source: Matogen AI case studies. Client names withheld.

Powered by Bonisa AI

The data scientists behind Titan Intelligence.

Bonisa AI is a Matogen AI company, best known for credit scoring, fraud detection and alternative data.

2007
Founded
250+
Projects completed
4 PhDs
In a team holding 44 university degrees
6
Continents with clients, with a US branch

Titan

Your single point of contact: contract, rollout, support and billing, plus Titan units, PSD and Halo.

Bonisa AI

Builds, hosts and runs the models, and keeps them retrained on your patterns.

Next steps

Four steps to your free proof of concept.

  1. 01Sign an NDA and data agreement
  2. 02Choose the locations, and name a data owner
  3. 03Agree the success and false-alarm measures
  4. 04Send 3 months of transactions, good and bad

The proof of concept has no cost and no obligation. Ongoing scoring is quoted for your business after the results readout.

We use your details only to respond to this request.

Prove it on your own transactions, free.

A blind test on 3 months of your past rentals, leases or deals. No integration, no credit data, no obligation.