Turning One-Off Agents into Scalable Workflows

Most teams get their first agent working in an afternoon. Then a rep runs it manually forty times, gets bored, and stops. The prompt was never the hard part. What's missing around it is: something to start it without a person, something to decide what happens next, and somewhere for the output to land. That's a Play. The agent is one part of it.

On Thursday, August 27, Sabrina Hu and Katie Aprile build three of them live in Nooks:

• Speed to lead. A demo request lands in Salesforce. That's the trigger. From there, a prospecting agent finds the other contacts at the account, an email agent drafts, and the last node assigns the rep and adds everyone to a sequence. Minutes, not Monday. This is one customers ask for most.
• Closed lost. Every rep knows closed lost is the lowest-hanging fruit in the CRM. Nobody works it, because the person who lost the deal left eight months ago and took the context with them. Here the trigger is an opportunity field changing, so the Play fires six months out from the account's renewal date, pulls the current contacts, and sequences them while the timing is still good.
• Weekly account prioritization. The agent Katie runs every Sunday night that ranks her book and names the prospect to open with at each account. In a Play, a schedule fires it, a research agent does the ranking, and a Slack alert puts the shortlist in front of her on Monday.

Plus: building a Play from a prompt inside AI Assistant, which shipped last week.

Sabrina Hu is a Product Manager at Nooks and owns Plays. She works with the customers pushing hardest on AI to deploy the complex workflows the rest of the market hasn't gotten to yet.

Katie Aprile is an upsell AE at Nooks. She spends her days talking to revenue teams about where AI is genuinely moving their numbers and where it isn't. She'll also demo the Plays she runs on her own book.

Live, in the product. No slides. 30 minutes + Q&A.