DIRECT ANSWER / 04
How do you compare a grant budget with its narrative?
Compare stable budget rows with specific narrative claims rather than comparing totals. Confirm each relationship by hand, recompute the visible arithmetic, and document every difference that a person has not resolved.
Why this review exists
A budget narrative explains the costs behind a line-item budget: what each amount represents and how it relates to the proposed work. The two documents should describe the same reviewed plan, and the places where they stop doing so are what a reviewer is looking for.
Why row-level comparison, not totals
A total-only check passes whenever the numbers happen to add up, which they usually do. What it misses is role, period, quantity, funding source and repeated-amount differences: the coordinator described at one rate in the table and another in the prose, the transit support written as ten months in one place and twelve in another.
In the sample review the personnel paragraph states 24,000 for a program coordinator across twelve months alongside 6,000 in other organisational funds, and the budget row B-0001 says exactly that. The value of the comparison is that both statements now point at the same identity, so a later edit to either one is visible.
Budget rows keep stable identities, B-0001 through B-0004 in the sample, so a mapping survives a re-import and a finding stays attached to the row it was about.
Claims are per paragraph, not per document
Narrative claims are extracted per paragraph and keep their surrounding context, their character offsets and the raw text as written. Claim C-0001 in the sample is the string twelve months from the personnel paragraph, with the sentence it came from stored alongside it.
That context is what lets a reviewer decide whether a figure is a line-item claim or a rollup. In the sample, the 46,500 in the evaluation paragraph is the proposal total restated, so it is marked as contextual rather than mapped to any single row.
Line-item review is therefore a decision about meaning, made by a person, recorded with a rationale. The tool provides the candidates, the arithmetic and the record.
Method and limits
Suggestions appear only where a claim's numeric value equals a budget figure within the tolerance you set. There is no fuzzy matching, no similarity score and no automatic confirmation.
The comparison establishes that two documents you supplied agree about their numbers. It cannot establish that either document is correct, complete, eligible or responsive to the opportunity.
Practical sequence
- Record the exact version label of both documents
- Give every budget row a stable identity
- Extract bounded numeric claims from each narrative paragraph
- Confirm each budget-to-claim relationship by hand
- Record every unresolved difference as a finding
What remains outside the app
Grantweft performs arithmetic and textual cross-reference on user-supplied material. It does not write proposals or determine eligibility, allowability, accounting treatment, compliance, completeness, accuracy, submission readiness or funding outcome.
The private desk keeps exact source labels, stable budget row identities, paragraph-based claims, confirmed mappings, visible arithmetic, findings, human decisions, review freezes and passive handoff files connected without an account or remote application database.