Close

Matt Tannyhill

Product Manager building agentic platforms

About Me

Product Manager with 7+ years of experience building AI-powered platforms, agentic systems, and data products at enterprise scale. Shipped new products from concept to widespread enterprise adoption, designed sensor-to-action architectures with governed tool execution and human-in-the-loop approvals, and scaled AI features across 2,600+ enterprise accounts.

Software development background with hands-on use of LLMs, Claude Code, and automation tooling to accelerate product delivery.

Experience

Senior Product Manager, Pattern Intelligence

Pattern

  • Driving strategy and delivery for Pattern Intelligence, an agentic automation platform unifying signals and capabilities across ~25 internal and SaaS systems; building a governed registry of 75+ tools that lets the assistant safely execute approved actions from chat.
  • Architected the sensor-to-action loop powering 1,000+ sensors; beta indicates ~15 hours saved per ecommerce manager per week and ~1,500 tasks created weekly today, projected to ~200,000 tasks per week at rollout to 350 core customers, with approvals and outcome tracking built in.
  • Launched a net-new Generative Engine Optimization (GEO) product from concept to 800 customers in 5 months; used Claude Code to prototype core workflows and accelerate shipping with a small team.
  • Built custom GPTs and n8n automation workflows that streamlined competitive analysis, PRD generation, and ClickUp backlog creation, cutting spec cycle time and reducing back-and-forth with engineering and design.

Product Manager

Domo

  • Owned the agentic analytics and visualization layer across 2,600 enterprise customers, enabling natural-language data queries, text-to-visualization generation, and LLM-assembled data applications; conversational answer accuracy improved 250% post-launch and adoption reached 85% of the customer base.
  • Directed a cross-team initiative to ship a markup language for dynamic app components, cutting data product build time from ~4 days to ~3.5 hours and adopted by ~7,000 weekly active users.
  • Built a governed semantic layer (ontology modeling, naming conventions, column definitions, join relationships) adopted across 115,000+ customer-modeled datasets; supported external LLM integrations and Domo-hosted models for data-sensitive enterprise customers and directly supported $25M in enterprise renewals.

Earlier Roles

Eide Bailly Technology, XLR8 Development, Mozenda

  • Eide Bailly Technology (2019–2021): Technical Program Manager and BI Consultant; helped stand up a PMO and resource planning across 50+ mid-market clients; built star-schema analytics models and Tableau and Power BI dashboards.
  • XLR8 Development (2018–2019): Software Developer and team lead; shipped 3 customer-facing products end-to-end; owned technical decisions and worked directly with stakeholders on requirements, scope, and release readiness.
  • Mozenda (2018–2019): Technical Project Manager; led intake-to-launch for 50+ customer integrations totaling $3M+; ran discovery and aligned requirements across clients and engineering through delivery.

Projects

Pattern Intelligence

Agentic automation platform unifying ~25 internal and SaaS systems into a single execution layer. Sensor-to-action loop with approvals, outcome tracking, and a governed registry of 75+ tools the assistant can call from chat.

Generative Engine Optimization (GEO)

Net-new product launched concept-to-adoption in 5 months, reaching 800 customers. Prototyped core workflows in Claude Code to accelerate shipping with a small team.

Vaccine Injury Site

Personal Next.js + TypeScript project. Source on GitHub.

View on GitHub

NBA Shot Predictor

Two scikit-learn models predicting NBA shot outcomes (AUC 0.634) and player salaries from box-score stats. Rebuilt in 2026 from a 2019 BYU class project after Microsoft retired the original Azure ML Studio backend. Stack: trained locally with Random Forests, served from Next.js with Python serverless functions on Vercel.

Live demo  ·  GitHub

Fighter Arena

Turn-based combat between Robots, Archers, and Clerics. Create fighters, pick two, and watch a step-through battle play out. Rebuilt in 2026 from a 2019 BYU CS 235 C++ class lab as a TypeScript port preserving the original damage and ability math. Stack: Next.js + Tailwind, fully client-side simulation.

Live demo  ·  GitHub

Northwest Labs — COVID-19 Campaign Analytics

Browse, analyze, and run a hit-goal predictor on ~4,800 GoFundMe campaigns scraped during the March 2020 COVID-19 surge. Unified rebuild of four 2020 BYU INTEX class repos (Django + React 16 + Azure ML + ASP.NET) into a single Next.js 15 app. Logistic regression re-trained locally and shipped as JSON weights for client-side prediction, replacing the Azure ML Studio endpoint Microsoft retired in 2024. Stack: Next.js + Tailwind + Recharts on Vercel, with all ~4,800 detail pages prerendered as static HTML.

Live demo  ·  GitHub

Skills

Education

Master of Science, Information Systems Management

Marriott School of Business, Brigham Young University

Bachelor of Science, Information Systems

Marriott School of Business, Brigham Young University

Get in Touch