Engineering
Engineering at Delivr.ai
How we build and run deterministic identity and person-level intent at scale. Real systems, real tradeoffs, real numbers.
- engineering processai
Ryan Spoone, CTO · · 6 min read
Agents implement, humans gate
Agents write most of our first-pass code. Humans review every diff and perform every merge. The interesting part is not the agents. It is the harness that makes their output safe to ship.
Read post - intentdata engineeringarchitecture
Ryan Spoone, CTO · · 7 min read
We rebuild the intent graph from scratch every day
Over 1.2 billion person-topic intent signals across roughly 27 million people and 2 million companies, thrown away and rebuilt whole every 24 hours. For a signal that decays in days, freshness is not a cadence choice. It is an accuracy claim.
Read post - ad measurementidentityattribution
Ryan Spoone, CTO · · 7 min read
View-through is a join, not a model
We put the same identity pixel inside the ad slot and on the advertiser site, then measured what display actually does. The typical payoff is a return visit three days later, which is exactly what click math cannot see.
Read post
