Luminance bets on in-house AI over ‘rented intelligence’

TL;DR:

• UK legaltech Luminance argues that owning fine-tuned AI models beats relying on “rented” frontier models for margins and accuracy.
• The Cambridge-founded firm has raised over £87.2m (5m) across two rounds in 12 months, the latest a £56.8m (m) Series C.
• It is targeting in-house legal teams with purpose-built agents that automate contract work end to end.

As legal AI fills with firms “simply putting their logos on technology developed by others”, Luminance is staking its future on the opposite approach: models it builds itself. Chief executive Eleanor Lightbody told City AM that proprietary, fine-tuned models are “really important” because they make a business’s margins “much more palatable and sustainable” than renting frontier intelligence.

The economics of owning the model

Luminance, founded by University of Cambridge mathematicians, aggregated customer data with consent to train its own models, then layered a blended approach — its own systems alongside rented tech — to convert the “language that defines a business” into intelligence the whole company can use. Its proprietary “Luna” series targets “low-value, high-volume” contract work, and Lightbody says the firm was the first to deliver AI negotiating against AI on both sides of a deal, starting with NDAs. The commercial logic is sharp: control the model, and you control both cost and the accuracy that lets you stand out.

The funding underlines the appetite. Luminance has raised over £87.2m (5m) in 12 months, including a £56.8m (m) Series C led by Point72 Private Investments, with around 800 in-house customers and roughly half its revenue from the US. It is a notable data point in a crowded market — legaltech startups raised around bn in 2025, with rivals Harvey and Legora absorbing much of the capital.

The build-versus-buy debate matters well beyond law, and Luminance’s bet helps explain why London is pulling ahead on senior legal-AI hires.

Looking forward

For UK businesses weighing AI strategy, Luminance’s argument is a useful counterweight to the default of building atop a handful of frontier models: owning the model can mean better margins, sharper differentiation and less exposure to a supplier’s pricing. The trade-off is the cost and talent required to build and maintain models in-house — a bet only well-funded firms can comfortably make.