News for Agent

Daily business news: funding, M&A, launches

Product Launches

Tursio Adds Per-Query Cost Controls and Mirrored Governance for AI Search

By ·

Key takeaways

  • Tursio added cost visibility and daily budgets to its AI search platform, with a typical query costing about three cents.
  • The mirrored governance feature reads existing access rules in systems such as SQL Server and Snowflake and enforces them in Tursio.
  • CEO Alekh Jindal will discuss the new features at VentureTech 2026, held November 16-18 in Fort Worth, Texas.

Tursio released cost controls and mirrored governance for its AI search platform, letting customers see the cost of each query and set daily spending budgets, with a typical query costing about three cents. The company announced the features on November 12, EIN Presswire reported.

How Tursio's Query Cost Controls Work

In the Tursio portal, customers can see the cost of each query as it runs, along with running averages. Admins can set a daily budget on the total cost of input and output tokens; once the budget is reached, Tursio stops processing new queries until the limit resets. Across Tursio deployments, a typical query costs about three cents, or about two cents when averaged across all databases. The company said estimates put the cost of a query through Fiserv's agentOS at about 20 cents, making Tursio roughly 10 times more cost-efficient.

Tursio keeps costs low by reusing context across queries instead of rebuilding it and by routing each question through a query plan that keeps model calls to a minimum. Tokens are billed as bare-metal LLM usage directly to the customer's existing subscription, such as Azure OpenAI, with no reselling. Co-founder and CEO Alekh Jindal tied the features to conversations at recent events, including Jack Henry Connect. "AI costs don't grow in a straight line; they compound quietly, query by query, until someone gets the bill," Jindal said. "Now our customers can see what every query costs and decide ahead of time how much they want to spend."

Mirrored Governance for Existing Access Rules

Tursio now reads the access rules customers already set in their source systems and enforces them, so users only see data they are authorized to see. Existing hashing, masking and redaction rules carry over as-is, and Tursio integrates with Microsoft Entra ID for identity and access management. The company built the capability at the request of a large enterprise customer with detailed access policies. Because Tursio runs entirely on-premises, the governance happens inside the customer's own security perimeter, and data never leaves their environment. "We pick up the rules they already have and enforce them, so the security they've built stays intact," Jindal said.

Tursio's Azure and Symitar Integrations

Tursio launched Azure VM Extension availability in February 2026, enabling deployment of Tursio Search directly within Microsoft Azure environments with native integration across Azure SQL, Azure Databricks, Microsoft Fabric, Azure AI Foundry and Entra, according to akta.pro company records. The same records list a native integration with the Symitar core banking platform for credit unions and a connection to Microsoft Teams via MCP. Tursio released v1 for the enterprise market in April 2025 with SaaS and on-premises deployment focused on regulated industries including BFSI and healthcare.

Corning Credit Union and VLDB Demonstrations

Corning Credit Union connected its core banking data to Tursio, enabling staff to query member data in natural language without tickets, EIN Presswire reported in July 2026. Tursio was accepted into the demonstration track of VLDB 2026, the 52nd International Conference on Very Large Data Bases, held August 31 through September 4 in Boston, Globalfintechseries reported.

VentureTech 2026 Session

Jindal will discuss the new features at VentureTech 2026, held November 16-18 in Fort Worth, Texas. Cost tracking, budgeting and mirrored governance features are now available to Tursio customers.

Companies in this story

Sources

Machine-readable: Markdown · JSON