<aside> 🧭

Grantmaking OS is a lightweight operating system for moving from a broad funding area to a concrete, defensible action: map the domain → find bottlenecks → generate interventions → find people/resources → execute → learn.

</aside>

Start here

1. Add one bounded area to Domains.
2. Collect evidence and conversations in Sources.
3. Write plausible constraints in Bottlenecks before deciding which are real.
4. Generate several candidate responses in Interventions.
5. Track relevant experts, operators, funders, and organizations in People & Organizations.
6. Turn the best next moves into Tasks.
7. Record important calls, forecasts, outcomes, mistakes, and process improvements in Decisions & Learning.

<aside> 💡

The databases are the source of truth. Notes and dashboards should synthesize them rather than creating duplicate records.

</aside>

Grantmaking model

0. Choose a bounded domain

If no higher-up has scoped the area, do a lightweight prioritization pass. Compare rough importance, neglectedness, leverage, actionability, information value, and your access/fit. Scope enough to start learning; do not wait for perfect cause prioritization.

1. Map & prioritize bottlenecks

Build a causal model of how the desired outcome happens. Gather evidence, add plausible bottlenecks to one registry, shallow-check them quickly, and deep-dive only the strongest candidates. Keep written reasoning and confidence rather than relying only on scores.

2. Generate & evaluate interventions

Generate multiple interventions before committing. Look for existing projects first. Compare expected impact, tractability, cost/resources, speed, downside risk, counterfactual value, opportunity cost, and uncertainty. End with a concrete recommendation.

3. Find & coordinate people and resources

Treat your network as both an information system and an execution system. Match problems ↔ people ↔ resources; share diligence; refer opportunities; recruit operators; coordinate and co-fund where useful.

4. Improve the system