How Nextdoor Built a Company-Wide AI Learning Loop with Glean

Nextdoor operates one of the world’s largest neighborhood social networks, connecting communities at a hyperlocal scale across the US and internationally. As the company’s teams scaled rapidly, knowledge fragmentation became a growing drag on execution: information was spread across Slack, Confluence, Google Drive, Coda, and other tools, and people were spending more time chasing answers than doing the work that mattered.

The root problem wasn’t a lack of information — it was the cost of finding the right information at the right moment with the right context. Projects moved faster than institutional knowledge could travel, and teams were onboarding and scaling without a reliable way to access what the organization already knew. Adding another point tool would have added friction rather than removed it.

Nextdoor connected Glean to the core systems employees already lived in, so answers and actions could happen in one place instead of across eight tabs. The rollout was anchored in four lightweight activation moves: a dedicated Slack support channel for open peer sharing, live office hours that began weekly and shifted to monthly as expertise localized, quick-win highlights in company bulletins, and purposeful gamification that recognized top Agent builders, Assistant power users, and Search adopters. Shweta Puri, Marketing Technology and AI Operations Lead at Nextdoor, described the approach as building a learning loop — making Work AI feel like how Nextdoor already works rather than a new system to learn.

Specific agents delivered results across functions. An HR open enrollment agent named Finn guided employees through benefit choices, reducing repeat People team tickets while improving employee confidence. A sales intelligence agent delivered tailored business signals and internal context to account executives before customer meetings, eliminating hours of manual research. A Trust & Safety agent assembled structured assessments from user history, policy links, and prior cases, enabling analysts to process more reviews with higher consistency. For engineering, an agent that answers natural-language questions against current code and technical documentation improved individual productivity by 2–3x compared to manual searching, with up to 2x time savings on specific tasks. A RevOps address enrichment agent reduced a 1–2 hour workflow across 50 accounts to approximately 5–6 minutes.

Nextdoor’s approach demonstrates that sustained AI adoption depends more on human-centered change management than on model capability. By embedding Glean directly into existing systems — Slack, Confluence, Coda, Google Drive — and building a cadence of peer learning rather than top-down mandates, the company created the conditions for Work AI to become habitual. The learning loop now generates its own momentum: early wins get shared, agents get remixed, and AI literacy compounds across the organization.