SplitEV — AI billing engine for shared EV charging

01Challenge

( 1 ) Challenge

Shared EV charging is growing fast, but the billing behind it is stuck in the dark ages. Multiple drivers on one charger, fluctuating electricity rates, overlapping sessions, and zero transparency into who actually owes what. The result? Manual reconciliation headaches, billing disputes, and a user experience that undercuts the promise of clean energy.

SplitEV came to us with a bold thesis: if you can make cost-splitting invisible and trustworthy, you unlock the next wave of shared charging adoption. They had Techstars backing, a top-8-of-7,000 ranking from Morgan Stanley, and partnerships with AWS and NVIDIA. What they needed was an AI engine and product experience that could deliver on that promise.

Shared EV charging is growing fast, but the billing behind it is stuck in the dark ages. Multiple drivers on one charger, fluctuating electricity rates, overlapping sessions, and zero transparency into who actually owes what. The result? Manual reconciliation headaches, billing disputes, and a user experience that undercuts the promise of clean energy.

SplitEV came to us with a bold thesis: if you can make cost-splitting invisible and trustworthy, you unlock the next wave of shared charging adoption. They had Techstars backing, a top-8-of-7,000 ranking from Morgan Stanley, and partnerships with AWS and NVIDIA. What they needed was an AI engine and product experience that could deliver on that promise.

02What we built

( 2 ) What we built

An AI engine that makes every charge explainable

We designed an AI engine that turns raw charging data (session metadata, user and vehicle IDs, charger location, timestamps, rate signals) into something people can actually trust: a clear, fair bill.

The system runs a three-stage pipeline:

Allocate. Splits costs across drivers and vehicles with an auditable, defensible breakdown. No black boxes.

Detect. Catches anomalies and edge cases before they become billing errors, flagging them for review instead of silently misbilling.

Surface. Translates every allocation into a plain-language output so users understand exactly what they owe and why.

An AI engine that makes every charge explainable

We designed an AI engine that turns raw charging data (session metadata, user and vehicle IDs, charger location, timestamps, rate signals) into something people can actually trust: a clear, fair bill.

The system runs a three-stage pipeline:

Allocate. Splits costs across drivers and vehicles with an auditable, defensible breakdown. No black boxes.

Detect. Catches anomalies and edge cases before they become billing errors, flagging them for review instead of silently misbilling.

Surface. Translates every allocation into a plain-language output so users understand exactly what they owe and why.

03How we got here

( 3 ) How we got here

1

Mapped every scenario that breaks manual splitting

Multi-driver sessions, mid-charge rate changes, shared infrastructure across sites. We started where the complexity lives and catalogued every billing edge case.

2

Separated prediction from deterministic logic

Demand patterns and cost optimization belong to the AI layer. Allocation rules and audit trails stay deterministic. That distinction shaped the entire architecture, keeping the engine smart without making the billing opaque.

3

Productized explainability

Every split is traceable, every charge is reconcilable, and every user sees a bill that makes sense. We didn't just make allocations accurate, we made them understandable. The external validation followed (Techstars funding, Morgan Stanley recognition, AWS and NVIDIA partnerships) but the real proof is in the product: billing that users don't have to think about.

1

Mapped every scenario that breaks manual splitting

Multi-driver sessions, mid-charge rate changes, shared infrastructure across sites. We started where the complexity lives and catalogued every billing edge case.

2

Separated prediction from deterministic logic

Demand patterns and cost optimization belong to the AI layer. Allocation rules and audit trails stay deterministic. That distinction shaped the entire architecture, keeping the engine smart without making the billing opaque.

3

Productized explainability

Every split is traceable, every charge is reconcilable, and every user sees a bill that makes sense. We didn't just make allocations accurate, we made them understandable. The external validation followed (Techstars funding, Morgan Stanley recognition, AWS and NVIDIA partnerships) but the real proof is in the product: billing that users don't have to think about.

AI Billing Engine for Shared EV Charging

An AI engine that makes every charge explainable

An AI engine that makes every charge explainable

Find Station

Secure Payment

Charging Status

Session Receipt

Nearby Stations

Let's talk about what's next