Todd Blocksom · Product leader, AI platforms · Seattle
Good AI makes the person using it harder to replace.
I’ve spent twenty-five years building it: Dragon Drive, Siri in the car, and Google Cloud’s analytics surface. Now I build agents, full time.
-
25Years shipping product
-
200M+Vehicles running Dragon Drive
-
$2B+Public-sector revenue unblocked at Google
-
4×Siri training data, privacy safe
Work
-
2022 to now47north TechnologiesFounder and principal consultant
An independent product consultancy and R&D lab. Everything below sits under it.
-
HIDE/SEEK 2026 to now
Designed, built, and shipped a location-based multiplayer iOS game solo, from product spec to public beta, in 41 active build days.
-
Totem Grid 2025 to now
Principal PM for an offline-first mesh platform. A 100-user Burning Man deployment converted into a Burning Man org partnership and a FIFA partnership supporting World Cup 2026 staff operations.
-
Apple, Commerce and Payments 2023 to 2024
Contract program manager. Scoped EU Digital Markets Act impact for the analytics organization and drove GDPR-aligned anonymization work.
-
-
2018 to 2022GoogleSenior TPM, Cloud Data Analytics
Sole TPM for a 65-person organization across a dozen analytics products. Earned adoption for a shared component library with no authority to mandate it, cutting front-end effort 80%, and ran the VPAT certification that unblocked $2B+ in public-sector contracts.
-
2016 to 2018AppleSenior engineering program manager, Siri
Turned Legal’s delete-everything stance into approved, PII-scrubbed training retention, growing usable data 4×. Took Do Not Disturb While Driving from a Craig Federighi directive to shipped in one summer, and am co-inventor on the resulting patent.
-
2013 to 2016NuancePrincipal PM, Automotive
Owned the roadmap for Dragon Drive, the conversational AI platform behind BMW, Mercedes-Benz, and Volkswagen voice systems. It runs in more than 200 million vehicles today and powers MBUX.
-
2001 to 2013Earlier
RealNetworks, where I pitched and led a six-designer UX studio. Tweddle Group, on Toyota Entune and Chrysler Uconnect. Loudeye, on the O2 Germany music store. Microsoft Office, where I hold a patent on the Persona Menu and built an event-condition-action automation prototype that looks a lot like agent tool calling, twenty years early.
HIDE/SEEK
A location-based multiplayer game that turns a neighborhood into a game board. I built all of it: the product spec, the game economy, the design system, the SwiftUI client, a server-authoritative Postgres and PostGIS backend, a web spectator view, and the release pipeline. It is in public TestFlight beta, with the App Store submission in review.
-
41Active build days
-
2,149Commits
-
37Tables, all with row-level security
-
477Database assertions
Point of view
For twenty-five years, I worked on the human end of hard systems: voice in cars that had to work at seventy miles an hour; Siri, where privacy and quality pulled in opposite directions; and a Google Cloud analytics surface shaped by a dozen teams with competing ideas of what it should be.
The pattern was always the same. The technology itself was rarely what stood between someone and the outcome they wanted. The gap was in what the system assumed about them, what it made them prove, and how much work it handed back.
That is the distinction behind the headline. The interesting question is not how much of someone’s job an AI product can do. It is whether the person using it can do things they could not do before.
I spent this past year testing that question firsthand. It changed how I build.
Contact
If you’re hiring for an ambitious AI product and need someone who understands both the technology and the people using it, I’d like to hear about it.
Write to todd@47northtech.com, find me on LinkedIn, or download my resume.