Oussama Maachi

AI GTM Engineer @ Spekit · I ship AI agents into the revenue engine every week · −80% SDR workload across 50+ brands · MCP · Claude SDK · CrewAI · LangGraph · RAG

Spekit 🐙Gex-Generale Energie GroupMex-Mutuelle familiale d'algerie

“

I build the AI layer of the modern revenue Most "AI for go-to-market" fails for the same reason: teams bolt an LLM onto a leaky pipeline and expect magic.

More

The model is the easy part. The plumbing underneath — enrichment, routing, handoff, attribution — is where revenue is won or lost. I build the AI agents and systems that make it actually move the Today I'm an AI GTM Engineer at Spekit, shipping AI-agent workflows into production every week: signal workflows, persona enrichment, nurture sequences, MQL processing, AE handoff. Owned end-to-end across RevOps, Marketing, Sales, and Before Spekit, I ran AI automation across 50+ subsidiaries at Generale Energie Group (North America, EMEA, UAE). Three numbers I can defend with receipts:→ 80% cut in SDR workload via an AI-agent stack→ +35% ROAS through server-side CAPI on high-ticket paid→ +25% MQL to SQL by replacing static round-robin with signal-aware routingEight years, one through-line: revenue systems that How I build:→ Production AI agents and multi-agent systems end-to-end: Claude Agent SDK, MCP, CrewAI, LangGraph, Hermes, RAG, Pinecone→ The full GTM stack — Clay, ZoomInfo, HubSpot, Salesforce, Outreach, n8n — wired into custom MCP servers→ Multi-tenant scale across NA / EMEA / UAE→ Trilingual: native French, fluent English and ArabicI write here about what I'm learning building this — the agent patterns that hold up in production, the ones that don't, and what it actually takes to put AI in front of revenue.