I Build AI Systems That Generate Pipeline, Not Slide Decks

Requirement: Deploy and Train AI SDR Agents

What I've Done

I haven't just evaluated AI SDR tools — I've built the infrastructure they run on and the qualification logic that feeds them. At LXT.ai I built an AI content engine that processes inbound signals and produces multi-channel output at 3x the volume with 80% less manual effort.

My property valuation AI processes 100+ analysis events daily with zero human intervention for standard cases — the same pattern as an inbound qualification agent: ingest signal, classify, enrich, route. When confidence is low, it flags for review. When it's high, it ships.

For the weekly refinement cycle: I run this loop daily on my own products. GTMIntel's classification accuracy improved from ~70% to ~92% over 90 days of iteration on prompts, training data, and routing logic. This is the same discipline the AI SDR role requires.

Requirement: AI-Powered Inbound Qualification

What I've Done

I built the AI Marketer funnel on this very site — a real-time inbound qualification system that captures leads, scores intent, and routes to the right next step. The form submissions feed into a structured pipeline with Slack notifications, CRM integration, and automated follow-up sequences.

At LXT.ai, I managed the content engine across 3 brands simultaneously — scoring which content drove demos vs. which drove MQLs vs. which was noise. Same skill set as real-time lead qualification: capture, score, route.

• ✓ Real-time lead capture — built on this site with Turnstile bot protection, Slack integration, and structured data routing
• ✓ Lead scoring logic — different CTAs, different intent signals, different routing for consulting vs. tools vs. workshop leads
• ✓ No lead waits, no lead gets lost — every form submission triggers an immediate structured notification with full context

Requirement: Automate CRM Hygiene, Call Summarization, Follow-Up

What I've Done

I cut weekly content production from 40 hours to 8 hours while tripling output at LXT.ai. The principle is identical: identify every minute of admin work that doesn't produce revenue, then automate it.

I've built production integrations connecting multiple systems via APIs — the same pattern as connecting Gong → Salesforce → Outreach. My tech stack speaks your GTM stack's language.

• ✓ HubSpot RevOps certified — I know the CRM, the workflows, the reporting, and the automation capabilities natively
• ✓ Production API integrations — this entire site (Rust/Rocket) connects to Slack, email systems, databases, and external APIs via custom code
• ✓ Automated call summarization pattern — I've built the same "ingest unstructured data → extract structured insights → route to system of record" pipeline for property analysis

Requirement: Architect the AI Layer Across the GTM Stack

What I've Done

The job post asks for someone who can make Salesforce the hub, with data flowing between Marketo, Outreach, Gong, and Qualified without manual intervention. This is literally what I do.

I architected the data layer for GTMIntel — pulling data from multiple sources, normalizing it, enriching it, and serving it through an API. The architecture pattern is identical to what Flosum needs: disparate systems → normalization layer → enrichment → single source of truth → agent consumption.

Requirement: Pipeline Math and Unit Economics

What I've Done

I built live calculators for this because I think in pipeline math. When someone says "we need 67 SQLs per quarter at 20% win rate and $75K ASD" I immediately see the levers:

67 SQLs at 20% win rate and $75K ASD = $1.005M quarterly pipeline. To compress: improve top-of-funnel lead quality (AI qualification), increase win rate (AI-assisted selling), or shorten cycle time (automated follow-up). I'd start by instrumenting the data to see where you're actually losing deals — then deploy AI at the bottleneck. Not everywhere. The bottleneck.

Requirement: Hyper-Personalized Outbound at 10x Volume

What I've Done

I built the Marketing Campaign Generator — a tool that takes ICP signals and produces structured, multi-channel campaign strategies. The same pattern scales to individual accounts.

For Koinly's Australian tax season campaign, I achieved 5x ROAS on YouTube by building automated creative testing with performance feedback loops — the same "research → personalize → sequence → measure → refine" loop that powers AI outbound.

My Customer Avatar Builder structures the exact ICP signals you mentioned — team size, current tooling, compliance requirements — into actionable targeting criteria.

Marketing Campaign Generator · Customer Avatar Builder

Requirement: Competitive Intelligence Engine

What I've Already Built

GTMIntel is literally this. A system that monitors market signals, structures competitive data, and delivers insights via API. I built it because I saw the gap in my own GTM work — I needed real-time competitive intel to write better outbound and position better against alternatives.

The weekly structured output you described? I've been running that loop myself for months. The system monitors, I review, I ship insights. Scale it to Flosum's competitive set and you have exactly what you described.

gtmintel.com · API Docs

Requirement: Closed-Loop Data Feedback System

What I've Done

I built RefreshAgent — an AI system that monitors Search Console and Analytics data 24/7, detects anomalies, and feeds structured insights back to the marketing team. The same architecture applies to BDR activity → pipeline outcomes → agent performance.

Every tool on this site follows the closed-loop pattern: input → process → score → output → measure → refine. I don't build one-way systems.

refreshagent.com

Requirement: AI Agent Orchestration

What I've Done

I currently run multiple AI systems in parallel, each handling different segments:

• ◆ PropertyValuationAI — autonomous analysis agent handling 100+ events/day with confidence-based routing
• ◆ GTMIntel — continuous monitoring agent scraping and classifying market data
• ◆ RefreshAgent — anomaly detection agent monitoring marketing data streams
• ◆ This site — 15+ SEO tools, each an AI-powered micro-agent handling a specific analysis task

Managing multiple agents means resolving conflicts, preventing duplicate work, and maintaining data quality at the hub. I've hit and solved these problems in production — not in a workshop.

Requirement: Write Production Code, Not Just Configure No-Code

Proof

This entire website is a custom Rust/Rocket web application I built from scratch. It serves 15+ interactive tools, handles form processing, webhook integrations, real-time event streaming, database operations, and API integrations — all written by me.

Requirement: Prompt Libraries, Training Datasets, Refinement Workflows

What I've Done

Every AI tool on this site is powered by carefully engineered prompts and processing pipelines that I iterate on continuously. My content engineering framework documents the exact process: brief → retrieval → generation → quality gate → output.

I treat agent training as a discipline. GTMIntel's classification accuracy went from ~70% to ~92% through structured prompt iteration, training data curation, and weekly evaluation cycles over 90 days. Not one-time setup — continuous refinement.

Content Engineering Framework · AI Marketer Methodology

Requirement: Small Team, High Intensity, Do the Work Yourself

This Is How I Operate

Every product I've listed on this page — GTMIntel, PropertyValuationAI, RefreshAgent, this entire website — I built myself. No agency. No large support staff. No delegation layer.

At LXT.ai I managed global content across 3 brands as essentially a one-person content engineering team, growing leads by 25%. I scale through automation, not headcount. That's not a philosophy — it's a constraint I've optimized for.

Requirement: Strong Opinions on AI vs. Human

Here Are Mine

Requirement: Salesforce Ecosystem Familiarity

What I Bring

I understand what your buyers care about because I've worked adjacent to the Salesforce ecosystem for years. CI/CD, data protection, compliance workflows — these aren't abstract concepts to me. I've built tools that serve similar audiences.

My HubSpot RevOps certification means I think in CRM architecture, not just marketing campaigns. I understand object relationships, workflow automation, reporting hierarchies, and data governance — the same principles that apply in Salesforce, just with different admin tools.

More importantly: I understand why enterprise Salesforce teams care about DevOps, data protection, and compliance. DORA isn't just an acronym — it's a trigger event that creates buying intent. I can spot those signals and build outreach around them.

The First 90 Days

Requirement-by-Requirement Match