From surprise reviews to continuous alignment
Problem
Reviews happen once or twice a year. Work changes every week.
Growth
is measured against generic, outdated career ladders.
Visibility
comes too late. Most problems don't show up until review time.
Priorities
change fast and get lost between managers and reports.
Solution
AI-generated performance narratives & a simple workflow that keeps managers & reports aligned.
Key innovations
An automated, evidence-backed, & employee-curated performance record
Built-in priority alignment between employees and managers
Pulse performance ratings so everyone stays on the same page
History that turns recurring conversations into 1-click reviews
Impact
84%
activation in month one among design partners
77%
of employees preferred Surface to self-written updates
+17 pts
jump in rated accuracy in month 1 (65% → 82%)
Kickoff
I facilitated a design sprint to kick the team off. We built rapid prototypes for testing for early validation.
Sprint to ship in 2 months
Core idea & iteration · 3 weeks
Design iterations · 2 weeks
Evals setup · 1 week sprint → ongoing
Phased design partner launch · 2 weeks
Evals
Version 1 of Surface was vibe-prompted.
Since then, I've built out AI prompting workflows that allow us to make more deterministic calls through decision trees, improve our outcomes, and lower token spend.
In one month, users rating Surface accuracy a 4 or 5 out of 5 jumped from 65% to 82%. Our team also moved to a smaller, cheaper model with no drop in quality.
