How HexaFlow built an AI feasibility platform for Dubai Municipality. Three modules: one reads a feasibility and fills the whole Excel model, one benchmarks it against world standards and returns a gap analysis, and one generates an entirely new professional feasibility from historical data.
A new feasibility study from local historical data and current market conditions.
Inside a municipal authority, every proposal has to be read, turned into the figures a decision rests on, judged against international standards — and sometimes written from nothing at all.
Feasibility assessment inside a municipal authority means working through large volumes of dense technical and regulatory documentation, where the material that determines a decision is buried across hundreds of pages — and arrives in whatever format the submitting consultancy chose to use.
The work does not stop at reading. Every assessment produces downstream artefacts: financial sheets, standardised forms, calculation tables. These were being filled by hand, transcribing figures out of one document and into another, which is slow and exactly the kind of task where transcription errors quietly enter the record.
Benchmarking was harder still. Judging a proposal against ISO, WHO and other international standards requires knowing which clauses apply and finding the corresponding evidence in the submission — expert work that varied depending on who performed it.
And commissioning a feasibility from scratch meant going outside: a professional study is a six-to-twelve-week external consultancy engagement, even when the authority already holds the historical data the study would be built from.
Not document automation. A feasibility platform that does the analyst's entire job — and then does the job the analyst could never do at all.
The analyst's reading and transcribing, done from the source.
Ingests any feasibility study, external or internal, parsing inconsistent structures into a common representation rather than requiring a fixed template. Every figure is extracted and the complete Excel model is filled — sheets, formulas and calculation chains computed from the source document rather than retyped from it.
| A | B | C | |
|---|---|---|---|
| 1 | Item | Value | Source |
| 2 | |||
| 3 | |||
| 4 | |||
| 5 | |||
| 6 | |||
| 7 |
Illustrative sample. Click any cell: an extracted value names the passage it came from; everything else is a formula.
Expert judgement, applied the same way every time.
Scores the submission against ISO, WHO and other international frameworks, and returns a full gap analysis: what is met, what is unevidenced, what falls short, and what it would take to close each gap. The same judgement applied the same way, regardless of who is reviewing.
Risk review happens once, at approval. Nothing covers construction or operation.
Define a review cycle through construction and operation, and who carries it out.
Illustrative sample, paraphrased — open a requirement to see its evidence and what closing the gap would take.
The job the analyst could never do at all.
Produces an entirely new professional feasibility study from the authority's own local historical data and current market conditions, written to Big Four consulting standards. Modules 01 and 02 make an existing process faster; this one removes the need to commission the study externally at all.
A six-to-twelve-week external consultancy engagement — even when the authority already held the historical data the study would be built from.
Generated in-house from that data and current market conditions, as a draft for the authority's own reviewer to sign off.
The case for the development, the headline financials and the recommendation, on one page.
Demand and pricing for comparable space, drawn from current market conditions.
How the authority’s own comparable projects performed, from its local historical data.
Costs, cash flows and returns — sheets populated and formulas applied, as in Module 01.
The proposal measured against ISO, WHO and other frameworks, with any gap stated.
A recommendation written for the reviewer to accept, amend or reject.
Illustrative sample. Tick each section as reviewed — sign-off stays locked until every one is.
Every output is built for a reviewer to interrogate — the extracted values trace back to their source, the benchmark shows its evidence, and the generated study reads as a draft to be signed off rather than an answer to be trusted.
Every extracted figure names the passage it came from — select any input cell in the Module 01 demo and the source lights up.
Each requirement carries what the submission shows, what it doesn't, and what would close the gap.
It is written for a reviewer to accept, amend or reject — and signed off by a person, not by the platform.
Two entry points feed one structured feasibility model. Submitted documents are parsed, chunked and passed through LLM extraction; historical data and current market conditions feed the generation path directly. From that shared model the platform drives form and sheet population, formula calculation, standards benchmarking with gap analysis, and — where no submission exists — a complete generated study for reviewer sign-off.
Two ways in, one model in the middle — so everything downstream works the same whichever way a feasibility arrived.
Extraction from both external consultancy submissions and internal documents
Forms and financial sheets populated and calculated automatically from source
Transcription errors removed from the calculation chain
Feasibilities benchmarked consistently against ISO, WHO and other world standards, with a gap analysis per submission
A complete professional feasibility generated in-house from historical data, without commissioning an external consultancy
Standardised reporting structure across all submissions
Delivered for a government authority
Construction takeoff and bidding from architectural PDFs, in minutes
Reads architectural PDFs, detects the scale on its own, measures only the trade you work in, and returns quantities, labour hours and a bid — then pushes the work into your CRM.
A walkthrough video in, a removals quote in seconds
The customer films their own home. Computer vision identifies and counts every object, estimates dismantling, handling and transport time, and returns a priced quote.
Security assessments that end in proof, not a PDF
A security assessment and remediation platform: nine frameworks, every gap routed to an owner, evidence checked and read by AI, and an auditor-ready closure record — in English and Arabic.