Confluence Technologies has launched Confluence POINT, starting with AI-enabled document validation and a conversational interface for its Revolution analytics platform.
Confluence says both capabilities are available immediately. POINT Validation checks financial and regulatory documents across multiple formats, including unstructured data. Its automated checklist completes validation in minutes, flags errors for review and records an audit trail for each finding, according to the company.POINT also adds a conversational interface to Revolution, Confluence’s multi-asset performance, attribution and risk solution. Users can ask questions and initiate processes in plain English through Microsoft Excel, without exporting data or moving between systems.
The company plans to extend POINT across its regulatory, analytics and investor communications products. Mark Evans, founder and chief executive officer, described the launch as a long-term commitment and said Confluence would add further AI capabilities across the suite in the coming months.
Connecting The Confluence Portfolio
To understand how POINT will develop across Confluence’s wider product suite, RegTech Insight spoke with Chief Product Officer Kathleen Keenan. She explained how the launch fits within a broader product strategy that includes rationalising a portfolio once spanning 45 products and moving selected products onto a unified data layer.
Regulatory and financial reporting products are among those being brought together first with plans to add analytics, allowing completed calculations and results to feed into regulatory reporting workflows. “Being able to leverage the calculations or the end results from our analytics platform into a regulatory reporting platform is really something significant for the industry,” Keenan said.Where products have not yet moved onto the unified data layer, Confluence is piping data from existing platforms into POINT.
Moving validation towards exceptions
POINT Validation applies AI to the checks performed after financial and regulatory reports are produced. A report can require hundreds of checks, followed by comparisons with other documents containing the same information. Keenan notes that while some firms have developed internal tools to automate parts of the process, much of the workflow remains manual. “Typically, the validation includes 200, 300 manual checks, in addition they need to reconcile against multiple other documents that include the same data,” she says.
Following the first review pass, POINT directs the user to any exceptions. The reviewer can inspect the source data, trace the calculation and locate corresponding figures in another report before deciding the result is correct.
The system records the actions taken, allowing control teams and auditors to reconstruct the process if an error passes through. Keenan said users could see “every step that was taken on that document”.
Human judgement remains part of the process. “Our AI tools are not intended to take the human out of the loop,” she says. They facilitate human judgment and make it more focused.”
Controlling Prompts for Repeatable Results
Confluence is also structuring how users prompt its conversational interfaces. Keenan said financial firms were unlikely to rely on prompting alone because open-ended questions can produce inconsistent answers. Processes across front, middle and back office must produce repeatable results.
Natural-language interfaces will include preset prompts for defined tasks, particularly in analytics. A user might request a portfolio view or initiate a calculation in plain English, while the system follows a repeatable process beneath the interface. Data lineage, auditability, repeatability are key design constraints.
Keenan said POINT Validation is currently in proof-of-value testing with around 20 customers. By the end of 2026, Confluence expects POINT to work across its analytics capabilities, providing a portfolio-level view, initiating calculations and making reports available through a common interface.
During 2027, the company plans to bring regulatory products into POINT. Keenan described the objective as a “move to being exception-based work versus doing everything”. POINT would initiate report production and surface exceptions, while users investigate and resolve them in the existing specialist applications.
Subscribe to our newsletter
