📍 Brisbane ↔ San Francisco | 🤖 AI Product Manager | ✍️ Agent Goose founder

10+ years across consulting, operations, and product. Scaled platforms from $6M → $21M ARR, grew ecosystems from 40 → 130+ integrations, and shipped AI features that cut review time by 50%. Now I vibe code AI agents, ship fast, and trust the process.

• 🦞 OpenClaw: Personal AI chief-of-staff system. Agnes orchestrates CPO, CTO, CMO, and CRO agents across Google Calendar, Gmail, Discord, and Google Sheets with automated reporting and a local Mission Control dashboard.
• 🪿 Agent Goose: Writing about the future of agentic AI and what it means to build, ship, and work alongside AI day to day

• 🐾 Bring Simba Home tracker: Internal tracker for moving Simba (Border Collie) from the US back to Australia. Tracks import requirements, vet timelines, and quarantine steps end-to-end.

• 🌐 Curated Website Projects: A collection of websites and interactive experiences designed and built using AI-assisted tools

  • 🏠 Floating House Design Concept: A 3D floating family home concept built with Omma. Exploring generative architecture and spatial design using AI-assisted 3D modelling tools
  • 🎮 Pokémon Fan Game: A browser-based game built around the original 151 Pokémon. Covers the full battle loop, type matchups, and classic mechanics from Gen 1
  • 🦷 Brisbane Dental Website: Clean, conversion-focused dental practice website designed and prototyped in Moda
  • 🪟 Viral Glass AI: Dark-themed landing page for an AI web design agency. Built with React, Tailwind, and Framer Motion. Features a liquid glass glassmorphism aesthetic with a cinematic hero backdrop

• ⛽ Watch My Fuel: Free live QLD fuel price comparison tool. Search by suburb or address, view stations on an interactive map with pricing indicators, and track price trends by fuel type and brand

• 🌐 Portfolio: A showcase of AI case studies, side projects, and my approach to building and shipping products fast. From idea to deployed in 24 to 72 hours using Claude Code, Cursor, and Vercel.

• 🧠 Brandon's PM OS: Open-source personal operating system for Claude Code. Built for PMs with a full workspace structure, 4 custom slash commands (/standup, /weekly-update, /new-post, /ship-check), a self-improving skill that extracts reusable knowledge from every session, and a growing library of PM-specific Claude Code skills

• 📚 OpenClaw Playbook: A curated collection of OpenClaw use cases. Browse real examples of AI chief-of-staff automation built on the OpenClaw framework

• 🗺️ Travel Log: Personal travel tracking app built with Next.js and Supabase. Deployed multiple AI agents via Pencil to design the first iteration. Globe and satellite map views, city cards with auto-fetched photos, trip type tagging, and live stats across countries, cities, and continents

• 🧪 growth-experiment: A Claude Code skill for Growth PMs. Give it a metric drop or funnel problem — it generates top 3 ranked experiment ideas using Akash Gupta's 10×10 behavioral trigger matrix, writes a full Atlassian-structured experiment brief, calculates sample size and runtime, and outputs a copy-paste ready Amplitude Experiment setup block

• 🏗️ Crevo Partners: AI marketing agency for local contractors. We build high-converting websites, run SMS and email automations, manage online reputation, and deploy Voice AI agents to catch and convert leads. Tradies spend less time chasing work and more time on the tools.

  • 🌿 Landscaping: SEO-optimised site for a local landscaping business
  • 🔧 Plumbing: Clean, conversion-focused site for a plumbing company
  • 💧 Pressure Washing: Landing page for a powerwashing business

• 🦉 Internal Helpdesk Agent: RAG-powered AI agent that resolves internal support tickets, routes to the right teams, and files bugs to Linear. LLM-as-judge scoring for relevance, accuracy, and citation quality

• 🐸 Eat the Frog: Android checklist app built around the Eisenhower Matrix. Prioritise by urgency and importance, tackle the hardest task first

• 🎯 Alice ABM Prototype: End-to-end account-based workflow covering ICP scoring, buying committee discovery, and coordinated multi-persona messaging with human review checkpoints

• 🚀 Retro Space Shooter: Built end-to-end with Cursor, Claude, and Vercel. Full game loop with movement, shooting, enemies, scoring, and boss battles, shipped in a weekend

• 🐍 Awesome Python Scripts: Fork of a curated Python automation scripts collection

• 🛍️ Product Mockup AI: Clickable prototype for an AI-powered product mockup feature inside a fictional website builder for small businesses and creators. Helps non-technical merchants turn a basic product photo into polished, in-context website visuals through a guided workflow with strong defaults

• 🧠 Go1 Content Intelligence Prototype: Interactive prototype showing how an AI layer (Morgan) makes learning recommendations visible, explainable, and improvable inside the LMS. Surfaces role signals, a dismiss feedback loop, and an agent activity log so admins can see why content is being recommended and course-correct in real time. Morgan also surfaces recommendations directly in Slack, meeting learners where they already spend their day rather than asking them to go back to the LMS.

• 🔬 11x Alice ABM Prototype: Built an end-to-end account-based workflow prototype showing how 11x Alice could move from target accounts to pipeline, with tighter control, better coordination, and measurable outcomes. Covers account selection and prioritisation based on ICP and intent signals, buying committee discovery, persona mapping, and account coverage. Finishes with coordinated multi-persona messaging and campaign execution, with human review checkpoints at each stage.

• 🛡️ Drata TPRM Continuous Monitoring: Prototype rethinking how Drata's Continuous Monitoring could move from a passive vendor change feed into an actionable risk system. Connects Drata's existing assets (vendor records, questionnaires, AI evidence review, and the risk register) into a post-onboarding vendor risk workflow that detects meaningful risk shifts, explains why they matter, and tells teams what to do next.

• 🎯 Amplitude Recommendations. Real time ranking engine that clusters users by behavioral signals, ranks items by likelihood to drive a target outcome, retrains hourly, and serves results via the Profile API in under a second. Worked with HBO to enrich content metadata with release date, category, and SKU so the model could surface new releases over old content. Result: +20% engagement, +15% content consumption. Lesson: recommendation quality is usually a data problem upstream, not a model problem.

• 🔮 Amplitude Predictions. Deep learning model (Nova AutoML) that scores every user on churn or conversion propensity within 7 to 90 days, refreshed daily or hourly. Chick-fil-A used it to get ahead of lapsing customers, suppressing promos for low risk users and triggering retention for the high risk segment before anyone churned. Lesson: shifts teams from reactive to proactive retention.

• 🎬 Amplitude Session Replay AI. Over 50% of new Replay users never opened the product. Those who did had to scrub through 10 to 15 minute videos to find anything useful. We shipped an LLM powered Summary Card grounded on structured event data (not raw video) to surface key actions and friction moments automatically. Built a labeled eval set from real customer replays to track groundedness, signal capture, and noise. Every prompt change ran against it before shipping. Result: 50% drop in replay review time, 17% lift in first time user activation. Lesson: grounding beats clever prompting, and AI alone does not fix activation. The workflow change mattered just as much as the model.

• 🌐 Amplitude Developer Portal. Amplitude was expanding from a single product to a multi-product platform with CDP as a company priority, but Sales couldn't build pipeline without ecosystem coverage and engineering was bottlenecked by custom integrations. I led the 0 to 1 developer motion: defined marketers as the initial ICP, prioritized audience activation APIs, and chose API key auth over OAuth to reduce partner onboarding friction. Built a self-serve portal with sandbox environments, data contract validation, and unified docs. Introduced a 4-tier partner model to align go-to-market investment to revenue impact. Result: 40 to 140 integrations, hit the 10 CDP rip-and-replace company goal. Lesson: ecosystem credibility is a sales problem as much as a product problem. The integration count changed how deals were won.

The tools I use to prototype, ship, measure, and collaborate across the product lifecycle.

• Designing a Production-Grade Internal AI Agent
• Maximize the Power of Your Data Ecosystem
• Is Your Data Actually Reliable?
• First-Party Data: What It Is, Importance, & How to Gather It
• 32 Product Management Interview Questions & How to Answer

📬 More at agentgoose.substack.com

• 🗞️ Forbes, TechCrunch, Engadget, Fox Business: Covered the launch of Uber Moments. Imagine booking a chocolate-making class or a rooftop yoga session directly inside the UberEats app. Think Airbnb Experiences, but we built it first. First experience marketplace on a food delivery platform, $109 avg ticket vs. $22 platform average
• 🎤 Featured speaker at Twilio SIGNAL 2025: Co-presented with Paige Wagner Slaboski from The RealReal on personalization at scale, powered by Twilio Segment + Amplitude. Amplitude data revealed that top customers treat TheRealReal like a social feed, browsing long sessions and tracking items via Obsessions lists. That insight shipped into Obsessions carousels, sort-by-Obsessions, and list sharing, driving a meaningful lift in GMV. Also showed how behavioral data surfaced Reconsign flows: flywheel customers revisiting past orders to consign again, making it easier for high-value users to stay engaged on both sides of the marketplace. Presented to 200+ attendees.
• 🎓 Golden Key Honor Award (top 15% at QUT)

1. Agents are becoming the primary users of software. We've crossed from agents that assist humans to agents that operate end-to-end, with sandboxed compute, persistent memory, file systems, and direct API access. Paul Graham said "make something people want." The new imperative is make something agents want. If your product requires a human to log in, click through a UI, or see an ad, agents will route around you. The platforms that are API-first and MCP-accessible will capture the next wave of enterprise workloads. The ones that aren't will slowly become invisible.

2. Data quality becomes the AI agent's competitive edge. AI commoditises software features. What it cannot replicate is proprietary data. The best agents will be the ones with access to the richest, cleanest structured and unstructured data. The data gap becomes the intelligence gap. An agent plugged into clean CRM history, product usage data, and past support tickets will outperform one working from fragmented, poorly labelled data every time. The companies that invest in data infrastructure now are building the foundation their agents will run on.

3. Outcome-based pricing replaces the seat. Intercom already made the move, charging per resolved conversation instead of per user license. When one person running agents does the work of five, a seat is no longer a meaningful unit of value. Klarna replaced 700 support agents and those Zendesk seats disappeared overnight. The SaaS companies that figure out how to price for outcomes and not headcount will win the next decade of B2B. The ones that don't will watch their seat count shrink as their customers scale with agents.

"Speed is strategy. Ship the smallest version that generates real behavior, measure it, and iterate. You're optimizing for learning velocity, not feature velocity."

• 🐾 Dad hooman to Simba, a border collie from Team Reeves Ranch. Extremely food motivated. Will sit, shake, and solve your problems for a kibble.
• 🎾 Tennis fanatic. Life goal is to attend all 4 Grand Slams. 0 down, 4 to go
• 🎮 Dota 2 noob. Pos 4/5 support, favourite hero is Shadow Shaman. Please be kind in chat
• 🃏 Collecting Pokémon cards. The hobby that started as nostalgia and became a problem
• ✈️ Avid traveller. 43 cities and 18 countries visited. Always planning the next trip