Stonly introduces Knowledge Agents, an AI capability that continuously monitors sources, detects knowledge gaps and changes, and drafts precise updates for customer service knowledge bases, ensuring accuracy for agents and AI.
Quick Intel
Stonly, the agentic AI and knowledge platform for customer service, today announced the launch of Knowledge Agents, a new AI capability designed to help support organizations keep the knowledge used by customers, agents, and AI 100% accurate and up to date.
Knowledge Agents go beyond basic generation and suggestion capabilities to address the harder problem of continuously monitoring source material and live support signals, identifying meaningful knowledge gaps, inconsistencies, and changes, and drafting precise updates to structured knowledge for easy human review and approval.
As companies scale AI in customer service, knowledge constantly changes due to policies, product updates, ticket-handling practices, compliance documents, and internal feedback. In most organizations, knowledge goes stale, gaps go undetected, and conflicts accumulate faster than teams can fix them. This creates a weak foundation for both human support and AI.
How Knowledge Agents Work
Teams connect their sources of truth, including resolved tickets, search and AI queries, SharePoint, Confluence, Google Drive, websites, PDFs, and other systems. Knowledge Agents continuously monitor these sources. When something changes, they trace the impact across support knowledge bases and help centers, identify every guide, workflow, and article where the change matters, and draft precise edits such as new branches, steps, insertions, or replacements.
All suggested updates appear in a dashboard where teams can review, adjust, approve, and publish changes.
Knowledge Agents also continuously audit content for broken links, conflicts, duplicates, and inconsistencies, providing a configurable Knowledge Health Score and detailed suggested fixes.
Additionally, teams can use prompt-based instructions for knowledge operations. Knowledge Agents can execute complex projects, such as updating specific references across the entire knowledge base, completing tasks that normally take hours in moments.
Alexis Fogel, Founder and CEO of Stonly, said: "Knowledge teams always tell us they are 'small but mighty.' They do not have a content generation problem, they have a change-management problem. Critical information changes constantly across dozens of places, with different owners, formats, and purposes. Knowledge Agents help teams keep their support knowledge current and accurate so both agents and AI can operate from what is actually true."
Why Accurate Knowledge Matters More Than Ever
When knowledge falls behind, agents take longer to find answers, escalate unnecessary issues, and risk mistakes, while customers experience frustration in self-service, leading to lower satisfaction and higher contact volume.
With the rise of AI in customer service, accurate structured knowledge has become mission-critical. AI operates strictly from the knowledge it is given and cannot independently detect inaccuracies. Inaccurate knowledge causes AI to scale errors across every interaction.
"At the same time that AI made accurate, structured knowledge essential," Fogel explained, "agentic AI enabled Stonly to develop advanced Knowledge Agents that automate the most challenging parts of knowledge management that were not possible before." The result is a foundation that keeps pace with the business, so every agent, every self-service experience, and every AI interaction operates from knowledge that teams can trust.
About Stonly, Inc.
Stonly is an agentic AI and knowledge platform for customer service. It helps support teams create, govern, and deliver accurate, structured knowledge for customers, agents and AI, making service reliable and scalable. Backed by top venture capital firms Northzone and Accel, Stonly serves complex support environments at companies including Carta, AMC, King, SimpliSafe, Siemens, and Estee Lauder. Stonly is HIPAA, GDPR, and SOC 2 Type 2 compliant.