Coverage matrix — stack layer × generation
Every US insurance stack layer crossed against the modern SaaS and AI-native generations. A cell with three or more tools is considered covered. A cell with one or two is thin. A cell with none is a research backlog item.
| Stack layer | Modern SaaS | AI-native | Row total |
|---|---|---|---|
| Policy admin | 14 | 0 | 14 |
| Rating | 1 | 2 | 3 |
| Underwriting workstation | 3 | 28 | 31 |
| Claims admin | 6 | 3 | 9 |
| FNOL intake | 1 | 1 | 2 |
| Damage estimation | 5 | 1 | 6 |
| Document processing | 2 | 6 | 8 |
| Fraud detection | 4 | 1 | 5 |
| Conversational AI | 0 | 3 | 3 |
| Data platform | 4 | 7 | 11 |
| BI / analytics | 0 | 0 | 0 |
| Actuarial | 6 | 1 | 7 |
| Reinsurance | 1 | 6 | 7 |
| Compliance / regulatory | 3 | 1 | 4 |
| CRM / distribution | 10 | 5 | 15 |
| Risk imagery | 2 | 6 | 8 |
Where the gaps are
The cells below are under the minimum coverage of 3 tools. They are the priority queue for new fiches. Empty cells rank higher than thin cells; within each tier, order is arbitrary.
- Policy admin×AI-nativeempty0
- Conversational AI×Modern SaaSempty0
- BI / analytics×Modern SaaSempty0
- BI / analytics×AI-nativeempty0
- Rating×Modern SaaSthin1
- FNOL intake×Modern SaaSthin1
- FNOL intake×AI-nativethin1
- Damage estimation×AI-nativethin1
- Fraud detection×AI-nativethin1
- Actuarial×AI-nativethin1
- Reinsurance×Modern SaaSthin1
- Compliance / regulatory×AI-nativethin1
- Rating×AI-nativethin2
- Document processing×Modern SaaSthin2
- Risk imagery×Modern SaaSthin2
Method
Each tool in the public registry is tagged with exactly one stack layer and exactly one generation. The layer is chosen from the sixteen canonical layers of the US insurance operational stack. The generation reflects the tool’s centre of gravity rather than its newest feature.
A cell is covered when three or more tools sit in it. The threshold is arbitrary but kept constant across the site, so that coverage statements stay comparable as the library grows. Thin and empty cells are published explicitly so readers can see what Phidea does not yet map.
Counts update on the next build whenever a new tool is added or an existing fiche’s tags are revised.
Why no Legacy column.Legacy insurance technology is rarely a purchasable SaaS — most of it lives on in-house mainframe systems, carrier-built portals, or COBOL-era vendor stacks that predate the cloud. Phidea’s library maps purchasable software, so the coverage matrix is scoped to the two generations where a coverage count is meaningful. The full generational ladder (legacy → modern → AI-native) is discussed in the classification essay.