LIVE · APPLICATION OPEN TULIO · AGENTIC REVENUE MARKETING MANAGER
NUMERO® · TAKAPUNA · AUCKLAND
● ROLE MATCH: 94% HUBSPOT n8n MAKE.COM ZAPIER CLAUDE / GPT / GEMINI META · GOOGLE · LINKEDIN · TIKTOK NOTION · CLICKUP

You're hiring
an agentic
marketer to hit
$10M. Most
will send a CV. I shipped the system.

This page is the portfolio. Every block you read is something I've built for service businesses at FlowTier — the exact GTM + RevOps + AI-workflow engine numero® is asking for, tailored to your $10M target. Talk to the chatbot at the bottom; it was built for this application.

Your job description, mapped line-by-line to systems I've shipped.

AGENTIC GTM · OUTBOUND✓ SHIPPED
"Build AI-assisted outbound sequences that combine automated research, personalisation, and human oversight."
Built an n8n + HubSpot-equivalent (GoHighLevel) pipeline that processes 2,400+ contacts per batch with a self-terminating pagination loop, AI enrichment, and a founder-tone email writer (60–100 words, strict JSON output). Human QA at the batch level, not per email.
LEAD SCORING & ROUTING✓ SHIPPED
"Develop and manage AI-powered lead scoring and routing systems."
Production VAPI inbound voice agent ("Daniel") with an educate-before-booking gate and tool-selection logic that routes to booking, knowledge base, or human escalation. Wired to CRM via $fromAI() dynamic variable injection.
CONTENT AT SCALE✓ SHIPPED
"Produce content at scale using AI for first drafts and research, with human editorial QA."
Structured prompt library + knowledge base RAG for FlowTier's outbound engine. Every output is JSON-schema-enforced so QA is a diff, not a re-read. Same architecture runs an IG sales agent ("July") that qualifies leads in ≤5 exchanges.
CRM / REVOPS✓ TRANSLATABLE
"Build and maintain lead scoring models, attribution frameworks, and pipeline reporting in HubSpot."
Architected GHL pipelines for multi-client onboarding with stage-based automation and a command-driven Claude agent that manages the full client lifecycle. Pipeline logic is CRM-agnostic — HubSpot migration is a tool swap, not a rebuild.
CEO PERSONAL BRAND✓ DONE FOR CLIENTS
"Manage and grow the CEO's personal brand across LinkedIn and other relevant platforms."
Run founder brands end-to-end: content engine, repurposing across channels, booking funnel. Richard's brand inherits the same playbook — thought leadership ladders to pipeline, not vanity metrics.
AI WORKFLOW DESIGN✓ CORE SKILL
"Build marketing automations using Google AI Studio, Make, Zapier, n8n, or equivalent platforms."
This is the day job. n8n is my primary IDE. Full-stack agentic systems: VAPI (voice), WPPConnect (WhatsApp), GHL (CRM), Claude/GPT (reasoning), vector knowledge bases (retrieval). Every workflow documented for team handoff.

Three systems currently running in production.

2024 — NOW
Founder · Operator

FlowTier — done-for-you AI Front Office for SMB services

Full-stack agentic systems for plumbers, HVAC, med spas, law firms, dental offices. I'm the operator and the builder: I sell it, I build it, I support it. The playbook below is the one I'd bring to numero® on day one.

  • Built multi-client cold outreach engine on n8n + GHL with AI email writer (founder-tone, 60–100 word JSON output) and self-terminating pagination.2,400 / BATCH
  • Shipped VAPI inbound voice agent (Daniel) with 3-sentence response cap, educate-before-book gate, knowledge-base tool, and dynamic CRM contact injection.SUB-500ms LATENCY
  • Designed Instagram sales agent (July) — ≤5-exchange qualifying flow, AI-reveal mechanic, strict KB-only pricing rules, direct funnel to demo.≤5 EXCHANGES
  • Architected WhatsApp infra (WPPConnect) with AI/human handoff via tag toggling, exponential-backoff reconnection, health checks, multi-instance redundancy.99%+ UPTIME TARGET
  • Built command-driven Claude Code agent that manages the full client lifecycle against CRM pipeline stages via a generic HTTP MCP layer.PM REPLACED
  • Positioned against price commoditization with a guarantee of 10 appointments or 20 hours saved in 30 days. Refund clause forces delivery rigor.30 DAYS · GUARANTEED
n8nMake.comVAPIGoHighLevelWPPConnectClaude / GPT / GeminipgvectorMCPNotionClickUpHTML / JS
// outbound_engine.v2.flowLIVE · 2,400 CONTACTS / BATCH
CRM QUERY tag: pending BATCH · 100 paginate ENRICH research agent CLAUDE WRITE 60–100w · JSON VALIDATE schema · len · tone SEND Gmail / ESP UNTAG remove pending LOOP · self-terminating when tag count = 0
// daniel.vapi.decision_treeINBOUND BDR · EDUCATE-BEFORE-BOOK
INBOUND CALL VAPI pickup · <500ms CLASSIFY INTENT 3-sentence cap EDUCATED? KB check QUERY KB answer & reframe send_booking_link $fromAI(contact_id) escalate_to_human if confidence < 0.7 CONFIRM + HANG CRM updated NO YES re-engage
// ai_front_office.n8n · 131 nodes · productionOMNICHANNEL · 6 AGENTS · 7 TOOLS
01 · CHANNELS · 7 ENTRYPOINTS 📱 LIVE CHAT webhook · new + re-eng 📸 INSTAGRAM DM · re-engagement 💬 SMS lead message webhook ☎️ VOICE (Daniel) VAPI · transfer-in 02 · ROUTING + BRAIN STATUS ROUTER ai / human / escalated ◆ AI AGENT CORE LLM: Anthropic primary fallback: Gemini ────────────── 6 parallel agents per-channel context 03 · TOOL LAYER · 7 CALLABLE book_appointment → calendar get_calendar_slots → availability update_contact → CRM fields human_escalation → staff ping future_follow_up → re-engage queue dnd_contact → suppress awaiting_reply → state flag 04 · MEMORY + KNOWLEDGE LAYER pgvector · RAG OpenAI embeddings · 3 indexes brand voice · pricing · FAQ Postgres chat memory persistent conversation state per-contact · per-channel n8n dataTables message queue · dedup 15 nodes · state ledger GoHighLevel CRM source of truth HubSpot-translatable // workflow stats 131 nodes 27 HTTP tools 9 state switches 6 agents · 2 LLMs 3 vector indexes SHIPPED ✓
SYSTEM DESIGN
AI ↔ Human Handoff

WhatsApp handoff state machine — what numero's JD asks for under "human oversight"

The JD says "automated research, personalisation, and human oversight." Here's what real human oversight looks like when you've shipped it — not a Slack notification, a state machine with explicit transitions, resume commands, and a lightweight staff dashboard.

// whatsapp.handoff.state_machineWPPCONNECT · MULTI-INSTANCE · 99%+ UPTIME TARGET
STATE_AI ◆ AI HANDLES tag: ai_active bot replies within 3s context from vector KB STATE_HUMAN ● HUMAN REPLIES tag: human_owned AI muted · context frozen dashboard shows conversation STATE_ESCALATED ● URGENT · NOTIFIED Slack + email ping SLA clock started supervisor dashboard staff tags "human_owned" /resume · /ai-on confidence < 0.5 · urgent keywords · refund · legal · angry acknowledged // infrastructure layer exponential-backoff reconnect · health checks every 30s · message queue on disconnect multi-instance WPPConnect redundancy · session state persisted · zero-downtime failover target
PRE-FLOWTIER
Operator · Affiliate · Marketer

8+ years in the trenches — courses, affiliate, physical product, paid media

Before I sold automation, I sold everything else. Digital courses, affiliate offers, and my own physical product (Triptofast — a supplement brand I built from formulation through launch). The JD calls for someone who's run paid, content, CRM, and pipeline end-to-end. I've done that for my own P&L — which is a different kind of rigor than running someone else's budget.

  • Triptofast (own brand): built a physical-product supplement business end-to-end — formulation, branding, funnel, paid acquisition, fulfillment, customer support.FULL P&L · SOLO
  • Affiliate marketing: ran multi-offer campaigns across verticals — the unforgiving school where CAC < commission or you're out of business by Friday.CAC < PAYOUT
  • Digital course sales: built funnels and paid campaigns for info products. Taught me that conversion copy is a math problem, not an art project.FUNNEL · COPY · PAID
  • Managed paid acquisition across Meta, Google, LinkedIn, TikTok — the exact channel matrix numero® is running.CPL · CAC · ROAS
  • Consulting engagements: positioning services, building pricing ladders, installing repeatable acquisition systems. Systems thinking isn't a trait I claim — it's what clients paid for.SYSTEMS · NOT TASKS
Meta AdsGoogle AdsLinkedIn AdsTikTok AdsGA4FigmaFunnelsCopywritingP&L ownership

The KPIs from your JD, translated into a first-90-days plan.

30d
Days 1–30 · Audit + Infra
HubSpot hygiene pass. Attribution model set. Pipeline stages defined with Richard. One agentic workflow shipped — probably AI-assisted outbound research — to establish the pattern.
60d
Days 31–60 · Channels
Meta + LinkedIn + Google live with testing discipline. Content engine running with AI drafts and editorial QA. Lead scoring model shipped. First cohort of MQLs through to sales.
90d
Days 61–90 · Compounding
Weekly pipeline report ritual. Prompt library documented and transferable. Dashboards live. AI discovery platforms (Perplexity, ChatGPT citations) tested as acquisition sources. Velocity compressing.
// $10M target · reverse-engineered from the bottom upPEOPLE → CUSTOMERS → REVENUE

Read top-to-bottom. Each stage is a headcount — the number of humans at that point in the journey — not dollars. The dollar math only appears at the final step, where closed customers convert to revenue.

WIDE · AWARENESS NARROW · CUSTOMERS ① PEOPLE WHO SEE YOUR MARKETING 250,000 people / month ad impressions · website visits · organic · AI discovery (Perplexity, ChatGPT) ↳ paid media is the biggest lever here 2% of them fill a form / request demo → ② LEADS · RAISED THEIR HAND 5,000 leads / month form fills · demo requests · AI-chat inbound · DMs ↳ now they exist in HubSpot with contact info 20% of leads look like real buyers (AI scoring) → ③ MARKETING-QUALIFIED LEADS (MQLs) 1,000 MQLs / month right company size · right industry · buying signals ↳ worth sales' time · marketing has done its job up to here 15% survive first sales call → ④ SALES-QUALIFIED LEADS (SQLs) 150 SQLs / month decision-maker engaged · budget confirmed · active deal 37% of sales-qualified deals close → ⑤ 56 NEW CUSTOMERS / MONTH
↓ PEOPLE → DOLLARS · THE FINAL STEP
56 new customers / mo  ×  $15,000 average annual contract  =  $840,000 MRR  ×  12 months  =  $10M+ ARR ✓

The one number I'm guessing at: the $15,000 average contract. That's a placeholder based on typical NZ SMB agency retainers. If numero®'s real number is $30k, we need half the customers. If it's $8k, nearly twice. Day one of the role: get the real number from Richard and rebuild this model for real.

COST TO ACQUIRE (CAC)
< $2,800 / customer
MQL → SQL RATE
15% now · 22% target
PIPELINE SPEED
45 days → 28 days
AI AUGMENTATION
80% of mkt tasks

↳ directional only · real model gets rebuilt in week 1 against numero®'s actual ACV, historical conversion rates, and sales capacity

Behavioural assessment, mapped to the JD.

CIS Assessment (DISC + Jung + Spranger · 99% validity index · 15 Aug 2024). Summary below; full report available on request. I've pulled the signals that matter most for this role.

// disc.profile · natural vs adaptedPROFILE: ID → DS (growth shift)
100 66 33 0 56 65 D Dominance 84 35 I Influence 33 60 S Steadiness 27 40 C Compliance Natural (ID) Adapted (DS)

Natural state: communicator, persuader, connector (I=84, D=56). Adapted state: more structured, patient, systems-driven (S=33→60, C=27→40). The shift itself is the signal — the JD asks for a "systems thinker" who still needs to own content + CEO brand. That's this profile.

★ HEADLINE COMPETENCIES
Enthusiasm94
Persuasion92
Extroversion92
Boldness76
Sociability65
Command61

Top-decile persuasion + enthusiasm maps directly to content engine, CEO brand, sales enablement.

✦ MOTIVATOR HIERARCHY
86
Economic · ROI
62
Theoretical
51
Political · Power

Economic 86 = ROI-native thinking. That's numero®'s brand. The fit isn't cosmetic — it's wired in.

◎ LEADERSHIP STYLE
Motivator 42% · Executive 28% · Methodical 16.5% · Systematic 13.5%

Leads through dialogue, momentum, and optimism — the right mode for marketing-to-sales handoff with a founder-led sales process.

In the cockpit
// off the clock
Licensed helicopter pilot — in aviation, the cost of imprecision is non-recoverable. Same discipline I bring to production AI systems. Ten years of kickboxing · moves fast on new tools (Make.com, Notion, ClickUp — added when a project needed them, not when a course demanded it).

Don't take my word for it. Ask the agent.

This chat is a live Claude deployment running in the resume itself. Ask about my stack, how I'd approach numero®'s $10M target, salary expectations, relocation, or the architecture of anything above. The answers come from me, through the model.

TULIO-AGENT · v1.0 · claude-sonnet-4
● ONLINE
TULIO-AGENT
Richard — or whoever's reading this — thanks for getting this far. I'm a Claude model briefed on Tulio's full stack, work history, and how he'd approach the numero® role. Ask me anything. I'll answer in his voice, cite what he's actually built, and flag anything I'm not sure about.
>

Nine years in the US. One country I've always meant to end up in.

Most applications for this role will come from people who'd take it because it's a good job. For me it's a good job and the country I've spent a long time planning to move to.

I moved to the US nine years ago. I built FlowTier here, learned the stack I'm now paid for, and came to know what I actually want the next decade to look like. New Zealand has been on that list the whole time — it's not a shortlist of three, it's this one.

Which is to say: the visa sponsorship isn't a perk I'm trading for, it's the thing that makes a move I'd already decided on possible. When someone shows up with a deliberate decision behind them, retention usually takes care of itself.

COMMITMENT
Long-arc
I moved countries once already. I know what it takes.
ALIGNMENT
In-person
Takapuna office is not a compromise — it's the point.

The quickest way to know if I'm right for this is a 15-minute call.

REF: numero® · AGENTIC REVENUE MARKETING MANAGER · TAKAPUNA · $120K NZD · VISA SPONSORED