Fundraising is a tax on high-impact work

I used an LLM to help draft this post and it likely contains >10% AI-generated text, but I’ve edited/rewritten it extensively and endorse it.

Access to capital, whether a venture cheque or a grant, runs through closed warm networks. The people doing high-impact work outside those networks stay invisible to the funders who would back them, and the work of reaching funders eats the time of the people we most want building. IRIS is the relationship-intelligence layer that removes that tax. The full theory of change is published as a structured argument map, linked below.

The problem

The knowledge problem is largely solved. A founder or nonprofit leader can build a pitch, get structured feedback, and draft a plan in an afternoon. What still decides outcomes is access: finding the right funders, earning a warm introduction, and getting from a first conversation to a closed round or grant.

That work is a tax. Every hour spent raising money is an hour not spent on the work the money is meant to fund, and it is regressive. The people without an existing network spend the longest to raise the least, so the tax falls hardest on the work that is most neglected. Funders face the mirror image, receiving far more inbound than they can evaluate and backing a small fraction of it. Both sides lose to the same fact: capital, whether a cheque or a grant, is gated by who you can reach. This is not a fairness complaint dressed up as an opportunity. It is a resource-allocation failure in one of the most consequential allocation systems we have.

Why it is neglected

Surely someone owns this already. They do not. Professional networks map who people are. CRMs store who an organisation already knows. Neither discovers new high-quality relationships, matches a builder to the right funder, or reduces the noise in inbound. The layer that would sit between the people doing the work and the people funding it does not exist as infrastructure. It exists informally, inside the inboxes of a small number of well-connected people, which is exactly why access is gated. The gap holds whether the capital is venture or philanthropic: one matching failure with two faces.

This is not a new observation here, which is part of why it is worth solving properly. The warm-introduction bottleneck has been named on this forum directly, and the stronger point, that capital is bottlenecked less on money than on the few people who can find and vet opportunities, has been made about grantmaking specifically (both linked in Further reading). Where IRIS goes further is in three places: it does the work end to end rather than only screening, it closes the loop by learning from real funded outcomes rather than from proposals alone, and its neutrality is structural rather than dependent on one well-connected person's goodwill.

What IRIS does

IRIS is a voice first, agentic AI relationship-intelligence system and fundraising agent. It does three things. It makes excluded builders discoverable, matching them to funders on thesis, stage and sector and surfacing those who never reach the usual pipelines. It carries the fundraising work itself, the coaching, targeting, scheduling and follow-up, so founders only meet funders who are active, relevant and ready, and get their time back. And it logs every introduction and outcome, learning from what actually closes rather than from generic data, so allocation is grounded in measured results.

Underneath all three sits trust, the precondition and the hardest thing to earn. IRIS's neutrality is mechanical, not promised: it works for the fit rather than for whoever pays, founders pay nothing, and your data, networks and training stay your own.

The theory of change, briefly

The goal is that by 2030, founders and nonprofit leaders without established networks reliably reach the RIGHT funders who would support and back them on their journey to impact, and the time they reclaim goes into their work. That goal needs three states to hold: discoverability, reclaimed time, and outcome-grounded allocation, with trust as the shared precondition.

The full map, with every link weighted and every assumption rated for confidence, is here: IRIS theory of change on fenc.es. Huge thanks to Florian Aldehoff-Zeidler for buildin this theory of change tool that will also help all the founders and funders on IRIS to maximise and scrutinise their impact. 

What we will measure

Measuring outcomes is the product, so the feedback loops are short. We will track the counterfactual question (builders who close funding through IRIS who would not otherwise have reached those funders), founder hours spent on fundraising per raise (women founders for example spend on average more than 40% of their time fundraising compared to the men peers, same problem if you're building in Nigeria rather than SV) , and warm intros convert 10 to 15x better than cold emails for securing investor meetings.

The honest risk

A system that learns from what historically closed can reproduce the exclusion it is meant to fix. The status quo already encodes those biases, invisibly and unaccountably, so this is a reason to build carefully rather than not to build: to measure against a fairness baseline, and to keep the discovery of excluded builders as a deliberate counterweight. IRIS is also a commercial entity, free to founders and funded by premium investor SaaS subscriptions, and I would welcome a direct conversation about how philanthropic capital best fits that structure.

Who I am, and what we are raising

I am CTO and co-founder of Cellcraft, a Cambridge AI and biomanufacturing company, where I raised five million pounds and scaled the team, I have gone through numerous top accelerator programmes and I built a 250-strong female founders' network in Cambridge alongside it. I have sent the cold emails and waited out the silence after pitch decks, and through the network I have watched how funding decisions actually get made. We are raising £500K to expand across Europe, starting with UK and France. 

What would help most from this community: feedback on where the theory of change is weakest, introductions to funders who think about capital access as a cause area, and a conversation with anyone who has funded philanthropic infrastructure. The argument map above is the best place to disagree with me precisely. I will be reading the comments.

Further reading

This post sits inside an active conversation here about how funding finds work. A short map of it:

On access being gated by networks

Issues with centralised grantmaking: shared proxies and networks mean good projects get systematically missed.

More Centralisation?: the decentralised model works well only if you already have the right connections.

What should EAIF Fund?: names the warm-introduction bottleneck and sketches an AI-assisted routing fix.

On funding concentration

What I'd like to see during Funding Strategy Week: the conversation about a single funder dominating the landscape.

On the bottleneck IRIS automates

AI safety is extremely bottlenecked on grantmakers: capital is stuck behind too few people who can find and vet opportunities.

On learning from the venture model

The Infrastructure Gap: philanthropy should adopt what venture capital does well: introductions, support, connections.

A funder already framing the problem

EA Animal Welfare Fund's 3-Year Grantmaking Strategy: names cultural, language and network barriers to funding, and active scouting as the response.

On theory of change as method

Nailing the basics: Theories of change: a theory of change as the business model of the nonprofit world.

Thank you for reading.