For candidates
Turn real projects into private feedback first, then share selected evidence when it helps you.
Private beta for AI-native builders
FirstRung
A private builder coach and evidence passport for young people learning to build useful software with AI.
- Private by default
- Local evidence first
- Candidate-approved sharing
Not a CV scraper
Evidence for how someone actually builds
AI makes polished project outputs easier to create. FirstRung focuses on the harder signal: framing, verification, iteration, responsible agent oversight, and the trajectory shown across real project work.
Candidates get private coaching first. Sharing is selective, evidence-backed, and designed around skill episodes rather than raw telemetry dumps.
For institutions
Help cohorts build stronger AI-native habits without turning student work into surveillance.
For employers
See concise evidence cards for judgement, verification, shipping, and improvement over time.
Phase 0 focus
Collector, coach, profile
FirstRung starts with a privacy-first local collector, a candidate-facing coaching layer, and a shareable evidence profile that a candidate controls.
Public core repository- Skill episodes from selected project work
- Verification signals from tests, checks, and reviews
- Candidate reflections on what changed and what could break
- Evidence tiers that avoid a single opaque score
Trust boundary
Open where trust needs inspection
Raw code, prompts, diffs, logs, and private project material stay local by default.
The open trust layer keeps collection, schemas, redaction, and basic rules inspectable.
The hosted passport only receives candidate-approved derived evidence and selected profile content.
Early partners