Real systems, shipped for real teams — not prototypes or proof-of-concepts.
After every sales call, reps had to manually navigate 130+ fields across HubSpot — cross-referencing their notes to fill in the right information for each client. The process was tedious, error-prone, and consuming 45–60 minutes of a rep's time per call. CRM data was often incomplete or inconsistent as a result.
An AI agent that activates the moment a Fathom call recording is transcribed. It reads the transcript, identifies the relevant information from the conversation, maps it to the correct HubSpot fields, and updates them automatically — with no rep involvement required. Built iteratively with direct feedback from the sales team to ensure the outputs matched their exact workflow.
Sales reps no longer touch HubSpot after a call. Fields are populated accurately within minutes of the recording finishing, giving the team cleaner CRM data and more time focused on selling rather than data entry.
Every time a new client was created, the finance team expected sales to add them to Invoiced with a complete billing address. Reps frequently didn't have the address on hand or simply forgot — creating ongoing friction between the sales and finance teams and leaving billing records incomplete.
An automated n8n workflow that runs every morning without any human trigger. It checks BigQuery for new Invoiced clients missing a billing address. It first looks in HubSpot — if the address exists there, it syncs it to Invoiced automatically. If not, it queries the OpenAI API to find and verify the client's address, then writes the verified result to both Invoiced and HubSpot, keeping both systems in sync.
The manual back-and-forth between sales and finance was eliminated entirely. New clients are reconciled every morning automatically — the finance team always has complete billing records without chasing anyone, and both CRM and billing systems stay in sync without any human intervention.