agentic GTM engine

Manual friction largely removed and increased capabilities compared to the existing outbound infrastructure.

The current outbound infrastructure I run still requires human orchestration between the automated workflows. I identified that the majority of these handoffs could be automated, and that the current infrastructure would benefit from more agentic execution.

Currently a work-in-progress, with the majority of features shipped and at prod-level. The system now runs nearly end-to-end, beginning with the user manually triggering it. The user then places pre-flight inputs such as ICP, segment, offer, etc. With these inputs, the system begins running lead-sourcing through to message-drafting, without any human intervention. Editing and review remain manual, before the approved batch is queued for "jitter-sending", where approved emails are sent at random intervals. The system is a mix of local LLMs, low-end GPT models, and Claude Code-run scripts, so as to maintain cost effectiveness.

A further 6-8 hours per week saved by reducing human-in-the-loop to only critical tasks. Context-switching significantly reduced, allowing me to maintain deep focus on tasks at hand while the system runs in the background.

Context switching was killing productivity. Continuously switching between tasks to ensure both progressed meant quality for both had tanked.