Vectara launches Tool Validator for agentic AI oversight

Vectara launched the Tool Validator, a governance component for enterprise agentic Artificial Intelligence (AI) that corrected agents' resource planning errors before those errors affected workflows.

A recent MIT report and other sources highlighted that many enterprise agentic systems fell short of expected benefits because workflow breakdowns, inadequate output quality and planning errors compounded into reduced reliability, increased costs, and compromised security, undermining return on investment.

The Tool Validator reviews the tools an agent proposes to use as part of a workflow, flags erroneous or irrelevant tool calls before execution, suggests adding missing relevant tools or omitting unnecessary ones, and logs errors and adjustments to ensure traceability; Vectara said the capability built on its work with Retrieval Augmented Generation (RAG) systems.

The Tool Validator was the second of Vectara's Guardian Agents and operated alongside the Hallucination Corrector to enforce governance and policies in real time across an organization's agentic systems and workflows, guarding against errors and hallucinations that emerged when agents planned workflows drawing on resources outside RAG pipelines such as web searches and external backend system queries.

“This reduces downstream errors and eliminates other major factors that have contributed to the 95% failure rate for enterprise AI deployments cited in the recent MIT report,” Vectara co-founder and CEO Adaptive Mesh Refinement (AMR) Awadallah said.

“Tool calls have real-world impacts: they might initiate a payment, update a customer record, or take action in a production system - not just generate a response,” Vectara Chief Product Officer Eva Nahari said. Vectara said the Tool Validator gives enterprises increased confidence to move agentic workflows into production by adjusting agents' mistakes before high-visibility detrimental outcomes while ensuring visibility and traceability.