Docebo ships AI agents and knowledge search for corporate learning

Docebo has launched two AI products for its Learn platform: Docebo Knowledge, a unified search layer over training content and business systems, and Docebo Agents, a set of AI agents that automate learning-related work. Both are available to Docebo Learn customers now.

The launch matters because it targets a gap Docebo itself has documented. Companies are spending heavily on AI tools while still building the foundations those tools depend on — skills data, trusted knowledge and workflow context. Docebo's 2027 Skills Intelligence Report found many businesses adopting AI faster than they're putting those foundations in place. Docebo is betting that learning platforms, which already hold skills and training records, are the natural place to close that gap.

By Sophie Lindqvist

What does Docebo Knowledge actually do?

Docebo Knowledge gives employees a single search interface across training material and connected business systems. It indexes content from Docebo's own platform — including SCORM and xAPI packages — plus approved company information held in SharePoint, Confluence, Google Drive and Slack. Searches return answers, procedures, videos and source links, all grounded in approved material.

The tool isn't just a search box. A user can move from an answer straight to a related course, certification, practice exercise or workflow. That turns a lookup into a learning action without leaving the interface.

Permissions carry over from each connected source, so users only see information they're authorised to access. In plain terms: the knowledge layer respects the walls each company has already built around its data.

What are the four pre-built agents?

Alongside the knowledge tool, Docebo introduced AI agents that automate tasks linked to learning and upskilling. The agents draw on a learner's history, identified skills gaps and organisational taxonomies, and can act across Docebo and connected systems. Four pre-built agents shipped at launch:

• Course creation agent — builds courses from source material
• Tutor agent — supports learners inside a course or learning plan
• Reinforcement agent — runs post-course practice sessions
• Marketplace agent — sources content from third-party providers

For customers who want something more specific, Docebo added a no-code builder. Organisations can set an agent's goal, instructions and permitted actions without writing software. That lowers the barrier for HR and learning teams that don't have developers on staff.

What did the pilots show?

More than 50 customers took part in pilots before launch, spanning transportation, retail, distribution and food service. Docebo said the tests produced early signs of time savings in administration and support. The company shared four concrete results:

• An international electronics distributor used the Marketplace Agent to cut manual content curation by about five hours a month per administrator.
• A global restaurant operator used the Reinforcement Agent to save store managers about 10 hours a month on training follow-up.
• The Tutor Agent at that restaurant operator handled 25% of frontline training questions without an administrator.
• A retailer recruiting 2,500 new employees a year built its own onboarding agent during the pilot and expects it to save sales managers one to two hours per new hire.

These are pilot figures, not audited results, and Docebo presents them as early signs rather than proof. But the pattern is consistent: the agents absorb repetitive administrative work — curation, follow-up, answering routine questions — that has kept learning teams busy without moving the business forward.

How much control do administrators keep?

Docebo said it built approval and oversight functions directly into the agent system. Administrators decide which agents operate, who can use them, and which actions require human sign-off before completion. Every action is logged and follows existing user permissions.

Customers can also track AI credit usage to monitor costs as deployment expands. That matters for enterprises worried about runaway spend on AI features billed by consumption — a common complaint as agent tools spread across workplaces.

Why does Model Context Protocol support matter?

The launch also extends Docebo's push to make its learning tools available outside its own interface. Through Model Context Protocol (MCP), customers can connect Docebo to AI assistants and workplace tools including Microsoft Copilot, ChatGPT and Claude.

The practical effect: employees can search for content, check their progress and enrol in courses without opening the Docebo platform. Administrators can handle reporting and enrolments through the same connected tools.

MCP is emerging as the standard plumbing for this kind of integration, and Docebo's support signals where enterprise software is heading — the learning platform becomes a service other AI tools call, rather than a destination users must visit.

What did Docebo's CEO say?

Alessio Artuffo, President & Chief Executive Officer at Docebo, framed the launch around a structural shift in how organisations operate. The changes, he said, reflect how businesses are beginning to manage both staff and software agents in day-to-day operations.

"Every organization is about to run two workforces: its people and the agents working alongside them," said Artuffo. "Both need the same things to perform well: trusted knowledge and an accurate picture of skills. That's the foundation Docebo is now built on. Agents run through the platform, taking on the manual work that has held learning teams back for years. Every employee gets development shaped around their own skills and goals. Less time on administration. Less budget spent on work that doesn't move the business. More capability where it counts."

The quote reads as a positioning statement for the whole category: if every company will run human and AI workforces side by side, the vendor that holds trusted knowledge and skills data for both has a strategic asset.

What's the competitive stake?

Docebo competes in the corporate learning management market, a space where AI features have become table stakes over the past two years. The differentiation here isn't a chatbot bolted onto a course catalogue. It's the combination of a permissions-aware knowledge layer, agents that can act across systems, no-code customisation and open protocol support.

The pilot numbers give Docebo early evidence to take to enterprise buyers who increasingly demand measurable time savings, not just feature checklists. If the five-to-ten-hour monthly savings hold at scale across the 50-plus pilot customers and beyond, the ROI case writes itself.

The open question is adoption pace. Docebo's own Skills Intelligence Report suggests many businesses lack the data foundations these agents rely on, and the company hasn't disclosed pricing for AI credit usage at scale. Watch whether the no-code agent builder turns learning teams into agent builders — that, more than the four pre-built agents, would signal whether Docebo's two-workforces thesis catches on.