EVA · Engine for Virtual Anticipation
EVA: Nereva’s anticipation engine, under development
Development is under way and the decision remains human.
EVA is the anticipation engine Nereva is developing to help teams use their operational data, requests, rotations, waits, cancellations, and identify certain situations earlier. This page sets out what exists today, what is under development and what remains a vision.
15-day free trial · 2 licences included · No payment method required
Where EVA stands
- Today
- The platform records operational data and analytics makes it readable. EVA builds on this base; its functions are under development.
- Under development
- Schedule proposals, tension detection, waste recognition on photos. Validated by your teams.
- Vision
- Broader anticipation, new variables, explainable proposals.
- Availability
- Announced when the functions are ready, in the tools your teams already use.
Why EVA exists
Frequent, variable, interdependent flows
Site transport is still decided over the phone: a delay, a cancellation, a skip full earlier than expected, and the day is reorganised by hand. EVA is Nereva’s research and development programme to help anticipate these situations from the data the platform already records.
Requests that move
Modified slots, cancellations, requests postponed by the driver: each change affects the carrier’s schedule and the site.
Limited resources
Trucks, drivers, skips, sorting centre slots: the dispatcher arbitrates every day with what they have.
Data already there
Requests, schedule, timestamped steps, waits, cancellations: the platform records what happened. That is EVA’s raw material.
Production, development, vision
Three statuses, clearly distinguished
Each function is presented with its actual status, identical in the site’s four languages.
What the platform does today, without EVA
- Record requests, the schedule, timestamped steps, photos, notes, cancellations and waits
- Make this data readable: dashboard, statements, per-truck reports, history (Analytics page)
- Show the schedule per truck and per driver, with each driver’s online or offline status
What is under development, under the EVA name
- A schedule proposal: mission order, groupings, slots, to be validated by the dispatcher
- Detection of the day’s tensions and adjustment proposals
- Waste recognition on mission photos
What remains a vision
- Broader anticipation of situations, with new variables
- Explainable proposals, whose reasoning teams understand
- Gradual integration into existing tools: schedule and apps, never an autopilot
EVA is under development: scheduling, assignments and day-to-day decisions remain, today, those of your teams.
EVA’s raw material already exists: the data of your operations
Each request executed in Nereva leaves a trace: site, client, truck, driver, timestamped steps, photos, notes, cancellations. The schedule and the dashboard make them readable right now, without automatic anticipation.
- Schedule per truck and per driver, in daily or weekly view, with the requests to be scheduled
- Dashboard: closed requests, kilometres driven, skips on sites
- Statements, per-truck reports and history per site, per client and per truck
Under development
What EVA is meant to bring, once ready
The functions below are under development. They are designed as proposals submitted to the teams: the dispatcher validates, modifies or ignores. Their availability will be announced when they are ready.
Schedule proposal
A mission order, groupings and slots proposed from the requests received; the dispatcher stays in control.
The day’s tensions
Spotting converging slots, likely waits, rotations to shift, and flagging them with a concrete proposal.
Mission photos
Recognising waste on the photos taken by drivers, to help document the full skip.
Data used
Which data, under which conditions
Nereva’s AI and data policy separates two zones. No training data is identifying; no sensitive data is used.
| Point | What the AI and data policy says | Source |
|---|---|---|
| Two separate zones | Zone A, operational and pseudonymised, reserved for providing the service; Zone B, irreversibly anonymised and aggregated, the only one used for training. The transformation from A to B is one-way and traced. | AI policy, art. 6 |
| Categories used | Requests (types, quantities, time windows); scheduling and execution (sequences, durations, steps, statuses); positioning, anonymised and aggregated; document types; technical usage data; sector context. | AI policy, art. 7 |
| Geolocation | Driver and vehicle identifiers removed, routes aggregated into segments with no possible reconstruction, degraded resolution, periodic re-identification tests. Never raw or pseudonymised data for training. | AI policy, art. 9 |
| Purposes | Improvement of features, development of new functions, sector statistical analyses. Never to evaluate, sanction, monitor or compare identified drivers. | AI policy, art. 10 |
| Where and for how long | Training on infrastructure in Switzerland or the EEA; anonymised training data kept for 36 months at most. | AI policy, art. 18 and 19 |
| Ownership | The Client’s data remains its property; anonymised derived data and the models belong to Nereva. | Terms of Use, art. 10 |
Human control
EVA proposes, your teams decide
Analytics and operational intelligence functions produce indications of an indicative nature: they inform the judgement of your teams, who keep the decision.
Indicative, never imperative
A proposal is validated, modified or ignored. Nothing is triggered without a person (Terms of Use, art. 14; AI policy, art. 22).
No automated decision about people
No decision producing legal or significant effects is taken automatically; geolocation is never used to monitor, rank or sanction a driver.
A declared framework
The models are classified as limited or minimal risk AI systems within the meaning of the AI Act; an AI referent answers at privacy@nereva.com; each model is validated before deployment and documented.
AI, data and privacy
The documents and pages to consult
- AI and data policyData lake zones, anonymisation, purposes, human control
- Data hostingWhere your data goes, who accesses it, for how long
- SecurityAccess, traceability, service security
- Analytics and reportsWhat the platform already reads in your data
- Forecasts and tensionsThe status of each anticipation capability
- Alerts and notificationsThe notifications that exist today
- Developments in progressThe product work in progress
Your questions
EVA, plainly
Does Nereva use artificial intelligence in production?
EVA, Nereva’s anticipation engine, is under development; its availability will be announced when the functions concerned are ready. The functions available today are those of the platform and analytics.
What does EVA mean?
Engine for Virtual Anticipation: Nereva’s anticipation engine. The name is not translated.
What will EVA actually do?
The functions under development: propose a mission order, groupings and slots from the requests; spot the day’s tensions and propose adjustments; recognise waste on mission photos. Always proposals: your teams keep the decision.
Does EVA take decisions instead of the teams?
No. The outputs of the models are of an exclusively indicative nature; they do not constitute automated decisions. The dispatcher validates, modifies or ignores a proposal.
Is our data used to train EVA?
Only irreversibly anonymised and aggregated data is used for training, in a zone separate from your operational data. Your identifying data never enters it; you remain the owner of your data. The AI and data policy describes each category used.
Does EVA learn over time?
The models are trained on the platform’s anonymised data; the more documented flows there are, the richer the training base.
When will EVA be available?
Each function will be announced when it is ready, and delivered in the tools your teams already use.
Does Nereva step into the commercial relationship?
No. Your partners and your business relationships remain yours. Nereva synchronises operations without stepping into the commercial relationship. It takes no commission on flows.
Other questions? The general FAQ repeats these answers; the Developments in progress page lists the product projects.
Go furtherYour operational data today will feed EVA, under development.
15-day free trial with 2 licences. No payment method required. Monthly subscription afterwards, no long-term commitment.