Why generic AI stalls and how governed agents like Cal move enterprises forward
Introduction
On 14 October, SIRV joined CSSC members at TWI, Cambridgeshire, to discuss AI in Risk Management: From ChatGPT to Super Intelligence. While the event was held under Chatham House rules, one theme from our session deserves to be shared, why so many enterprise AI pilots stall, and what governed AI agents like Cal do differently.
1. Why Generic LLMs Stall
Governance gaps
LLMs can’t show where their answers come from. Without source citations or provenance, enterprises can’t trust or audit outcomes.
Reliability risks
Without retrieval from approved data, even minor hallucinations can undermine confidence.
Integration barriers
Generic LLMs operate outside enterprise systems and compliance frameworks. They don’t align with access controls, reporting standards, or audit trails.
2. The Governed Alternative
Governed AI agents like Cal are built for enterprise realities:
-
RAG (Retrieval-Augmented Generation): Answers grounded in your verified data.
-
Policies: Responses filtered and constrained by approved rules.
-
Logs: Every interaction cited, timestamped, and reviewable.
Together, these create what we call governed speed, the ability to act faster without losing control.
3. From Experiment to Adoption
The next stage isn’t “bigger models,” but better governed ones. By combining retrieval, rules, and records, Cal allows organisations to automate reports, briefings, and risk monitoring; all with accountability built in. To experience this in practice, sign-up to our governed AI Sprint.
Author bio: Andrew Tollinton
Andrew Tollinton is Co-Founder of SIRV, the UK’s enterprise resilience platform. A leader in risk management technology, he chairs the Institute of Strategic Risk Management’s AI in Risk Management group and regularly speaks on AI and resilience at global conferences. A London Business School alumnus, Andrew brings 20+ years’ experience at the intersection of technology, compliance and security.
"SIRV helped us move beyond basic reporting into a system that actively supports decision-making". Les O'Gorman, Director of Facilities, UCB - Pharma and Life Sciences
