Every inspection, incident, and repair on one evidence trail — AI-screened.
- Company
- Aslan Kurierdienste GmbH
- Where
- Germany · Amazon DSP / last-mile
- Fleet
- 36 vans · 42 drivers · 3 dispatchers
- Plan
- Enterprise — all modules on
- Live since
- mid-March 2026 (~four months)
On 10–20% margins, undocumented damage is the cost you can't see.
Amazon DSPs run thin, typically 10–20% net, and non-warranty repairs and excess wear are the operator's bill — paid out of pocket, charged back when the van is returned. Industry analysis of last-mile fleets puts the average van at around eight separate damage points at any time, against a roughly 20% annual accident rate per vehicle. When damage isn't captured the moment it appears, a DSP pays three ways: leakage it can't pin to a shift, unplanned downtime at €410–€700 per van per day, and disputes lost to whoever holds the evidence. The job Aslan hired Fleet to do: capture defensible condition evidence on every shift, and triage it without burying a three-person back office.
Pilot to full yard in two weeks.
| Week of | Shifts | Drivers active | Vans used |
|---|---|---|---|
| Mar 16 (pilot) | 3 | 1 | 3 |
| Mar 23 (pilot) | 2 | 1 | 1 |
| Apr 20 (pilot) | 2 | 1 | 2 |
| Apr 27 (rollout) | 47 | 26 | 23 |
| May 4 | 133 | 30 | 31 |
| May 11 | 104 | 28 | 25 |
| May 18 | 120 | 31 | 29 |
| May 25 | 114 | 33 | 29 |
| Jun 1 | 112 | 28 | 32 |
| Jun 8 | 120 | 28 | 28 |
| Jun 15 | 122 | 29 | 25 |
| Jun 22 | 91 | 23 | 23 |
| Jun 29 | 89 | 22 | 22 |
| Jul 6 | 97 | 24 | 27 |
| Jul 13 | 89 | 24 | 24 |
Evidence discipline became automatic
Drivers captured at least one inspection photo on 99.6% of completed shifts, and six or more on 97.8% — 7,710 time-stamped photos across four months. Every van now carries a continuous shift-over-shift before/after record. Damage has a provable origin window instead of a shrug.
AI carried the triage load
When AI photo review switched on in mid-May, it compared inspections against the van's previous shift: 670 unique shifts were auto-reviewed, 347 localized findings were stored with 331 flagged as new versus the previous inspection, and 16,778 image-analysis inputs were processed. Analysis inputs can include reanalysis; 967 unique shifts had at least one AI analysis.
Incidents and repairs on one trail
Aslan documented 8 incidents and 8 repair jobs — €8,105 of repair work against €8,220 in recorded damage cost, including a €7,080 collision repaired across two jobs. Seven jobs went through Aslan's own workshop and one to an external body shop; every case is photographed and claim-ready instead of a month-end scramble. Five of the eight incidents predate go-live — historical cases backfilled during onboarding — so only three fall inside the four-month live window. That is too small a sample to call an accident rate, though it sits in the range of the industry's ~20% annual benchmark.
in addressable damage exposure per year — recoverable, charged back, or caught early on a fleet this size.
That's Aslan's real activity costed at published industry benchmarks: roughly €16,500–€28,900 a year for a DSP this size. A mid- to high-single-digit return on an Enterprise subscription. These euros are an illustrative estimate, not money Aslan has booked.
How we modeled this (assumptions & sources)
Benchmarks: ClearQuote (fleet damage), Automotive Fleet (~20% accident rate), Michelin & Penske (downtime cost), Route Consultant & Tullis (DSP margins), Amazon DSP Fleet Program. Full citations on file.
Damage you can prove is damage you can recover or refuse to pay for. Damage you can't prove, you eat.
What Aslan isn't doing yet
Today Aslan uses Fleet as a condition-evidence and incident system of record. The step that turns this evidence into booked euros — formal insurance-claim records, claim-pack submission, at-fault liability, and recovered-amount tracking — isn't being captured in-app yet. With four months of airtight evidence already stacking up, that's the natural next phase, and where the modeled numbers above become real ones in the next revision of this story.
Aslan Kurierdienste is Fleet’s development partner, and a Fleet stakeholder holds an interest in the company. The figures shown are Aslan’s own operational data.
“Fleet replaced our WhatsApp groups, our spreadsheet, and three different paper forms with one workflow the drivers actually use. The photos are the evidence — there is no more arguing about damage.”
Run your DSP like Aslan does.
Photo-based shift starts. AI-triaged inspections. Incidents and repairs on one evidence trail.