Impact accounting doesn’t function without AI. That is the premise Sedoha
is built on. Financial transactions arrive already priced; impact evidence arrives as
meter readings, surveys, program reports, and published valuation factors. The
frameworks and factor libraries made the field theoretically possible, but impossibly
difficult to implement for years. Turning that evidence into balanced, auditable books
takes a volume of classification and a density of methodology judgment no team could
afford. It’s why impact has frameworks and disclosures, but has never had books.
AI is what closes that gap. It carries the whole methodology in working memory,
applies each ruling across the entire ledger, and holds the books to a
rigor no team could sustain by hand. For us, for Sedoha, the point is not efficiency.
The point is that the books can finally exist, and that they get better every cycle.
Our methodology, our engagements, and the company’s own
operations are all built to be read and worked by humans and AI together. Every
accounting decision is recorded once and becomes precedent. The AI keeps that memory,
reasons over the books, and brings our impact accountants and bookkeepers the new
decisions that need their judgment, while handling the volume work itself:
classification, reconciliation, drafting, cross-checking. People rule on methodology
and on every material accounting call; the system carries each ruling forward and
applies it when the question recurs. The work is infinite in every direction. The learning
compounds.
We run our own methodology on ourselves. The same improvement cycle we deliver to
clients is how Sedoha itself operates. The company is its own first proof of
concept.
None of this loosens the accounting. The deliverable of every engagement is a
double-entry, auditable ledger, and every number in it traces to its source. The
discipline is in the books, not in a trust-us claim about the tools.
The evidence base is designed to compound: each engagement strengthens the evidence our
recommendations draw on: patterns of what worked, never a customer’s data.
Each client’s books stay their own.
