The Grant Application Is Dead. What Comes Next?

How federated protocols, local agents, and organisational self-sovereignty could replace the broken funding model. Instead of organisations applying to funders, flip the relationship. Each organisation maintains a living, structured, machine-readable profile of itself — not written for any specific funder, but maintained for its own purposes. Funders become discoverers rather than gatekeepers, searching across a landscape of organisations and reaching out when they see alignment. Ideas become first-class objects in the system — structured, tagged, place-based, loosely costed — sitting visibly in the network, evolving over time. The core architecture is a local agent running on or for each organisation: pulling evidence from existing systems, maintaining the living profile, managing ideas, handling access control, and federating outward via open protocols (ActivityPub or AT Protocol). No centralised platform — just a network of self-sovereign nodes.

LLMs have made it trivially easy to produce polished grant applications. Funders are drowning in submissions — many indistinguishable from one another. We are building machines writing applications for machines to read, and the human decisions funding is supposed to enable are getting further away. The application form was a technology for filtering in a low-bandwidth information environment. That environment no longer exists. The form conflates identity, evidence, intent, and fit into a single high-stakes document. Organisations with capacity to write well are structurally advantaged. Smaller organisations doing strong work become invisible. Every funder asks for the same information in different formats, creating enormous uncompensated admin burden falling hardest on those least able to bear it. Fully relational funding (no forms, just relationships) risks reverting to the boys' club — access to funder networks is overwhelmingly a function of privilege.

Shifts power from funders to organisations. Makes organisations discoverable regardless of existing funder relationships — building equity into the infrastructure. Ends the performative absurdity of the application form. Changes temporality of funding from artificial deadlines to organisational readiness. The same structured data satisfies regulatory requirements while also making organisations discoverable, connecting ideas to collaborators, and telling richer stories. Reduces duplicated admin burden across the sector.

Define a protocol-agnostic data model: organisational profiles (identity, evidence, governance, culture) and ideas as structured objects. Build a reference local agent that pulls from existing systems (CRM, finance, impact data) via MCP integrations, maintains the living profile, and federates via ActivityPub. Start simple — static profiles auto-populated from Charity Commission, Companies House, and websites (building on llmstxt.social). Test with a willing funder, a small cohort of organisations, and a genuine alternative to a funding round. Building blocks already exist: llmstxt.social for machine-readable profiles, OpenIdeas.uk for visible/connectable ideas, MCP for system integration, 360Giving and Open Referral for data standards.

Equity and capacity — even with low-friction tooling, some organisations won't maintain a profile, potentially just moving the privilege barrier. Preventing surveillance — access control and audit trails are essential, not optional. Interoperability across fragmented CRM landscape (Salesforce, Lamplight, Charitylog, Airtable, spreadsheets). Funder adoption — funders need to shift from designing calls to discovering and entering dialogue. Coexistence with traditional grant-making during a long transition.

Schema definition with input from organisations and funders. A reference agent implementation. Fellow collaborators across the social sector, funding infrastructure, and federated protocol communities. Political will to try something genuinely different rather than patching what's broken.