On June 24, 2026, UK law firm Shoosmiths launched Project Apollo, a self-developed generative AI contract review platform built with Microsoft’s support and running on Azure. The platform marks a significant step in the legal sector’s adoption of AI, combining the power of large language models with rigorous governance to deliver transparent, efficient contract analysis. Deployed across the firm’s operations, Project Apollo is not merely a technology experiment but a fully integrated tool designed to redefine how legal professionals interact with documents.

Shoosmiths, a firm known for its innovation in legal services, has spent over a year developing Project Apollo in close collaboration with Microsoft. The result is a bespoke solution that leverages Azure’s cloud infrastructure and AI capabilities while addressing the unique demands of legal work: precision, confidentiality, and explainability. Unlike generic AI tools, Project Apollo was built from the ground up with legal workflows in mind, incorporating feedback from lawyers and clients to ensure it meets real-world needs.

The platform’s core function is contract review, a traditionally time-consuming process where lawyers sift through hundreds of pages to identify key clauses, risks, and obligations. Project Apollo uses generative AI to not only flag important provisions but also to suggest improvements, generate summaries, and even draft alternative language. However, what sets it apart is its emphasis on transparency. Every AI-generated output is accompanied by an auditable trail showing the source data, reasoning steps, and confidence levels, allowing lawyers to verify and trust the results.

The Technical Backbone: Microsoft Azure

Project Apollo runs entirely on Microsoft Azure, utilizing a suite of AI and cloud services. Shoosmiths opted for Azure due to its enterprise-grade security, compliance certifications, and the ability to isolate data within the firm’s tenancy. The platform leverages Azure OpenAI Service, which provides access to OpenAI’s advanced models while ensuring data privacy—critical for handling sensitive client information.

The architecture incorporates Azure Cognitive Search for indexing and retrieving relevant contract precedents, and Azure Machine Learning for fine-tuning models on Shoosmiths’ proprietary legal data. This hybrid approach balances the power of foundation models with domain-specific adaptations, reducing hallucinations and improving accuracy. Microsoft engineers worked alongside Shoosmiths’ in-house AI team, providing guidance on model optimization, responsible AI practices, and scalable cloud architecture.

Azure’s built-in responsible AI tools, such as content filters and model monitoring, were customized to meet legal standards. For instance, Project Apollo includes automatic redaction of personally identifiable information and a “human-in-the-loop” checkpoint system that prevents automated decisions from being finalized without lawyer review. These features align with the Solicitors Regulation Authority’s guidelines on AI use in legal practice.

AI Governance and Transparency at the Core

Shoosmiths has positioned Project Apollo as a model of “Transparent AI Governance”—a term the firm coined to describe its multi-layered approach. At the heart of this are “GenAI playbooks,” detailed protocols dictating when and how AI can be used in contract review. These playbooks, developed jointly with Microsoft, cover prompt engineering, model selection, data handling, and output validation.

The platform’s user interface displays an “AI explainability panel” for every review. When the system highlights a clause as high risk, it shows the specific legal principles, past contract examples, and model predictions that led to that conclusion. This not only builds trust but also educates junior lawyers, serving as a training aid. Shoosmiths reports that early pilot users saw a 40% reduction in time spent on first-pass contract reviews, with no increase in error rates.

Critically, Project Apollo does not make decisions autonomously. It is designed as an advisory tool, with all outputs requiring human ratification. The firm has implemented a continuous feedback loop where lawyers can correct or refine AI suggestions, which are then used to retrain and improve the underlying models. This feedback is anonymized and processed in isolated Azure environments to maintain client confidentiality.

The Development Journey and Microsoft Partnership

The genesis of Project Apollo dates back to early 2025, when Shoosmiths’ innovation team identified contract review as a prime candidate for AI augmentation. The firm already had a data lake of anonymized, consented contracts—over two million documents—which provided a rich training ground. However, off-the-shelf AI tools lacked the customization and transparency the legal profession demands. That led to the decision to build a proprietary solution with Microsoft’s backing.

Microsoft provided not only technical resources but also strategic input through its Industry Solutions division. The collaboration included a three-month “AI design sprint” where lawyers, data scientists, and cloud architects defined the platform’s requirements. The project proceeded in phases: first, a proof of concept using a small subset of commercial contracts; then, a pilot within the real estate and corporate teams; and finally, firm-wide rollout.

One of the biggest challenges was fine-tuning the AI to understand the nuanced differences in contract language across various practice areas and jurisdictions. For example, an indemnification clause in an English law commercial contract carries different implications than one in a U.S. employment agreement. Shoosmiths’ legal knowledge engineers worked with subject matter experts to create practice-specific taxonomies and tagging systems, which were then used to train Azure-based classifiers.

Generative AI for Contract Review: Use Cases and Benefits

Project Apollo covers a wide range of contract types, from simple NDAs to complex M&A agreements. Key capabilities include:

  • Automated clause identification: The AI scans contracts and categorizes clauses into standardized types (e.g., force majeure, limitation of liability) with high accuracy.
  • Risk scoring: Each clause receives a risk score based on the firm’s playbook, indicating potential issues or deviations from standard language.
  • Comparison and benchmarking: Users can compare clauses against the firm’s repository of accepted clauses to see how a given draft aligns with market norms.
  • Drafting assistance: Lawyers can request alternative clause wording, which the AI generates along with explanations of the trade-offs (e.g., increased protection vs. commercial viability).
  • Summarisation: The platform can produce executive summaries of key terms, obligations, and deadlines, saving hours of manual review.

In a recent client matter involving a complex supply chain agreement, Project Apollo identified 47 risk clauses in under two minutes—a task that would have taken a senior associate over four hours. The tool also flagged a subtle inconsistency between two related clauses that had been missed during human review. These results highlight how generative AI augments, rather than replaces, legal expertise.

Client feedback has been positive. Early adopters appreciate the transparency—they are no longer presented with AI “black box” suggestions but can see the reasoning behind recommendations. This builds confidence in the technology and helps in-house legal teams justify decisions to their business stakeholders.

Challenges and Ethical Considerations

Despite its promise, deploying generative AI in law is fraught with challenges. Shoosmiths has been candid about the hurdles, including model hallucination, data privacy, and the risk of over-reliance on automation. Project Apollo’s governance framework addresses these head-on:

  • Hallucination mitigation: The platform uses retrieval-augmented generation (RAG) to ground responses in actual contract data, and all outputs are cross-referenced against the firm’s legal knowledge base before being shown to users.
  • Data privacy: Client data never leaves the firm’s Azure environment, and the AI models do not retain or learn from client data in real time. All training is done on consented, anonymized sets in isolated workspaces.
  • Over-reliance: The platform includes mandatory “human review gates” and requires lawyers to explicitly confirm they have reviewed and agree with each AI-generated finding before it can be incorporated into client work.

The ethical dimension extends to billing practices. Shoosmiths has committed to transparency with clients about when and how AI is used, and it does not charge standard hourly rates for work done by the AI. Instead, the firm is exploring value-based billing models that reflect the efficiency gains while ensuring fair compensation for human oversight.

Shoosmiths’ Culture of Innovation

Project Apollo is the latest in a series of tech-driven initiatives from Shoosmiths, a firm that has long punched above its weight in legal innovation. Its dedicated innovation function, “Shoosmiths Create,” focuses on process improvement, legal engineering, and technology adoption. The firm was an early adopter of Microsoft 365 copilot tools and has integrated AI into various back-office functions.

This cultural foundation made Project Apollo possible. Lawyers were involved from day one, contributing to the training data, testing prototypes, and shaping the user experience. The firm’s leadership, including CEO Simon Boss, has championed the project as a strategic priority, allocating resources and setting clear success metrics. The result is a tool that feels organic to the firm’s operations rather than a bolt-on experiment.

The platform’s name, Apollo, reflects the ambition: like the moon landing, it represents a giant leap for the firm, enabled by careful planning and cutting-edge technology. It also underscores the collaborative effort between Shoosmiths and Microsoft, mirroring the public-private partnership that characterized the original space program.

Industry Context and Future Implications

Shoosmiths is not alone in exploring AI for contract review. A growing number of law firms and legal tech vendors are deploying generative AI, but Project Apollo stands out for its bespoke nature and transparency. By building its own platform, Shoosmiths avoids generic AI solutions that may not align with legal best practices and can tailor the system to its specific client base.

The move also signals a shift in how law firms think about technology: not just as a cost-cutting tool but as a competitive differentiator. In an increasingly crowded legal market, the ability to offer faster, more insightful contract reviews could win clients—especially those frustrated by slow, manual processes.

Looking ahead, Shoosmiths plans to extend Project Apollo to other document-intensive tasks, such as due diligence, regulatory compliance checks, and litigation discovery. The platform’s architecture is designed to be modular, so new capabilities can be added as the technology matures. The firm is also exploring the potential of “agentic AI,” where multiple AI agents collaborate on complex tasks, although such applications would require even more robust governance.

For Microsoft, Project Apollo serves as a powerful case study for Azure’s role in the legal sector. The partnership demonstrates how Azure’s AI services, combined with responsible AI tools, can meet the most stringent professional requirements. It also opens the door for similar collaborations with other law firms and professional services firms, potentially creating a new ecosystem of industry-specific AI solutions on Azure.

Conclusion

Project Apollo is more than a technology deployment; it is a statement of intent from Shoosmiths. By marrying generative AI with rigorous transparency and governance, the firm has created a tool that not only boosts productivity but also upholds the highest standards of the legal profession. As AI continues to permeate white-collar work, Shoosmiths’ approach offers a blueprint for how professional services firms can harness the technology responsibly.

The launch on June 24, 2026, will likely be remembered as a milestone in legal AI. But its true test will be in everyday use: whether lawyers trust it, whether clients value it, and whether it delivers measurable improvements. If the early results hold, Project Apollo could redefine what it means to be a modern law firm.