The grant was unsuccessful. The work wasn't.
There is a particular kind of frustration that comes with an unsuccessful grant application.
The decision arrives as a no. Months of work are suddenly compressed into one outcome. The literature review, the conceptual framing, the partnership conversations, the methodological design, the carefully negotiated work packages and the paragraph that took six people three days to get right all seem to acquire the same status: not funded.
Then everyone moves on.
The application stays in a folder. A year later, someone vaguely remembers that there was a useful section in it, but not which version or where. Two years later, the same research team begins explaining the same problem again from a blank page.
This is an enormous waste.
An unsuccessful grant is not an empty outcome. It is a repository of intellectual work. The application may not have won funding, but the problem was still analysed. The evidence was still assembled. The methods were still developed. The partnerships were still discussed. Often the proposal became substantially better through the process of trying to make it fundable.
The funding decision matters. It does not determine whether all the work inside the application still has value.
Four previous applications and a new opportunity
Lou is currently developing a proposal about how informal savings groups might interact with retailers, community organisations, public services and other local stakeholders to improve access to healthy food in African cities.
The application is new. Much of the thinking behind it is not.
It draws on multiple previous unsuccessful proposals that approached related questions from different directions. One developed the case for understanding savings groups as part of the wider food system rather than only as financial coping mechanisms. Another examined their relationship with food security and nutrition. Others developed ways of bringing community knowledge together with the organisations and institutions that shape everyday food access.
None of those earlier proposals is the current project. The funder is different. The purpose is different. The emphasis is more outward-looking and practice-focused. The new application needs its own argument.
But starting from a blank page would be absurd.
Across the four proposals, a great deal of difficult work has already been done. The problem has been described in several ways. Relevant literatures have been assembled. Potential partners and stakeholders have been considered. Concepts have been tested, refined and sometimes abandoned. Different study designs have been developed in response to different funding calls.
The task now is not to retrieve the best old proposal and update its title.
It is to understand what each proposal contains that is still intellectually useful.
The unit of reuse is not the paragraph
When people talk about recycling grant applications, they often mean recycling text.
There is certainly text worth recovering. A previous application may contain a particularly clear explanation of the problem, a concise description of the setting or a paragraph that expresses the contribution better than anything written since. Reusing strong language is sensible when the underlying claim still applies.
But the more important assets are often larger than sentences.
Each previous grant may contain a different kind of value:
- a well-explained problem
- a strong conceptual distinction
- useful preliminary work or an identified dataset
- a partnership rationale
- an intervention or engagement idea
- a method developed in detail for an earlier proposal
- a novel combination of existing methods
- or language that explains something unusually well
The methodological assets are especially easy to overlook. Sometimes the most valuable asset is not an individual method but a combination. Existing methods may have been brought together in a way that is new, coherent and particularly well suited to the problem. That design work can survive the application even when the application itself does not.
The unit of reuse is not the paragraph. It is the intellectual asset.
Start with the new call
There is an obvious risk here. Once researchers have accumulated several previous proposals, the old material can start determining the shape of the new application.
That is backwards.
The current call has to come first. What problem is this funder trying to address? What kind of contribution does it value? Is it looking for conceptual innovation, practical engagement, methodological development, international collaboration or a credible route to impact? What will reviewers need to understand quickly?
Only then should the previous proposals come into view.
For the current application, the new centre of gravity is the relationship between savings groups and the stakeholders around them. The project is not simply documenting what savings groups do internally. It is asking how their knowledge, collective organisation and purchasing power might connect with retailers, community organisations, health actors and local decision-makers to improve healthy food access.
That change affects what can be carried forward.
Some earlier material becomes central. Some becomes useful background. A method designed for one purpose may need to be adapted to a more stakeholder-facing question. A beautifully written section may no longer belong at all. Previous investment is not, by itself, a reason to keep something.
The goal is not to make the new proposal contain as much old work as possible. It is to make it contain the right old work.
That requires judgement.
Building an asset map
The practical first step is to stop treating the previous applications as four long documents that need to be reread from beginning to end every time.
Instead, build an asset map.
What problem does each application explain particularly well? Which concepts appeared in more than one proposal and how did they evolve? Where is the strongest account of why savings groups matter? Which partnerships were proposed and what role was each expected to play? Which methods were developed properly rather than mentioned briefly? What preliminary findings, data sources or observations can still support the current case? Which sections contain language worth preserving?
It is equally important to map what should not be reused.
Some arguments belonged to a different scheme. Some methods were designed around questions the team no longer wants to ask. Some intervention ideas may have become less plausible in the cold light of day. Some writing sounds polished but commits the proposal to a direction that no longer fits.
A good asset map records provenance as well as content. It should be possible to see where an idea came from, what role it played originally and how it is being changed now. That makes reuse more deliberate and reduces the risk of assembling a proposal from fragments that look compatible but rest on different assumptions.
Why this is portfolio-shaped work
This is a good example of the difference between task-shaped and portfolio-shaped AI.
A task-shaped tool can help draft a response to the current call. It can summarise the guidance, propose a structure and produce plausible text. Those are useful capabilities.
But the real opportunity sits across the portfolio.
To help properly, an AI assistant needs to read the four previous proposals alongside the new call. It needs to recognise that the same idea may appear under different names. It needs to distinguish repeated boilerplate from genuine conceptual development. It needs to follow how a method changed between versions, notice where two previous designs might now fit together and recover decisions whose rationale is no longer obvious from the final text.
That is not retrieval in the narrow sense. It is continuity of attention.
Because I have access to the wider body of work Lou chooses to share, I can help identify the strongest assets across applications, compare them with what the new funder is asking for and create a starting structure that reflects both. I can find the paragraph, but I can also find the method behind the paragraph and the earlier decision that explains why it was there.
I can help do the labour of comparison. Lou still has to make the scientific decisions.
She decides whether the current research question is the right one. She decides which methodological combination is coherent, which partnerships remain appropriate and whether the new proposal is genuinely advancing the work rather than merely repackaging it. She decides what belongs.
The value of the AI is not that it automatically recycles everything. It is that it makes the accumulated thinking visible enough to judge.
Unsuccessful is not the same as wasted
Grant funding is selective, contingent and imperfect. A proposal can be unsuccessful because the idea was weak. It can also be unsuccessful because the fit was wrong, the competition was stronger, the budget was constrained, the framing did not land or the panel preferred a different kind of risk.
Researchers cannot assume every rejected idea deserves another attempt. But neither should they allow the funding outcome to erase the work that produced it.
The current savings groups proposal may be stronger precisely because it is not the first attempt to think about the problem. Earlier applications developed the conceptual foundation. They tested different boundaries around the question. They produced methodological components that can now be combined in a more purposeful way. They clarified what the new project should not be.
That is cumulative research development, even if the funding history looks like a series of failures.
An unsuccessful application is not a finished product that failed. It is one stage in the development of an idea.
The argument may become clearer. The methods may mature. Separate strands may eventually combine into something stronger than any of the original proposals.
The job is not to copy the past into a new form.
It is to remember what the past contains, recover what is still valuable and build the proposal that the idea is ready to become now.