AI revenue management for car rental
The autonomous revenue operating system for car rental.
Quantum Kraken continuously shops the market — and is designed to combine multiple commercial sources — pairing those observations with historical performance, booking pace against history, fleet pressure, rental length, booking window, local events and demand signals, brand strategy and operator-defined safety rules. It then recommends — or safely executes — the right price for each exact scenario: brand, location, category, pickup date, rental length and booking window.
Every decision is explainable, governed, auditable and progressively confidence-scored. Human revenue managers still own strategy, exceptions and commercial priorities — the machine provides the coverage no team can sustain by hand.
Every market. Every location. Every category. Every rental length. Every booking window. Re-evaluated many times a day.
Operating in a live multi-brand, multi-location rental environment.
The reference operation includes Green Motion Morocco and U-Save Morocco — owned by the same parent company, yet run as two distinct commercial brands with separate positioning and separate pricing logic. The platform is built for operators globally.
Decision pipeline
live signal graph- Market observations
- Historical performance
- Booking pace vs history
- Fleet pressure
- Rental length & window
- Events & demand
- Brand strategy
Exact-scenario decision
versionedbrand · location · category · pickup date · LOR · window
synthetic illustration · illustrative confidence
Protected: floors, corridors, caps and anomaly checks
Controlled execution with explanation and audit
- Born inside live rental operations
- Multi-brand, multi-location, exact-scenario aware
- Continuous shopping and governed execution
- Explainable today, learning by design
The problem
Car-rental revenue management breaks at scale.
The challenge is not simply finding a competitor's price. It is deciding whether that price should influence your own rate for this specific brand, location, vehicle category, pickup date, rental length and booking window — while considering booking pace, future fleet availability, historical performance and commercial strategy. That decision must then be executed, monitored and repeated across an entire network, many times every day.
Combinatorial scale overwhelms manual teams
Millions of live and future combinations of brand, location, category, pickup date, rental length and booking window cannot be continuously inspected by humans, however skilled the team.
Signals remain disconnected
Market prices, bookings, pace, fleet position, events, demand and commercial rules usually sit in separate tools, so no single system sees the whole decision.
Pricing and fleet are inseparable
A rate that looks profitable in isolation can still create a shortage in the wrong city or the wrong category days before its strongest demand window.
Reactive workflows are structurally too slow
Manual shopping, spreadsheet reconciliation and repetitive publishing always lag behind market movement — the network moves faster than the workflow.
Pricing in car rental is not a lookup. It is a continuous, constrained operating decision across market, demand, inventory, time, location, category, brand and distribution channel.
The market contains valuable rate-shopping, forecasting and revenue-management products. Quantum Kraken was created because the available solutions did not match the granularity, configurability, automation depth, safety model and economics required by the reference operation.
How Quantum Kraken works
Observe. Normalize. Understand. Decide. Protect. Execute. Verify. Learn.
A closed commercial loop designed around how rental networks actually run — not around how a dashboard looks. The deterministic shopping, normalization, decision, protection and execution core operates today, while deeper operational context, verification and learning are being integrated.
Observe
Market rates and availability are shopped continuously today. Historical performance, forward capacity, event calendars and external demand signals are being integrated into the same observation layer.
Live market shoppingHistory (integrating)Capacity (integrating)Events (integrating)Normalize
Suppliers, vehicle groups, station structures, brands, currencies, rental lengths and booking windows are reconciled into one decision-ready model.
Supplier mappingVehicle groupsStationsCurrenciesUnderstand
The system determines what changed, whether it matters, which dates are exposed and which combinations deserve attention right now.
Change detectionMaterialityExposed datesPriorityDecide
A specific recommendation is produced for the exact brand, station, category, pickup date, rental length and booking window — with reasons and alternatives.
Exact scenarioReasonsAlternativesConfigurable logicProtect
Brand strategy, floors, corridors, caps, anomaly detection, category relationships and operator-defined rules are applied before anything can exist.
Floors & capsCorridorsAnomaliesOperator rulesExecute
Decisions publish automatically when configured deterministic safety conditions are met, and otherwise route to human approval. Confidence-aware routing is entering the decision loop.
Controlled publishingApproval routingModesInstant pauseVerify
Decisions are compared with the resulting bookings, market movement, utilisation and realised outcomes so effect is measured, not assumed.
BookingsMarket movementUtilisationOutcomesLearn
Recommendations, overrides and outcomes are structured as future training feedback. This learning layer is being developed and is not yet fully autonomous.
Feedback structureOverridesCalibrationRoadmap
Platform modules
Six modules, one closed loop.
Each module is independently useful and jointly required — market observation, operational truth, decisioning, governance and accountability. Labels show what is live today and what is entering the loop.
Autonomous Market Shopping
Continuous market shopping and availability collection by location, pickup date, rental length and booking window, normalized into one comparable model — with an architecture designed to orchestrate multiple commercial sources.
Revenue Decision Engine
Exact-scenario recommendations for brand, location, category, pickup date, rental length and booking window, using operator-configurable logic.
Historical & Booking-Pace Intelligence
Historical performance, booking pressure and pace measured against expected curves, entering the decision context alongside the market view.
Fleet & Network Intelligence
Future capacity, city pooling, category substitution, upgrades and one-way flows — with fleet and defleet recommendations on the roadmap.
Governed Automation & Execution
Advisory, approval-based and controlled-autonomous modes per brand, station or category, with guardrails enforced before publishing.
Decision Ledger & Explainability
Complete evidence, rules, constraints, versions, approvals and outcomes for every decision — with step-through replay.
Built for car rental's real complexity
Eight dimensions, priced together — many times a day.
Rental revenue lives in the interaction between these dimensions. Quantum Kraken treats them as one exact-scenario decision space instead of eight disconnected reports.
42,310,800
Decision cells in a representative two-brand network — 23 locations × 28 categories × 90 rental lengths × 365 pickup dates — before booking windows, channels and repeated daily refreshes are added.
Exact scenario
Every decision is taken for one specific brand, station, category, pickup date, rental length and booking window — never for an averaged segment.
Repeated refresh
Humans define strategy and exceptions. Machines provide complete coverage and consistent execution across every refresh cycle, every day.
Dimension 01
Locations
Airport, city and resort stations behave like separate markets with separate demand curves and separate supply.
A rate that is correct downtown can be wrong at the airport terminal 12 km away.
Ask Kraken · the end-state interface
Run the business through conversation.
Every rental owner should be able to ask what is happening, why it is happening, what will happen next and what the business should do — without navigating every dashboard.
Try a question
Ask Kraken
synthetic illustrationWhy are Marrakech compact cars underperforming?
Ask Kraken is the natural-language operating layer above every module — market signals, decisions, fleet pressure, governance and the decision ledger — answering in plain language with the evidence behind each answer.
- Native-language and voice interactionroadmap
- Companion mobile app for alerts and approvalsroadmap
- Advanced scenario simulation and executionroadmap
- Grounded answers across every modulelive in demo
Product truth
Deterministic intelligence today. A learning layer being developed.
Quantum Kraken is explicit about what runs in production, what is being integrated into the decision loop, and what remains ahead. The deterministic core is in operation; the machine-learning layer is being built on top of the data it already produces.
Live today
in operation- Autonomous market shopping and normalization
- Multi-brand, multi-location pricing logic
- Decisioning by location, category, pickup date, rental length and booking window
- Rental-length coverage up to 90 days
- Booking-window refresh schedules many times a day
- Floors, corridors, caps and sanity checks
- Controlled price execution
- Decision archives, audit history and Control Tower monitoring
- Production operation inside the reference two-brand environment
Entering the decision loop
integrating- Historical performance
- Booking pace versus historical pace
- Capacity snapshots and fleet pressure
- Event-calendar context
- External flight and demand signals
- Confidence scoring
Next-generation decisioning
roadmap- ML-assisted recommendations
- Outcome-based confidence calibration
- Scenario simulation before execution
- Human-approved strategy experiments
- Price elasticity and demand-response learning
- Fleet and defleet recommendations
- Upgrade, transfer and city-pooling constraints
- Ask Kraken native-language and mobile companion workflows
Capabilities listed as entering the decision loop or as roadmap are being developed and are not presented as live production features. Quantum Kraken does not claim that machine learning is fully live, nor that the system already learns autonomously.
Safety and trust
Autonomy that operators are willing to switch on.
In revenue management the hardest part is not producing a price — it is guaranteeing that a wrong price cannot escape. Blocked, pending and published outcomes are all first-class results, each recorded with its evidence. Safety is the core differentiator, not a settings page.
Preserve current price
When inputs conflict or constraints cannot all be satisfied, the existing price stands. Doing nothing is a valid, deliberate outcome.
No update on weak evidence
Thin market coverage, stale observations or low confidence produce a recorded no-update rather than a guess.
Last-good-data policy
Feed failures, timeouts and partial collection fall back to the last known good state instead of reacting to broken inputs.
Versioned decisions
Every decision carries its version, inputs and applied constraints, so any rate can be reconstructed after the fact.
Traceability
Revenue managers can follow a price from signal to rule to constraint to execution without reading code.
Human control modes
Advisory, approval-based and controlled autonomous modes can be set per brand, station or category — and paused instantly.
Confidence-aware automation, scenario simulation before execution and required human approval for higher-risk network and fleet actions are roadmap capabilities. They are being developed and are not presented as live today.
Founder
Nawfal Taki — Founder & Product Architect
An operator who built the system he needed. Quantum Kraken was not designed from the outside — it was built by someone accountable for revenue, fleet and pricing decisions inside a live rental network.
Nawfal Taki
Founder & Product Architect, Quantum Kraken
12+ years · travel marketplaces · logistics · car-rental operations · ex-Booking.com
Revenue operator. Ex-Booking.com. Builder of a production pricing system born inside real car-rental operations.
Nawfal Taki is the founder of Quantum Kraken. He brings more than twelve years across travel marketplaces, logistics and car-rental operations, including nearly a decade at Booking.com working on commercial, supply and marketplace problems at global scale, followed by hands-on responsibility inside a live multi-brand, multi-location rental operation.
He currently carries commercial responsibility across Green Motion Morocco and U-Save Morocco — two distinct brands run by the same parent company — spanning revenue management, broker distribution, fleet planning, network performance and customer quality.
That combination is unusual: marketplace pricing intuition from one of the world's largest travel platforms, paired with the daily reality of fleet utilisation, one-way flows, category substitution, brand positioning and station-level shortages. He designed the product vision, decision architecture, safety framework and data flows behind the platform, and built the system itself using Python, cloud infrastructure and AI-assisted engineering.
Quantum Kraken exists because the pricing problem was lived operationally before it was engineered — and because no available system could price a network at machine scale while respecting how the operation actually runs.
- Product vision, pricing architecture and decision logic
- Safety framework, guardrails and execution governance
- Data architecture, pipelines and operational integration
- Python, cloud infrastructure and AI-assisted engineering
Company
From a live two-brand operating system to global vertical SaaS.
Quantum Kraken was built and validated inside a real, live, two-brand and multi-location car-rental operating environment. The same combinatorial pricing problem exists for independent operators, franchise networks and regional groups in every market — and the platform is being productized as global vertical SaaS to serve them.
Productization is concrete work, not a rebrand: configurable tenants with isolated data, brand- and operator-level rule sets, connectors for market and operational sources, governance and permission models, guided onboarding and configuration, and a repeatable deployment path — so each new operator is a configuration, not a custom rewrite.
Pre-seed investors
Backing the transition from a founder-built production system into a standalone global SaaS platform.
Accelerators
Programs focused on vertical AI, travel technology and applied revenue management.
Cloud and compute partners
Infrastructure and credit programs to support the data platform and the developing ML layer.
Early design partners
Independent operators willing to shape the productized platform around real operational constraints.
We are opening conversations with pre-seed investors, accelerators, cloud partners and early design partners to productize the platform, deepen the machine-learning layer and onboard independent operators.
Contact
Request a private demo.
Tell us who you are and what you want to see. Submitting opens a prefilled email draft to our founder — no data is stored on this site.
Direct
nawfal@quantumkraken.appFor operators: we can walk through decisioning, safety constraints and automation control against your own structure. For investors and programs: we can share the architecture, the operational validation and the productization plan.
Validated in a live multi-location rental operation