Shoosmiths

City law firm Shoosmiths launched Project Apollo, a generative AI tool built with Microsoft to assist junior lawyers in contract review. Follows year-long pilot; part of multi-million pound AI investment.

Leadership read: Project Apollo's operational weight sits not in the automation it delivers but in the institutional knowledge it encodes. Shoosmiths spent a year extracting deal-making logic from senior M&A lawyers, structuring it into a system that explains not just what changed in a contract but why. That is a fundamentally different data architecture than a generic LLM wrapper; the firm has now committed to maintaining, versioning, and governing a proprietary knowledge layer that will drift from the underlying model over time and require continuous expert curation. The senior sign-off requirement is both a risk control and an implicit acknowledgement that the system's outputs carry the firm's liability, not Microsoft's. The related signals set for this period is broad across product launches but thin in professional-services AI specifically; this is one of the cleaner legal-sector deployment signals in the last 90 days, sitting alongside the Pinsent Masons hallucination incident and the Lush trial model development as evidence that UK law firms are moving from experimentation into production, with governance failures visible in real courts. The pattern is consolidating around two poles: firms building proprietary models trained on internal expertise, and firms absorbing the reputational cost of deploying generic tools without adequate review infrastructure. Across firms at this stage of proprietary AI deployment, the functional pressure concentrates in legal-tech product ownership (people who can manage a knowledge system as a product, not a project), AI governance and risk operations, and the training and curriculum design capability to make junior development through AI a measurable outcome rather than a marketing claim.

curated · 2026-06-24 · context →