# Build vs Buy AI Agent Scorecard

Canonical: https://pharosproduction.github.io/build-vs-buy-ai-agent-scorecard/
Methodology: https://pharosproduction.github.io/build-vs-buy-ai-agent-scorecard/methodology.html
Published: 2026-08-25
Decision model version: 1.0.0
Export schema version: 1.0.0

## Direct answer

Build when differentiated behavior, controlled data and deep integration justify durable internal ownership. Buy when the capability is standard and speed matters more than control. Use hybrid when commodity infrastructure can be bought while proprietary policy, adapters and evaluation remain owned.

## Model contract

- 12 criteria with total signed weight 31.
- Buy answer = -1, hybrid = 0, build = +1, unknown = null.
- Decision index = round-half-away-from-zero(100 x signed known points / known weight).
- BUY at -30 or lower, BUILD at +30 or higher, HYBRID between the thresholds.
- Minimum 75% known-weight coverage plus known data control, team capability, lifecycle ownership and compliance path.
- A result within seven index points of a threshold carries a boundary warning.
- The index is decision support, not a probability, confidence score, benchmark or forecast.

## Open data

- /data/criteria.json
- /data/tco.json
- /data/methodology.json
- /data/claims.json
- /data/sources.json
- /data/profiles.json

## Citation boundary

The model's weights, thresholds and bundled dollar values are authored defaults. External sources support factor relevance and lifecycle or TCO concepts; they do not validate this model's numerical behavior.