AI Landscape Radar · how it works
How changes are reviewed
How a change in the AI-agent landscape becomes a reviewed, versioned methodology update. Seven steps, every one human-approval-gated.
1. Source reviewed
An approved public source is reviewed for material changes.
2. Signal extraction
The fetch is reduced to one or more candidate signals — short, structured records describing what changed.
3. Status assignment
Each signal lands in 'Watching', 'Under review', 'Approved', or 'Affects reports'.
4. Proposal
Approved signals become proposals — concrete suggestions for methodology / scoring / question changes.
5. Human approval gate
No proposal modifies the methodology until a human reviewer approves it. Review happens on an internal, founder-only console — nothing changes your score or methodology without that human approval.
6. Publication as a new methodology version
Approved changes are published as a new, versioned methodology release. Old reports continue to reference their original methodology version.
7. Reassessment trigger
When a published change affects an existing report's control, the customer sees a 'Reassessment recommended' notice with a link to review and re-run when they choose.
Review happens on an internal, founder-only console, so a human always signs off before any change reaches your report.