Notion People Team Ships Smilers Agent, Saves 30% of Environment Team Time.

Notion's People team has been quietly building AI agents internally for nearly a year, starting with Smilers, a custom tool that fields questions about the company's physical workspaces. The agent operates in Notion's ask-Environment Slack channel and responds to questions about office layouts, amenities, and workspace details at what the team calls "the speed of thought."

According to a recent LinkedIn post from Katy Shields, Notion's VP of People, Smilers has reduced busywork for the Environment team by roughly 30%. That time savings allowed the team to focus on higher-value work, like designing employee experiences and planning events.

Smilers wasn't a one-off experiment. After proving the concept, Notion's People team built additional agents, including Gatsby, which handles events planning, and another tool focused on seat assignments. The post highlights a recent Product Design Team event that brought together crews from Notion and Anthropic, organized with help from these internal agents.

The agents appear to be custom-built for Notion's internal workflows rather than products intended for external customers. Shields noted that the team has made a Smilers package available for other teams interested in reducing repetitive tasks, though details on how it's shared or deployed weren't included in the post.

The move to build internal agents aligns with Notion's broader push into AI tooling. The company recently introduced external agents that integrate with tools like Claude and Cursor, and launched a Developer Platform that lets teams sync external data sources into Notion databases and build custom tools for their agents.

Internal agents like Smilers offer a different value proposition. They don't require third-party integrations or public-facing features. Instead, they solve specific, high-frequency problems for internal teams, automating the kind of repetitive questions that pile up in Slack channels and pull employees away from strategic work.

For the Environment team, that meant answering the same workspace questions multiple times a day. Smilers handles those inquiries automatically, freeing the team to focus on projects like event planning and workspace design.

Notion's decision to share details about Smilers suggests the company sees internal agent development as a model worth promoting. The post frames the project as both a time-saver and a morale booster, emphasizing that the Environment team can now "lean into the work that lights them up."

It also signals that Notion is treating AI agents as tools for any team, not just engineering or product. The People team built Smilers without waiting for a formal AI product launch or external vendor. They identified a problem, prototyped a solution, and iterated on it.

That approach mirrors how Notion has positioned its Developer Platform: as a set of building blocks that let teams extend the product to fit their workflows. Whether those workflows involve syncing support tickets from Zendesk or answering office questions in Slack, the underlying idea is the same. Teams should be able to build the tools they need without starting from scratch.

Notion hasn't announced plans to release Smilers or similar agents as packaged products, but the company's emphasis on making the Smilers package available internally suggests it's testing how these tools might scale across organizations. If other companies adopt similar patterns, internal agent development could become a standard part of how teams use platforms like Notion, Slack, and Airtable.