Government
The policy is in one folder, the spending in a spreadsheet, the law in the gazette, and the lessons in a review from five years ago. Bwat helps officers pull that record together and turn it into briefs and analysis anyone can check, with a source on every claim.
Bwat helps with reading, analysis and drafting. Decisions and accountability stay with public officers.
What was promised, what was paid for, and what was delivered?
An officer transfers and years of context leave with them. Put a programme's documents in a Project now, and every new question starts from the record. A Bwat Analytics pilot goes further: it connects the people, programmes, places, spending and results into an ontology, a digital twin of how the programme really ran.
Give Bwat the solicitation, the published criteria and the bids. It sets out what each bidder offered against each criterion, with the page it's on, and marks what's missing. It gives no scores and makes no recommendation. That stays with the evaluation committee.
Bids against the published criteria
| Criterion | Bidder A | Bidder B |
|---|---|---|
| Similar projects | 3, listed on p. 12 | 5, listed on p. 8 |
| Key staff CVs | Provided, p. 20 | Not provided |
| Tax clearance | Provided, p. 31 | Provided, p. 17 |
| Price | UGX 1.42bn | UGX 1.31bn |
Bidder B has no key staff CVs. No scores given.
Set the current framework, what happened in practice, what stakeholders said, and the options side by side.
Lay bids out against the published criteria and flag gaps, without making the award.
Read hundreds of submissions, group the real positions, keep the minority views, and link back to each one.
Search Ugandan legal material, compare instruments, and keep the source open for legal review.
Cabinet briefs, reports, letters and plain-language summaries from reviewed material, in your format.
A searchable collection around each programme, so new questions start from the record.
Use documents you're cleared to share, say where the decision boundary is, and ask for sources in the answer.
“Using the policy, implementation reports, consultation responses and legislation I've provided, prepare an options brief on [issue]. Separate the current legal position, what implementation shows, stakeholder positions, costs stated in the record, and open questions. Cite every material claim, and don't recommend an option until the evidence table is complete.”
“Read the solicitation, the published evaluation criteria and the bids I've provided. Build a comparison table that quotes or cites the evidence each bidder gives for each criterion, and marks anything missing or unclear. Give no score and draw no conclusion the documents don't support. The evaluation decision belongs to the appointed committee.”
“Analyse these consultation submissions. Identify the distinct positions, the evidence behind each, who is putting them forward, the important minority views, and proposals that are answering different versions of the problem. Link every theme back to the original submissions, and don't treat how often a view appears as proof that it's right.”
“Rebuild the history of [programme] from the approved documents I've attached. Give me a dated timeline of decisions, budgets, procurement, changes to delivery, reported results and reviews. Cite the source for each event, and separate official decisions from later accounts of what happened.”
The first decisions are about the records and the people, long before anyone writes a prompt.
One department, one programme, or one collection of documents.
Who can see what, how long it's kept, where it's hosted, and which AI models may read it.
With the officers who do the work, on questions they already know the answers to.
Record where it fails and when it must hand over to a person, before anyone relies on it.
We'll scope the records, the controls, the test cases, the training and the human review a responsible pilot needs.
