Single Rulebook and Sigma AI have partnered to shorten the time between an exchange changing its rules and the affected trading business responding. Exchanges can issue circulars that alter trading conditions each day, leaving firms to identify affected desks, assess the operational consequences and document decisions across disconnected teams. The firms aim to deliver source-grounded analysis to the responsible business function within seconds, with the evidence needed to defend each decision.
The partnership, announced on 15 July, combines Single Rulebook’s structured regulatory intelligence with Sigma’s artificial intelligence (AI) research and analytics. Single Rulebook, part of Kaizen RegTech Group, brings exchange rulebooks, circulars and regulatory updates into one environment. Its RegPulse engine captures and classifies changes. The wider platform assigns ownership, tracks implementation and retains an audit record.Sigma provides analytical technology developed for financial-market research, execution, risk and surveillance. The combined service is intended for compliance, trading and operations teams.
Single Rulebook CEO Chris Dingley described how relationship began with a meeting at the IDX conference followed by several months of discussions on how Single Rulebook’s regulatory content and workflow could work with Sigma’s analytical capabilities.
Dingley positioned the target use-case as sitting below the major regulatory initiatives managed through central compliance programmes. Exchanges and other trading venues issue rulebook amendments and circulars that can change the terms under which firms trade. These changes may demand a rapid operational response while remaining subject to the governance and evidence standards applied to regulatory change.Firms must determine which notices matter, identify the affected desks and establish whether systems, controls or operating processes need to change. Fragmented ownership and manual document reviews can delay that assessment.
Single Rulebook provides the structured content, regulatory context, source references and workflow. Sigma divides questions into more granular analytical tasks and presents the results to business users. Together, the firms intend to connect a change in a rulebook or circular with the employees responsible for assessing and implementing it. “This is landing right here in the front office,” Dingley said.
The front-office focus extends the use case beyond regulatory research. A venue change may affect a trading desk or technology control before it enters a firm-wide compliance programme.
Testing the Answer
An AI-generated answer has little value if a regulated firm cannot establish how the system produced it. “Most people talk about Gen AI being probably right. And we don’t deal with probably right in a regulatory world,” Sigma AI founder and CEO Andy Simpson said.
Simpson said the combined architecture goes beyond placing regulatory documents in a large language model. It uses an adversarial AI network containing a group of “AI judges” to check the output. He described the objective as bringing “a deterministic artificial intelligence capability to a world that absolutely demands precision”.
Dingley said the architecture records the system state, the user request, the tools applied and citations to the underlying text. It also links the response to its original online source. The audit trail follows the query through Sigma’s analysis to the response returned to the user. “Every step of that is available on a full audit trail,” he said.
Giving Specialists Their Time Back
The business case also rests on cutting the document work that consumes regulatory specialists’ time. Simpson recalled mapping the effects of the first Markets in Financial Instruments Directive (MiFID) across a boardroom wall before its implementation in 2007. Firms still compare document versions, interpret amendments and trace changes through connected rules and processes.
“This solution allows people in that profession to focus on their day jobs, on their real job of understanding strategic intent of policy and how the business responds to it,” Simpson said.
Keeping the process on a controlled platform could also reduce the risks created when employees build regulatory processes around personal prompts and configurations in general-purpose AI tools. Such arrangements can lose support when an employee leaves or a firm changes model provider. The partnership is intended to retain the analytical method, sources and decision history within an institutional process.
Dingley said the solution is currently in beta whilst discussions with banks is uncovering additional use cases. The need for faster analysis will grow as markets extend their operating hours and trading passes between Asia, Europe and North America with fewer pauses. “The RegTech world has to be able to respond to that change in cadence, the change in frequency and the change in expectation,” he says.
Subscribe to our newsletter
