ActiveViam partners Databricks on real-time risk analytics

ActiveViam has partnered with Databricks to offer real-time risk analytics on the Databricks Data + AI Platform, targeting financial institutions using AI agents for risk work.

The integration links ActiveViam's Atoti Intelligence architecture with Databricks, allowing banks and other financial firms to run risk calculations inside their own environments. Institutions can use approved large language models while keeping data, governance controls and intellectual property within their existing Databricks set-up.

AI agents can call deterministic risk tools directly, rather than generate SQL queries or code against raw tables. These tools include value at risk, expected shortfall, Fundamental Review of the Trading Book, xVA, and liquidity coverage and net stable funding ratio calculations.

The focus is on risk functions that require repeatable calculations and audit trails. Each calculation can be traced back to the underlying trade and remains subject to Databricks governance, lineage and access controls.

The move extends an existing relationship between the two technology stacks. ActiveViam already supports DirectQuery for Databricks, giving customers a live connection to governed Databricks data without moving it into a separate environment.

Together, these capabilities are designed to help financial institutions build agentic workflows for capital-markets risk on a single data foundation. That is significant for banks trying to move beyond pilot AI projects in areas where controls, model oversight and data handling face close scrutiny.

For existing ActiveViam users looking to consolidate data and analytics workloads on Databricks, measures, hierarchies and workflows can carry over. That means institutions do not have to rebuild risk frameworks from scratch when shifting to the Databricks platform.

For Databricks customers that do not already use ActiveViam, the integration provides a way to add specialist capital-markets risk analytics to a lakehouse environment without building those systems internally. It also avoids moving data off-platform, a sensitive issue for institutions managing market, treasury and liquidity risk.

Risk workflow

The announcement reflects a broader push across financial services to combine governed enterprise data with AI tools in ways that meet internal control standards. In risk and finance functions, that typically means using AI systems to orchestrate or query approved analytics rather than letting models produce calculations independently.

ActiveViam positions Atoti as a semantic and analytics layer for financial institutions handling complex risk and treasury data. Its tools support real-time and historical analysis across trading, risk and treasury management, with pre-packaged support for regulatory and internal risk metrics.

ActiveViam has recently received industry recognition, including inclusion in the RiskTech 100 and category leadership in the Chartis Credit Risk Report 2026 and the 2026 Energy50 rankings. The accolades reinforce its focus on specialist financial risk software at a time when large data platform providers are seeking closer ties with sector-specific applications.

Databricks has been expanding its reach in regulated industries as customers look to centralise data, analytics and AI development on fewer platforms. For capital-markets firms, one challenge is connecting that platform strategy with long-established risk engines and regulatory reporting processes that depend on consistency and approval workflows.

In that context, the partnership is intended to keep the risk logic separate from the language model layer. Instead of asking an AI tool to derive answers directly from source data, institutions can direct the agent to call a defined risk service and return a validated output.

Alan Lee, Chief Product Officer at ActiveViam, said: "Databricks gives banks one governed home for their data, while Atoti turns it into risk intelligence they can trust and sign off on, without copying data out. The result is the same number every time, traceable to the trade and precise enough to sign off. Everything runs inside the firm's own environment and under its existing Databricks governance, so its data and IP never leave. We're proud of this integration with the Databricks Data + AI Platform and look forward to seeing our clients take full advantage of this new capability."