The AI Funding Playbook

Funding advice usually stops at "apply for more grants." That advice produces burnout, scattered outreach, and reactive cycles — because applying is the last step of a system nobody built.

Non-dilutive capital goes unclaimed at scale. A seed-stage startup trades meaningful equity for its first institutional round; federal R&D programs will fund comparable work and take none — NSF states plainly that it takes no equity, and awardees keep full ownership of their company and intellectual property.

On the nonprofit side the money is real and moving: in a single 2025 announcement, one foundation committed $75.8 million across 149 grants in 13 countries. For many major funders, application quality is only part of the equation — visibility, relationships and institutional fit determine whether you reach the application stage at all.

Each layer feeds the next. Skip one and the layers below it produce motion without money.

1. 01 Capital Map

   Establish which of five funding tiers you actually qualify for today.

2. 02 AI Systems Layer

   Fit mapping, funder-language analysis, adversarial review, compliance checking, reporting.

3. 03 Outreach Engine

   Program-officer contact, pitch sequencing, visibility-building for invitation-only funders.

4. 04 90-Day Build

   Registration, mapping, contact, one properly built application, then stacking.

5. 05 Go / No-Go Scorecard

   Eight checks run before every submission.

Money is not one pool. It arrives in five tiers, and the strategy at each tier is completely different.

Tier One — $50K–$150K · Entry points & credits

Small federal awards, low-end foundation grants (published awards in the 2025 portfolio started at $50,000 and $52,500), state portals such as grants.ca.gov, and cloud/compute credits. Strategy: volume and speed. Win something small, then cite it everywhere.

Tier Two — $150K–$325K · SBIR & STTR Phase I

America's Seed Fund is coordinated by the SBA across eleven agencies and is equity-free. As of April 2026, agencies may issue a Phase I award up to $323,090 without SBA approval. Under NSF solicitation 26-510, Phase I proposals may run up to $305,000 over six to eighteen months, inclusive of all costs.

NSF states historical Phase I funding rates between 10% and 20%, with panel review one to three months after the deadline and notification at five to seven months. The program funds high-risk R&D, not incremental product development.

Tier Three — up to $500K/year · The credit that requires winning nothing

The PATH Act enacted IRC §41(h) and §3111(f), letting a qualified small business apply part of its research credit against payroll taxes. The Inflation Reduction Act raised the annual maximum from $250,000 to $500,000 for tax years beginning after December 31, 2022. The credit first reduces the employer's social security tax up to $250,000 per quarter, then Medicare tax; the remainder carries forward.

Tier Four — $500K–$2.15M · Phase II & institutional philanthropy

Track A: as of April 2026 the general Phase II ceiling is $2,153,927. NSF Phase II runs to $1,250,000, typically over 24 months, for Phase I awardees filing six to twenty-four months after the Phase I start. Phase IIB adds $50,000–$500,000 matched against investor or customer funds; TECP supports up to 20% of the Phase II award.

Track B: the Patrick J. McGovern Foundation announced $75.8 million across 149 grants and roughly $500 million over the past decade. Published awards ran from $50,000 to $1.25 million, with the bulk between $200,000 and $750,000. This portfolio is not built from cold applications — it is built from organizations already visible in the field. Budget six to eighteen months for access.

Tier Five — $1.55M–$30M · The ceiling

NSF Fast-Track combines both phases: $400,000 for the Phase I component plus up to $1,155,000 for Phase II, by invitation, with the full proposal due within four months. Strategic Breakthrough awards — created by the reauthorization — are invitation-only, for existing Phase II awardees, up to $30,000,000. That is a relationship instrument, not an application.

NSF relaunched its SBIR/STTR programs following the Small Business Innovation and Economic Security Act. Solicitation 26-510, posted May 2026, replaces the 2024-era solicitations entirely.

• A new $30M Strategic Breakthrough category exists, by invitation, for Phase II awardees.
• Submission volume is capped: one proposal per organization at a time, one PI per proposal.
• Voluntary committed cost sharing is prohibited — do not volunteer match.
• The PI's primary employment must be with the small business (at least 51%); outside work over 19.6 hours per week conflicts.

A prompt is not a system. The hour figures below are workflow estimates for the manual version of each stage — the shape of the saving, not a promise of it.

• Stage 1 — Fit mapping (~15h). Write a one-page technical narrative: the uncertainty you are resolving, the method, and why it might fail. Map it against agency mission language for a ranked shortlist with stated reasons.
• Stage 2 — Funder language analysis (~10h). Pull published awards for your target program and extract the distribution, vocabulary, and shape of a fundable project.
• Stage 3 — Drafting against criteria (~10h). NSF evaluates on Intellectual Merit, Broader Impacts and Commercial Impact. Draft section by section, then run an adversarial pass — the rejection list is worth more than the draft.
• Stage 4 — Compliance (~5h). Only compliant proposals reach reviewers. Convert the solicitation into a literal checklist and verify line by line.
• Stage 5 — Reporting. Build the reporting workflow before you win, and one award becomes a renewable relationship.

At NSF you cannot reach a full proposal without a Project Pitch and an official invitation. The conversation is the gate. If declined, a new pitch must be submitted and invited, and a resubmission must be a significant revision with a written change description.

NSF invites declined applicants to email the Program Director for a debrief. A declined proposal with a debrief is a roadmap; without one it is just a loss.

For invitation-only funders, access comes from visibility: published work, conference presence, and partnership with existing grantees. Read the published grant list, find the three grantees nearest your work, and build a legitimate reason to be in contact.

Automate: opportunity monitoring, deadline alerts, CRM hygiene, follow-up sequencing, research-backed first drafts, reporting reminders. Never automate: the Project Pitch, the first message to a program officer, anything to an invitation-only foundation, or any claim about your own capability.

• Days 1–14 — Infrastructure. SAM.gov and UEI, Grants.gov, SBIR.gov, state portal. Nonprofits: confirm 501(c)(3) documentation and clean budget figures.
• Days 15–30 — Map and qualify. Write the one-page narrative, run fit mapping, produce a shortlist of no more than eight targets across at least two tiers.
• Days 31–45 — Contact and pitch. Submit the NSF Project Pitch, email program officers, open contact with two existing grantees of invitation-only funders.
• Days 46–75 — Build one application properly. One. Draft against the criteria, run the adversarial pass, run compliance line by line, get a human review.
• Days 76–90 — Stack. Take the R&D credit position to your CPA, set monitoring for the next cycle, layer state incentives with federal awards.

Run this before every submission. Below five, fix the infrastructure instead of writing.

• SAM.gov registered with an active UEI
• Technical narrative written, specific, and one page
• Genuine technical uncertainty — not incremental engineering
• Target program identified, with a named program officer
• Pitch submitted or contact made well before the deadline
• Eligibility confirmed against the solicitation, not a summary
• Budget defensible line by line, with no unrequested cost share
• Reporting capacity exists if you win

Every figure was verified against the agency, the funder, or the statute: NSF Program Solicitation 26-510; the Patrick J. McGovern Foundation 2025 grant announcement; SBA America's Seed Fund; NSF SBIR/STTR applicant guidance; IRC §41(h)/§3111(f) and IRS IR-2025-99. Confirm against the controlling document before you submit.