GTM Engineering & Revenue Operations

Design the system. Scale the revenue.

I engineer the agentic systems that turn your GTM strategy into predictable revenue.

R$16B+

AUM served

3

built-to-handoff systems

30%+

peak reply rate

The system I build, end to end

Built revenue systems inside

Monte Bravo Partner & Head of BI · R$16B+ AUM
XP Investimentos BI & Analytics
Rico BI & Analytics

What I do

GTM systems that actually work

I design and build the technical layer between your team and your revenue targets.

01

GTM Engineering

Automate outbound, inbound, and PLG motions with enrichment pipelines, multi-channel sequencing, and AI-powered workflows.

02

Revenue Operations

CRM architecture, lifecycle design, and unified reporting. Built around how your team actually sells, not how the vendor demo works.

03

Data & Enrichment

Turn raw contact and account data into prioritized pipeline with ICP scoring, intent signals, and research at scale.

The architecture

One system, layer by layer

Every engagement is a variation on the same blueprint. The tools change with the budget and the motion — the layers don't.

Build layer
Claude Code Purpose-built agents

I architect and ship the entire stack below from an agentic workbench — which is why a system like this goes live in days, not a quarter, and keeps adapting as the motion changes. The agent layer is how I build; the layers below are what it produces.

  1. 01

    Source

    ApolloSales NavigatorGoogle MapsIntent signals

    Pull the right accounts and people — not just more of them.

  2. 02

    Enrich

    ClayHunter.ioWeb scraping

    Fill the gaps that actually decide fit, with waterfall fallbacks.

  3. 03

    Qualify

    ClaudeICP gateAA/A/B/C scoring

    An LLM scores fit and tiers every record before a human spends a minute.

  4. 04

    Activate

    InstantlyWhatsApp / Z-APILinkedIn

    Multi-channel outreach, personalized per record, paced to stay deliverable.

  5. 05

    Measure

    DashboardsSQLLooker

    Funnel visibility leadership trusts — the BI layer most pipelines skip.

Orchestration & state

n8n moves work between layers; Supabase holds durable state so nothing double-fires across long cycles. This backbone is what turns a pile of tools into a system that runs without me babysitting it.

Selected work

Systems I've built

US fractional CFO firm (construction finance)

Three-way beachhead split for a construction finance firm

Signal-driven outbound architecture for a US fractional CFO firm serving construction contractors. Refused the single-platform framing, shipped a composite signal model anchored on a 90/90 overlap validation, full architecture handoff to the client's internal automation engineer.

2,350+

Capterra rows analyzed

Apify Claude Code Supabase n8n +1

Internal portfolio (4 niche content sites)

From WordPress migration to autonomous content operations across 4 properties

Migrated 3 WordPress sites to Astro with zero SEO regression on a property doing 4.8k clicks/quarter, then layered an autonomous content + SEO pipeline with a Claude critic gate across the full 4-site portfolio. Self-running, ~$3/month total cost, weekly digest as the only inbox-touching surface.

4

self-running properties

Astro Cloudflare Workers Claude API Python +2

Nordic spa & sauna manufacturer (US)

Multi-vertical outbound system that grew with the business

Built a multi-vertical outbound pipeline for a US specialty manufacturer — 9,000+ leads processed, 18 campaigns across 15 verticals, evolved from volume outbound into signal-based prospecting and conference re-engagement layered on top. Built-to-handoff — client's internal team now operates the system.

9,000+

leads processed

Clay Instantly Hunter.io Firecrawl +2

US Coffee Tech

Signal-based outbound system for a specialty coffee tech company

Designed ICP signal architecture using Google Maps, AI site analysis, and competitor detection. Highest-value move was blocking a premature CRM integration — and coaching the RevOps lead from CRM admin into systems architect along the way.

0 → V1

outbound system designed

Clay HubSpot Google Maps API Hunter.io +1

Monte Bravo

AI recruiting pipeline for a R$45B+ wealth management firm

8,000+ candidates analyzed with AI scoring, 1,100+ qualified leads, 30%+ WhatsApp reply rate. Sourced advisors including one with R$1.5B+ AUM from a top-tier global bank.

8,000+

candidates scored

n8n Supabase Apollo OpenAI +1

Hospitality group (São Paulo)

AI concierge taking a founder out of the reservation inbox

WhatsApp AI agent handling guest inquiries, reservations, modifications, and confirmations for a São Paulo wine bar — replacing the founder, who was personally answering inbound messages despite operating a multi-venue group.

24/7

first-touch coverage

n8n WhatsApp OpenAI Redis +2

Online mentoring & education

Inbound AI SDR closing the full enrollment loop on WhatsApp

AI SDR for an EdTech mentoring company — picks up inbound leads on WhatsApp, qualifies, handles objections, confirms program fit, and closes the enrollment without manual handoff. HubSpot stays the source of truth.

Full close

AI handles lead → enrollment

HubSpot n8n WhatsApp OpenAI +2

Open source

Tools I've built and shipped

Agentic systems released under MIT for anyone to fork and run.

Open source · MIT

Outbound Campaign Crew — claim-traceable cold email sequences in CrewAI

CrewAI planner→copywriter→critic crew that turns an Account Research Agent brief into a 3-touch sequence where every factual claim is traceable, in code, to a line of research evidence — and refuses to write at all if the account scored out of ICP. The governance pattern from the LangGraph trilogy, ported to a second framework.

~$0.15

per 3-touch sequence

CrewAI Flows task guardrails Python +2

Open source · MIT

eval-watch — discipline layer over the agent stack

Meta-runner that wraps each sibling's existing eval entry via subprocess adapters, tracks regression and drift in SQLite, and commits a STATUS.md report back to the repo on a monthly GitHub Actions cron. Four siblings across two frameworks (LangGraph + CrewAI). Sixth instance of the code-enforced-rule pattern.

~$35

per year operational cost

SQLite GitHub Actions subprocess adapters Python +1

Open source · MIT

Meeting Prep Agent — composes ARA + SM into a pre-meeting brief

LangGraph agent that produces a 1-page sales-prep brief by composing the Account Research Agent and Signal Monitor via subprocess. Four code-enforced clamps (evidence, length, talking-point provenance, spend) make the brief safe to read at face value. Third repo in the pre-outbound stack.

$0.40

per brief (ARA + SM + MPA combined)

LangGraph Anthropic SDK Pydantic Claude Sonnet +3

Open source · MIT

Signal Monitor — typed buying-window signals with code-enforced clamps

LangGraph agent that watches a watchlist of companies for six typed buying-window signals and emails you a weekly digest. Three code-enforced clamps (evidence, spend, dedup) make it safe to leave running unattended. Companion to the Account Research Agent — modular, not coupled.

$0.56

first 5-company run (live)

LangGraph Anthropic SDK Pydantic Exa +8

Open source · MIT

Account research agent with code-enforced disqualifier policy

LangGraph agent that takes a company domain and produces a decision-ready ICP-fit brief — with a self-correcting critic loop and a hybrid disqualifier policy enforced in code, not in the prompt.

$0.30

per account brief

LangGraph Anthropic SDK Pydantic Firecrawl +3

Let's talk

Let's build your revenue system

I'm available for GTM engineering and RevOps projects. Tell me where revenue is leaking and I'll tell you what I'd build.