I built this to solve a problem we kept hitting: deals would go quiet, and by the time we noticed, the context for a good follow-up was scattered across HubSpot, Slack conversations, and old call transcripts. The agent runs daily and:

* Pulls stale deals from HubSpot (configurable pipeline stages)

* Researches each prospect by aggregating context from multiple sources: HubSpot contact/deal/account data, internal Slack discussions about the prospect, Fireflies call transcripts, and recent company news via web search

* Uses Claude to generate personalized follow-up emails that reference specific pain points or initiatives mentioned in past conversations

* Delivers a daily HTML digest for review before sending

It’s essentially giving your sales team a research assistant that remembers every interaction and can craft contextual outreach at scale.

Tech stack: Python, HubSpot API, Slack MCP, Fireflies API, Claude API. Setup takes about 10-15 minutes if you have the accounts already connected.

GitHub: https://github.com/sneurgaonkar/sales-followup-agent

I’m actively working on this and would love to hear what features would make it more useful for your workflow. What other data sources would be valuable? Different notification methods? Integration with other CRMs?

Happy to answer questions about the implementation or help anyone get it running.