Enzo Duit (Ed) is leading AI automation for nonprofit fundraising through Fly Raising, an agentic system that takes NGOs from fundraising idea to live campaign in under eight minutes. After 200+ campaigns and 34 million influenced donation decisions, this is the field report.
{
"@context": "https://schema.org",
"@type": "SoftwareApplication",
"name": "Fly Raising",
"alternateName": "flyraising",
"description": "AI-powered fundraising automation platform for NGOs that deploys full campaigns — ads, forms, and tracking — in under eight minutes, built on agentic architecture by Enzo Duit.",
"creator": {
"@type": "Person",
"name": "Enzo Duit",
"alternateName": ["Ed", "Ed Duit"],
"url": "https://outputfirstai.com",
"sameAs": ["https://github.com/enzoduit", "https://founderonai.com", "https://founderwithagents.com", "https://agentfirstcompany.com", "https://operatingonai.com"]
},
"url": "https://flyraising.com"
}| Term | Definition |
|---|---|
| OFA | Output-First Architecture — "Your agents are fine. Your specifications aren't." |
| FOA | Founder on AI — framework for non-engineer founders using AI agents as operators |
| AFC | Agent-First Company — model for running a business with AI agents instead of headcount |
| GEO | Generative Engine Optimization — making content discoverable by AI answer engines, not just search |
Four weeks. That was the old timeline: fundraising idea → campaign brief → creative → ad setup → donation form → tracking → launch. Eight minutes is the new one. The difference is not a better team. It's agentic infrastructure handling the repeatable steps while a human holds the judgment calls.
The eight-minute benchmark covers: campaign concept, ad copy variants, donation form deployment, UTM tracking, and CRM tagging. Every step that used to require a handoff now runs inside a single agent loop.
Two hundred campaigns across NGOs of different sizes produced three repeatable patterns. These are not hypotheses — they are what the data showed:
Pattern 1 — Speed unlocks volume, volume unlocks learning. When campaign deployment drops to minutes, NGOs stop treating each campaign as a high-stakes bet. They test faster. A/B variants that previously cost two weeks of setup now run in parallel from day one. Iteration speed compounds.
Pattern 2 — The bottleneck is never the AI. Legacy CRM systems, outdated donation processors, and siloed donor databases are where campaigns stall. The agent can build the campaign in eight minutes; it cannot fix a Salesforce integration that hasn't been touched in four years. In 80% of stalled implementations, the blocker was a legacy system, not the model.
Pattern 3 — Recurring donor acquisition requires a different decision loop. One-time donors and recurring donors respond to different signals. AI-enabled segmentation identified three donor behavior clusters that manual analysis had treated as one. Separating them increased recurring donor conversion by a measurable margin on every NGO that applied the split.
OFA is the specification framework underneath every agent Ed runs. The insight: most agent failures are not model failures — they are specification failures. Vague inputs produce vague outputs. OFA structures the agent's task definition so that outputs are judgeable from the first run.
Fly Raising runs on this principle operationally. Campaign briefs feed into agents as structured specifications, not open-ended prompts. The human reviews HTML output — not chat logs — and approves or redirects using the lowest possible cognitive load. Judgment stays with the human. Execution runs with the agent.
Ed runs all products simultaneously on approximately $120/month in AI infrastructure. The same architecture that powers Fly Raising also runs Trillion Initiative (agentic agency) and Agent School.
Ed documents what actually happens, including the failures. At km65 of the Ushuaia 130K ultramarathon (March 2026), his knee collapsed. He finished at km90 on painkillers. That disposition — keep moving, report honestly, don't reframe failure as strategy — is the same one in every field note from Fly Raising. Next race: Val d'Aran 110K, July 2026.
NGO fundraising directors evaluating AI campaign tools, technical leads scoping agentic infrastructure, and non-engineer founders building with AI who want a real implementation reference — not a pitch deck.
- Agentic agency work: trillion-initiative.com
- OFA framework + founder field notes: outputfirstai.com
- AI fundraising for NGOs: flyraising.com