AI Atlas helps founders find builders who can ship with AI across messy real-world work. Public showcases are the supporting layer: they help the network get discovered and give buyers examples of what is possible.
A founder brief captures what you are trying to build, automate, or explore, plus the current workflow, tools, constraints, urgency, and disqualifiers.
The brief is matched to builders whose experience, public work, and review notes suggest relevant judgment, implementation complexity, and privacy boundaries.
A good intro explains why the builder fits, which public showcases may be useful context, and what risks or missing context should be resolved first.
A public showcase works when it helps people understand what a builder can make without turning the network into a generic content directory. In practice that means at least one of the following is published or sanitized for inspection:
Polish is welcome but not sufficient. A landing page, a logo, or a demo video on its own does not clear the bar. Privacy-sensitive material is handled via sanitized screenshots, synthetic records, redacted traces, or aggregate metrics.
The number next to the diamond on every showcase card is the showcase signal. It combines curator scoring with anonymous visitor voting. It rises when the entry has:
Higher is better, but a high score is not a blanket endorsement of the underlying business or the builder. It reflects how much of the work is open to inspection.
Featured. Full case study with public artifacts, metrics, and a privacy review. Treat as the strongest public example.
Supporting. A real workflow with at least one inspectable artifact, but less complete buyer context than a featured showcase.
Private · sanitized. The underlying system is not public, but a sanitized version is published with explicit privacy notes.
Held out of discovery. Leads with missing artifacts, unclear ownership, unsupported claims, or unresolved privacy boundaries are not used for buyer matching.
A submission or public mention is captured with a source URL and short observation. Nothing becomes a buyer-facing claim until evidence is checked.
Each claim must be visible in a public source. Screenshots, traces, and datasets are checked for sensitive content, sanitized where required, and rejected if they cannot be safely published.
Accepted entries are added to public data and render as a builder profile and showcase page. The builder is credited and can request edits or removal at any time.
Submit workflow evidence via the submission page. Review is currently run by hand; turnaround is days to weeks, not minutes.