Subject John F. Kim
Executive Director,
Collective Impact Labs
Current Location Campbell, California
Sheet revised 2026-09-25
Progress declared by the author. Not audited,
not peer-reviewed.

John F. Kim

I work on the infrastructure that lets independent organizations move together.

Eight people in a room agreed on what needed to happen. They came from different contexts, but aligned around a shared mission. Six months later, nothing had materialized. Not due to laziness or disagreement — there was clarity, there was will, but there was no mechanism. The gap is almost never intelligence or goodwill. Everything below is an attempt at the mechanism, recorded with what is actually true about each one.

Show 9 rows · 9 declared limits

The sheet

Live · 501(c)(3)

Collective Impact Labs — shared infrastructure for mission-aligned ecosystems

Small team, real output. The structural problem is the one the whole sheet is about: backbone work is the least fundable category in philanthropy. Funders fund programs. Almost nobody funds the connective tissue — which is why it usually gets done off the side of someone's desk until that person burns out.

Honest limitGeneral operating support is harder to justify to a board than a named program, and I know I'm asking for the harder thing.

Progress 4/5 — operating,
not proven at scale
What it needs
  • Unrestricted operating support. Multi-year beats larger-but-once.
  • A funder who'll argue with me about it. Twenty minutes on what backbone funding would have to look like to clear your committee.
  • Board and advisory candidates with nonprofit finance or governance depth who'll ask uncomfortable questions.
Start there
Live · maintained map

Field Atlas — a working map of the faith-and-technology field, and where it overlaps

It exists so funders stop paying three teams to solve one problem separately, and so the people building the eleventh version of something can find the other ten. A map is only as good as its corrections. The most useful thing anyone does for it is tell me where it's wrong about them.

Honest limitThe dataset isn't openly published yet. Some organizations in it work in places where being listed carries real risk, and that has to be fully resolved before anything opens. Walkthroughs happen live, with me, for that reason.

Progress 4/5 — working,
not yet publishable
What it needs
  • Corrections about your own organization. Five minutes. Genuinely the best contribution available.
  • A long-term steward. The structural risk is key-person concentration.
  • Data-protection counsel on lawful basis and on records relating to people in sensitive contexts. This gates publication.
Ask for a walkthrough
Research · mechanism study

Consent-aware coordination — a self-tested model of the signal layer

As more coordination gets executed by software agents rather than people, what any given party actually consented to stops being a footnote in the terms and becomes the whole design problem — consent, delegation, revocation, and who is allowed to speak for whom. It's deterministic, runs in about twelve seconds, and surfaces a residual risk I have not designed my way out of.

Honest limitA working, self-tested agent-based model of the signal layer — a mechanism study. Empirical field validation is future work. Anyone describing work like this as validated is rounding up.

Progress 2/5 — mechanism only,
no field validation
What it needs
  • Someone to tell me where it breaks. A paragraph is more useful to me than a call.
  • An academic co-author in multi-agent systems, mechanism design or governance who'd co-write rather than endorse. I don't have a research CV and that's a real gap.
  • Research funding — modest and specific: a methodologist and real agent experiments rather than more simulation.
Ask for the write-up
Prototype · end-to-end demo

OpenBook — an archive that answers, without giving itself away

Decades of careful work that people used when they knew where to look. Now most of them ask a chatbot instead. The obvious fix is to hand the corpus to someone with a large model; that costs you the thing, the attribution, and eventually the reason anyone needs you. This is the other path: the archive becomes answerable while the people who made it keep it, keep the credit, and keep the ability to say no.

Honest limitIt's a prototype, not a product. The commercial model is genuinely unsettled and I won't pretend otherwise to get a pilot.

Progress 3/5 — works end to end,
model unsettled
What it needs
  • Two or three archive holders willing to test. A publisher, seminary, denominational archive or research library.
  • Rights and licensing counsel who has actually negotiated corpus licensing and can tell me which assumptions are naive.
  • An engineer who has shipped retrieval at scale — a few hours of review, not a job.
Send a document
Prototype · rubric v0.9

Mission Collective — six dimensions, thirty-one indicators, no score

Each indicator is anchored to a written rubric designed so two people scoring the same network should land in roughly the same place. It returns a profile, never a grade — there is deliberately no single composite score, because one number lets everyone skip the conversation the profile is for. What it needs is not funding. It needs reality.

Honest limitThe rubric is version 0.9 and unvalidated. Anyone who uses it now is helping me test it as much as the other way around, and I say that up front every time.

Progress 2/5 — v0.9,
inter-rater untested
What it needs
  • Networks willing to be read. Forty-five minutes, free, one-page profile back. Two at a time, then it queues.
  • A second rater to score the same networks independently. The single highest-value contribution anyone could make here.
  • A measurement methodologist to tell me where the construct validity is weak.
Take one of the two
Research · measurement method

Edge City India — testing the rubric on people who owe me nothing

I'll be at Edge City India this fall, and while I'm there I'm hoping to take part in a small, unfunded research collaboration on measurement method: does the network-health rubric hold up when the people scoring it have no relationship to me? That is the honest test the diagnostic has never had. Everything I've measured so far has been inside my own network.

Honest limitA research collaboration, not a program and not a convening: no registration, no seats, nothing to sponsor. The terms aren't settled yet. If they aren't settled before I arrive, I attend as a participant and this row comes down.

Progress 1/5 — proposed,
not yet confirmed
What it needs
  • Independent raters on the ground. People with no relationship to me who'll score a network against the written rubric, so agreement gets measured honestly.
  • A conversation if you'll be there. Especially if you work on coordination or measurement in a context different from mine. Disagreement is the useful input.
  • A return path. Anything scored goes back first to the people it describes.
Say you'll be there
Prototype · pre-announcement

An open-licensed library of AI capabilities for nonprofits and ministries

The practical things a small organization needs and cannot afford to develop, released so that the tenth organization doesn't have to rebuild what the first one did. The reason it's a commons rather than a product is the whole point: every one of these organizations is small, and the economics of each of them separately commissioning the same capability are indefensible.

Honest limitThe arrangement behind it is under a mutual NDA and pre-announcement, so I'm deliberately vague about who is involved until they say otherwise. If that vagueness is a problem for you, wait for the announcement rather than taking my word for it.

Progress 3/5 — first wave
being prepared
What it needs
  • Pilot organizations willing to use an early capability and say honestly whether it helped or wasted their afternoon.
  • Technical reviewers who'll read what's being released with an adversarial eye before it goes out under an open license.
  • Open-source licensing counsel. Getting the terms right at the start is much cheaper than fixing them after adoption.
Put your hand up
Live · own name, not CIL

Consulting — ecosystem design and capital architecture

I help funders, networks and the organizations between them figure out how to actually coordinate, and how to pay for it. Four areas: coordination design, capital architecture and funding strategy, early-stage advising, and AI readiness and governance. Fees depend on scope, and I hold a lower rate for mission organizations. The full list is on the consulting page.

Honest limitIt's one person with a hard time ceiling. Network-health diagnostics run through Collective Impact Labs, not this practice. I don't take fractional-executive roles, give investment advice, broker capital, or do anything that would make me a finder.

Progress 3/5 — open,
no engagement yet
How this starts
  • Two paragraphs. What is happening, and what you want to be different.
  • A yes or a no within 72 hours, with a date. No calendar widget, no form, no discovery sequence.
  • A fixed term, renewed in writing. Never open-ended.
Send the two paragraphs
Live · cottage food

Love Your Neighbor Baking Co. — sold in pairs, on purpose

Everything else on this sheet is infrastructure, which is invisible by design. This part you can hold. The motto is be good to one another, and when you can — be sweet. Everything is sold in pairs: you keep one, and you give one to somebody with something worth saying attached to it. The pastry is the excuse; the sentence is the product. The longer plan is workforce development — baking is a real trade, and a kitchen is a good place to come back to work in after a transitional or rehab program.

Honest limitIt's a cottage-food operation, not a storefront. Capacity is my actual kitchen.

Progress 5/5 — operating at
kitchen capacity
What it needs
  • A kitchen, and a hub. If you run a community hub, a rehab, or a transitional program and you've ever thought a bakery could anchor it — that's a conversation I'm actively looking for.
Say hello
Off the sheet

I cofounded Christians in Web3 and sit on its board. I write occasionally, at no particular cadence: The Glue Is the Job and Why I'm Writing About Coordination.

I'm a Christian. Most of what I work on serves churches and mission organizations, and that isn't incidental to why I do it. I don't think you have to share that for the maps to be useful to you.

I read everything. I'm slow to reply — I'd rather tell you that than pretend otherwise. If it's time-sensitive, say so in the subject line and I'll move it up.

john@collectiveimpactlabs.org