Large Kuwaiti IT systems integrator · Digital platforms

Project margins, vendors and delivery risk in one portal, with a chat that answers leadership's questions.

  • AI data analysis
  • Dynamics 365
  • Microsoft sign-in
  • OpenAI
  • Risk scoring
Executive Pulse dashboard with revenue, profit, cost and margin by year, on demo data
  • ~15dashboards over ERP project data
  • 18 daysfrom first commit to the plain-English chat
  • 0–100health score for every project
  • 89commits
Before

Project figures lived in ERP Excel exports: a project summary, a sales-and-cost transaction sheet, a vendor mapping and a separate account-manager list. They needed cleaning before anyone could analyse them.

Reporting looked backwards, with no way to drill down to an individual project, no export and no benchmarks.

What we built
Executive Pulse
Revenue, profit, cost and margin by year and department, including which project years drive each year's revenue.
Project drill-down
A searchable portfolio with CSV export, and per-project P&L, delivery progress and a health score.
Risk scoring
A 0–100 health score from margin, delivery progress, account-manager changes and profit trend, with a watch list of the worst large projects.
Vendor and customer intelligence
Vendor margins and concentration flags, customer growth and churn, and a whale curve.
Clean data from messy exports
Rules that skip cancelled projects, normalise names and allocate uncategorised revenue in proportion to cost.
Risk scoring page with health bands and a watch list of large projects
Risk scoring: the worst large projects, on a watch list.
Vendor ecosystem dashboard with margins by vendor and a manager-by-vendor heatmap
Vendor ecosystem: margins, concentration and who sells what.

Where AI works in the solution

answering questions

Ask in plain English, get a table back

Executives ask questions in plain English. The model first matches any person, customer or vendor name, then requests a query against the project data and returns a formatted table and summary in Kuwaiti dinars. Every query is checked before it runs: reads only, two project tables only, and capped result sizes.

Integration

ERP exports in, answers out

  1. existing systemDynamics 365project and transaction exports
  2. we builtData loadercleaning and allocation rules
  3. we builtAnalytics portaldashboards, drill-down, chat
  4. AI modelOpenAIquestions to read-only queries
SystemDirectionMethodWhat flows
Dynamics 365inExcel exports, loaded and cleanedProject summaries, sales and cost transactions, vendor categories, account-manager status
OpenAIoutAPI · function callingQuestions, name matches and query results
Microsoft Entra IDinSingle sign-onStaff sign-in
next case studyAI-read dental order forms, from phone photo to paid lab order
Read it →