Protect the thesis after you buy
You paste this in as a decision
“Long NVDA through 2027. The CUDA ecosystem is a durable moat, and competitors stay roughly two years behind on training performance.”
- Competitors stay ~2 years behind on training workloads
- Data-center gross margins hold above 70%
- Hyperscaler capex keeps growing through 2026
Vigil extracts these — you confirm, edit, or add before anything goes under watch.
You point Vigil at
- The company's SEC filings index
- The quarterly earnings transcripts page
- Competitor newsrooms and benchmark results
The scheduled check
A competitor posts benchmarks at parity, six months ahead of your model. Vigil names the assumption it breaks — with the sources cited. You read it the next morning, not in a post-mortem two quarters later.
Where you see it
On your Overview, on the decision's page with the evidence attached, and in your weekly brief.