Notion — searchable YouTube transcripts & AI Q&A

How Candidly makes coding agents multiplayer in Notion

Candidly's chief product officer reports that integrating AI coding agents into their workflows increased value delivered to customers by 73% per person. He demonstrates how Notion serves as a shared surface where humans and agents collaborate: he can tag a coding agent in a ticket or move it to a ready state, and agents can manage the board overnight, allowing him to review completed work in the morning. He also notes that the same board can tag human engineers, highlighting the flexibility of the system.

First Block: Interview with Jamie Hall & Steve Hind, Co-Founders of Lorikeet

The founders argue that startups should choose early design partners who are motivated by unmet needs, not those who will give glowing feedback, to avoid building undifferentiated products. They emphasize aligning pricing with customer success, such as only charging when the AI resolves a ticket, to ensure incentives are aligned. They also caution against building cool technology for its own sake, advising founders to focus on painful user problems and to be disciplined about where they spend their time.

Notion’s Most-Used Agent: A Personal Data Scientist for Every Notion Employee

Notion's most-used agent, Data Scout, lets employees query Snowflake data in plain language, reducing data team requests by 80% in two months. The agent interprets raw numbers using documented context in Notion, such as adoption definitions and customer-specific Slack conversations, to generate actionable insights like meeting prep pages. This self-service approach shifts data scientists to higher-leverage work, though the speaker notes the workflow was built to solve Notion's own problem.

How to Build a Self-Improving Answer Bot with AI Agents

The speaker argues that AI agents can transform knowledge management by not only answering questions but also identifying gaps and proposing updates to a company's wiki. He demonstrates a system where a Slack-based agent answers sales questions, a reporter agent drafts new articles from Slack discussions for human approval, and a verifier agent flags outdated content. The key takeaway is that this human-in-the-loop approach keeps the knowledge base accurate and self-improving, while freeing humans for higher-level work.

Designing Employee Onboarding in the AI Era

Notion's people team argues that onboarding can be automated to scale delightful experiences, with the key challenge being exceptions like off-cycle starts or name changes. They built a centralized database that updates downstream systems automatically, such as rescheduling calendar invites when a start date changes, and an agent that generates personalized task lists and schedules one-on-ones. This frees up HR staff to focus on the human experience, though they note that the system still requires human oversight for some exceptions.