Snapshot product update: per-model verdicts and alternatives drift
What shipped this month, what changed about the deliverable, and what is coming next for teams running their first Snapshot.
Two changes shipped in the last sprint that should make the Snapshot more useful for teams trying to act on the findings quickly.
Per-model verdicts
Every prompt is now sampled against four models in parallel. You see the verdict for each one, not a single blended score, so you can prioritize fixes by where your customer actually prompts.
Alternatives drift
We now track which alternatives appear across runs and flag the ones gaining ground. If a competitor starts showing up in three more prompts than last month, you will see it in the report.
What is next
Team-led readouts are moving from optional to default for every Snapshot. The goal is to make sure no team ends a report wondering what to ship first.