1 · Grants

The Working Model

The home page is the pitch. This is what sits underneath it: what you choose to collect, the two standards of rigor that hold a grant application between them, and the five-step lifecycle that turns one round into a record the next round builds on.

Narrative, Instrument, or Both

What You Collect Is a Choice

The home page shows the contrast. Here is the decision behind it: which to use, what each is good for, and what each costs you. It used to be settled quietly, and you found out which you collected only after the decisions were made. Now it is an explicit choice, made for each thing you collect.

As a Story

What You Collect
the project in its own words
Best For
context, motivation, judgment calls
The Reviewer's Job
weigh the story
The Tradeoff
rich, but cannot be compared or shown wrong

As an Instrument

What You Collect
comparable, checkable data
Best For
comparison, verification, tracking across rounds
The Reviewer's Job
check each claim against named evidence
The Tradeoff
comparable and verifiable, but each measure has to be defined up front

Most rounds want both. What is new is that the choice is explicit, made for each thing you collect, so you get a story where you want a story and data where you need data, instead of finding out which you got after the fact.

The Rigor Sandwich

Two Standards of Rigor, One Grants Application Between Them

and are cross-domain: they bring the same rigor in any domain. and are what that rigor looks like applied to grants.

CRAFT

the meta-standard · cross-domain

The cross-domain rigor on whether a domain's data can be checked at all. Detailed on Cross-Domain →

the domain-specific filling · swappablenow: grants

CROSS

grants · normative

Common Reporting Outcome Standards Schema. What a grant round must commit to before applications open: , , indicator fields, organizational identity, concurrent-funding disclosure, public benefit mechanism. CRAFT applied to grants. Read CROSS →

GRAIN

grants · abstraction layer

Grant Representation And Interchange Notation. The abstraction layer beneath the grants standards: it represents what a grant program is in one common notation, so everything above it, CROSS included, reads the same description the same way. Its building blocks were not invented; each was drawn from real funding programs and kept only if it appeared in more than one independent source. A grant comes apart into these pieces and rebuilds from them. Descriptive, not normative. Read GRAIN →

WALKRI

field and instrument quality · cross-domain

The cross-domain rigor on whether a field produces data or noise. Detailed on Cross-Domain →

The bread never changes; only the filling does. CRAFT and WALKRI bring the same rigor in any domain. Whatever domain you work in slots into the middle. Grants is simply the first filling built; swap it out and the bread still holds.

CLEAR

The Round Lifecycle

Five steps that close the loop instead of resetting it each round. The shape never changes; what a program declares at each step is what differs.

C

Commit

Publicly commit to the obligation architecture before any application opens. The public commitment is what everything afterward is checked against. Nothing moves until this is written and published.

L

List

List the specification publicly and collect structured applications against it. The program lists its obligations; applicants list their projects. WALKRI-conformant intake runs at this stage.

E

Evaluate

Assess applications against the committed specification. Reviewers work from declared criteria, not impression. Verdicts are instrument-grade.

A

Attest

Verify delivery against milestone commitments. Evidence submitted, attestation entries written, determinations on record.

R

Register

Formally register outcomes as Cohort Position. The institutional record that feeds the next round's Commit stage. This is what closes the loop instead of resetting.

The Same Word, Different Programs

The five steps never change. What each program declares at each step is what differs, the decision step (the A) most of all.

Prospective grant

decides by verifying what was delivered

C
Commit
L
List
E
Evaluate
A
Attest
R
Register

Quadratic funding

decides by splitting a pool by formula

C
Commit
L
Launch
E
Evaluate
A
Allocate
R
Register

Retroactive program

decides by rewarding work already done

C
Commit
L
Log
E
Evaluate
A
Award
R
Reconcile

Funding lottery

decides by random draw among the eligible

C
Commit
L
List
E
Evaluate
A
Allot
R
Register

Theory of Change

Built Into All Four Standards, Not Bolted On

Most programs treat a theory of change as a document you attach. Here it is part of the machine, and each standard carries a piece of it.

GRAIN declares the pathway

The numbered causal claim from the work to the change, the assumptions it rests on, and the level it aims at, from an immediate output up to a long-term goal.

CROSS holds a grant only to the level it controls

A single grant can show a near-term outcome; a long-term goal is an argued contribution, never pinned on one project, and each gate asks for the kind of evidence that level can really bear.

WALKRI turns an outcome into a measure

Not the word "impact," but what changed, for whom, by how much, over what time, and against what baseline.

CRAFT feeds each cycle's result back

Its sixth condition, feedback with propagation, carries the result back into the theory, so a wrong assumption gets corrected instead of quietly repeated.

Declare the pathway, test it against what actually happened, revise. That loop is the theory of change doing work, not sitting in a folder.

The Data Layer

Built for What Collects It Next

The standards sit in front of whatever collects the data and behind it, never in the collection itself. They set what has to be captured before anything is gathered, and hold the result to a standard after. The collection method in the middle is swappable: a form today, an agent reading documents or a data feed tomorrow.

Because every field is an instrument in one shared representation, what comes out is machine and AI readable by construction. That is what lets a tool, or an AI, triage applications, compare rounds, or assist a review, working from the data instead of guessing at prose. The round you run now is ready for tools that do not exist yet.

Where This Comes From

Read Across Every Tradition at Once

None of this began as a design. It began as an analysis of where grant funding breaks down, read across every tradition at once instead of staying inside one.

200+
programs, rounds, and instruments analyzed
86
funding and evaluation frameworks
1,100+
cited sources
web3 roundsphilanthropic foundationsgovernment and multilateral fundersimpact investinghumanitarian and evaluation bodieslegal and tax frameworks

A pattern that shows up in a foundation, a development bank, and an on-chain round is structural, not a matter of taste. It is what the grants standards are made from.

Compatibility

Works Alongside Any Program, Platform, or Sector

Every sector that runs grant programs faces the same intake and reporting problems:

international developmentpublic healthclimatecivic techarts and cultureWeb3

These are not Web3 standards and not a single-platform integration. They are a common data layer any grants program can adopt without changing its existing tools.

What It Opens

Worlds That Could Not Read Each Other Now Can

Compatible with everything is not a feature, it is the effect. The same grant, described once, is legible on both sides of lines that used to block it.

Web3 Grants

Gitcoin · Optimism · Octant · DAOs

CROSS+WALKRI

one common layer

The Larger Grants World

foundations · government · multilateral · impact investing

is the crossing: it makes the bridge possible, not the funding automatic.

Comparable across the whole fieldAccess for small and new programsMachine and AI readableHarder to inflate or fakeTools and analysis built on top

Everything here is CROSS and GRAIN, the grants filling. The rigor they inherit comes from CRAFT and WALKRI, two standards that know nothing about grants and reach far past them. That is where this goes next.