September 29th, 2026
New AI lab brings continuous, governed data remediation into the customer’s own environment, helping enterprises build confidence in their data and realize more value from AI investments.
TQStarling today announced the public launch of Starling Labs by TQStarling, its applied AI research and product arm, and the availability of Starling Sovereign, a platform for continuous enterprise data remediation.
Starling Sovereign addresses a persistent challenge for enterprise AI: the data underpinning automation and decision-making often requires people to manually check, reconcile, and correct records before they can confidently act on them. Periodic cleanup can improve a dataset, but maintaining its quality requires an ongoing process as records and systems change.
Starling Sovereign brings that process inside the customer’s own environment. It scores data health, identifies defects, and applies corrections under confidence and risk controls, while retaining evidence behind the decisions.
“Enterprises are investing in increasingly capable AI, but people are still doing the work of checking and correcting the data those systems depend on,” said Joe Peters, President of Starling Labs by TQStarling. “We created Starling Labs to address that gap. With Starling Sovereign, we’re helping organizations make data remediation a continuous capability, so they can act with greater confidence and get more value from their AI investments.”
Starling Sovereign extends data cleaning beyond correcting a value. Its patent-pending provenance ledger retains an auditable record of what changed, why it changed, the source supporting the correction, and who or what approved it. Decisions to leave records unchanged are recorded as well.
Routine corrections can be applied automatically when they meet confidence and risk criteria. Higher-risk changes, including every merge, go to a person for review. When no verified replacement value is available, the record is escalated with a rationale rather than assigned an invented answer.
“A correction needs a defensible basis, and confidence alone is not permission to change a record,” said Hugo Delgado, Chief Data Scientist at Starling Labs by TQStarling. “We check proposed corrections against the customer’s source-of-truth baseline and govern them through confidence and risk criteria. When a value cannot be verified, the system escalates it. Preserving that uncertainty—and the evidence behind each decision—is fundamental to trustworthy remediation.”
Starling Sovereign is now publicly available. To explore the platform, review model cards, or discuss a Value Audit, visit tqstarling.ai.
Starling Labs by TQStarling is TQStarling’s applied AI research and product arm. It develops specialized small language models and governed AI systems for enterprise use. Its Starling Sovereign platform delivers continuous data remediation inside the customer’s environment, with evidence retained behind every decision.
