Adopting AI at Spiral
Capturing and sharing knowledge
One of the simplest but most impactful changes has been how we handle meetings. We now record meetings and generate AI-assisted summaries, which are automatically saved into the relevant section of our knowledge base. This means institutional knowledge is captured consistently, rather than sitting in someone's notebook or memory. Our team can then query this knowledge base using a dedicated agent, making it far easier to find decisions, context, and history.
We've also started generating project status reports, weekly and monthly, with AI assistance, freeing up time that was previously spent compiling updates manually.
Supporting software development
Our developers use AI in focused, practical ways and always with human review as the final step.
Code generation tools like Copilot help with routine tasks such as generating view models and comparing data models. Rather than writing boilerplate from scratch, developers get a useful starting point and retain full control over design decisions and implementation. We're also exploring AI-assisted unit test generation, which reduces repetitive work and helps developers think through edge cases they might otherwise miss.
On the learning side, we use AI to power daily development quizzes with progressive difficulty - a lightweight way for the team to sharpen core skills over time.
For database work, AI has helped generate batch SQL scripts, including execution plans and supporting documentation. These outputs are always reviewed carefully before use, particularly where production data or regulated environments are involved.
Strengthening quality and compliance
Quality is non-negotiable in clinical trial software. We maintain an extensive library of Standard Operating Procedures and have introduced a compliance agent that monitors our tools and flags issues that require attention. Combined with our automated testing suites, this gives us layered assurance without adding manual overhead.
Accelerating customer solutions
When a new project comes in, AI helps us move faster in the early stages, from evaluating protocol documents to support quoting and estimation, and producing prototypes to map out solution architecture before development begins. This means we can give research teams clearer, more accurate proposals sooner.
Getting the foundations right
Perhaps the most important lesson we've learned is this: AI is only as good as the information it has access to. That's why we're investing significant time in curating and structuring our knowledge base - making sure our agents have dependable, well-organised information to draw from. Without that foundation, even the best AI tools fall short.
“We'll continue to adopt AI where it genuinely adds value - carefully, practically, and with people at the centre.”
