kama.ai launches SourceLock to keep AI answers inside approved sources

kama.ai introduced REACTT SourceLock on November 12, 2026, in Toronto, giving organizations a way to make AI agents answer only from their own website or approved documents. The company says the feature is designed to reduce hallucinations, improve auditability, and support brand-safe AI governance.

Why it matters: - Organizations want AI answers that stay inside approved information, especially for customer-facing use cases. - REACTT SourceLock is designed to reduce hallucinations by limiting responses to content the organization chooses and controls. - The feature also aims to improve accountability with traceable answers and a full audit trail. - The launch matters for teams that need brand-safe AI without opening access to the broader internet.

What happened: - kama.ai launched REACTT SourceLock, a set of features for its REACTT Agent platform. - The launch took place on November 12, 2026, in Toronto. - The feature points an AI agent at an organization’s own website or Trusted Collections, which are source documents the customer authorizes. - kama.ai says AI agent setup and launch can be completed in a few hours.

The details: - REACTT SourceLock restricts the AI agent to the sources the organization selects. - Each answer can be traced back to the chosen content through a full audit trail. - The system is meant to support brand safety, consistency, and accountability. - Customers can choose the large language model and provider they want. - REACTT SourceLock keeps the model within approved rails. - The feature locks out action mode and training across models. - kama.ai says this prevents source information from being leaked and blocks rogue actions. - REACTT SourceLock integrates with Trusted Collections and customer website content as a core part of the REACTT Agent platform. - AI agents can be up and running within an hour when using website content or authorized documents through Trusted Collections. - The platform is aimed at manufacturers, municipalities, colleges and universities, professional services firms, non-profits, and Indigenous communities. - kama.ai describes itself as a Canadian, Indigenous-owned Responsible AI solution provider.

Between the lines: - The launch reflects a push to move enterprise AI away from open-ended internet answers and toward governed, source-bound deployments. - The emphasis on human-in-the-loop management suggests customers may want tighter oversight as AI systems are deployed more broadly. - The pitch also positions governance as a product feature, not a separate compliance layer. - Brian Ritchie, CEO and Founder of kama.ai, said the company is building AI that gets answers right and improves with human-in-the-loop management and AI governance.

What's next: - kama.ai is directing interested customers to more information or direct contact for a discussion. - The company is also offering media contact through Charles Dimov, CMO, at Charles.Dimov@kama.ai and +1 (647) 702-1494. - kama.ai says the REACTT Agent platform will continue to center on governed sources and customer-controlled AI deployment.

The bottom line: - REACTT SourceLock gives organizations a way to keep AI answers anchored to approved content while preserving control over the model, the source set, and the output trail.