A guarantee fund is easy to describe on a slide and hard to run on a spreadsheet.
Most national credit guarantee institutions in Africa run on the same infrastructure: a shared drive, a set of spreadsheets maintained by two or three people, and an email thread with each partner bank for anything that needs a decision. It works, until it doesn’t — until a donor asks for exposure figures across five funds, or a board meeting needs a sinistrality rate nobody had time to compute, or two people in the office update the same eligibility rule with slightly different numbers. Garantia, the guarantee fund management platform we built, started from a simple observation: the problem isn’t that these institutions lack discipline, it’s that spreadsheets were never designed to hold a process this specific.
What a guarantee fund actually has to track
A credit guarantee fund’s job sounds simple: a bank or leasing company makes a loan, the fund guarantees part of the risk, and if the borrower defaults, the fund pays out its share. In practice, that single sentence hides an entire lifecycle — from the moment a partner bank submits a request, through eligibility checks, scoring, approval, repayment monitoring, and, when things go wrong, a loss declaration. Garantia digitizes that full lifecycle, with every dossier moving through nine tracked states between request and closure. None of this is exotic engineering. What makes it hard is that the workflow has to match the institution’s actual rules exactly — not a simplified version of them — because the numbers that come out the other end are the ones a board, a regulator, or a donor will act on.
Why one fund’s rules can’t be another’s
Most guarantee institutions don’t run one fund — they run several, often for different sectors: SME lending, agriculture, energy, exports. Each fund typically has its own eligibility criteria, its own guarantee ratio, and its own budget ceiling, set independently of the others. Off-the-shelf case management software tends to assume one workflow fits all; Garantia was built the opposite way, with multi-fund management as a first-class concept rather than a configuration afterthought. That’s less a software feature than a modeling decision, and it’s the kind of decision that’s expensive to get wrong after the fact — one more reason institutions in this position benefit from custom software built around their actual rules rather than adapted from something generic.
Two audiences, one system
A guarantee fund isn’t used by one type of user. Partner banks and leasing companies need to submit requests and track their status without seeing anything that belongs to another partner. The fund’s own back-office team needs the opposite view: everything, across every partner, with the ability to score, approve, and monitor exposure. Garantia handles this with a dual workspace — a partner extranet and an institutional back-office — built on the same underlying data so the two sides never fall out of sync. That last part is the harder engineering problem. It’s the same challenge we described when writing about why connecting systems is harder than building them: two interfaces reading and writing the same records will eventually disagree unless the underlying model was designed for that from day one, not patched in afterward.
Building for the audit, not just the transaction
The institutions that run guarantee funds don’t answer only to their own management — they answer to boards, to donors, and often to a banking or financial regulator, all of whom want the same three things: a clear sinistrality rate, real-time exposure by fund and by sector, and a decision trail they can actually audit. That last requirement shapes the software more than people expect. Garantia’s scoring model evaluates every dossier across three dimensions — project, sector, promoter — and keeps a full record of how each decision was reached, not just what the decision was. It’s a distinction that mostly matters to public-facing and donor-funded institutions, which is also why this kind of build sits close to the work we do in public sector information systems: the technical requirements are ordinary, but the accountability requirements around them are not.
None of Garantia’s individual pieces — workflow states, multi-fund rules, dual access, an audit trail — is unusual in isolation. What made it worth building as dedicated software, rather than another set of linked spreadsheets, is that a guarantee institution needs all of them to hold together at once, under rules that don’t compromise. That’s usually the actual argument for custom software over a generic tool: not that the generic tool can’t do any of the individual tasks, but that it can’t hold the whole shape of the problem without quietly losing something at the edges.