A Gulf tax authority - Enterprise generative AI, analytics platform
Government / Tax and Customs · Gulf
The shape of this problem
The answer exists and nobody can find it
The problem
a Gulf tax authority needed a governed enterprise capability spanning employee support, document processing, risk analysis, compliance, stakeholder engagement, and operational insight across diverse tax and customs data. Each of those needs was being met, where it was met at all, by a separate manual pipeline with its own extraction, its own access rules, and its own review step. Effort therefore scaled linearly with the number of use cases, and nothing built for one workflow could be trusted or reused in the next.
What we built
We delivered an end-to-end generative AI platform spanning data ingestion, document intelligence, retrieval-augmented generation, conversational applications, model deployment, governance, and monitoring - supporting HR Assist, Agent Assist, Executive Assist, summarisation, translation, classification, anomaly detection, sentiment analysis, and content generation.
What moved
| Measure | Before | After |
|---|---|---|
| Document-processing time across supported workflows | † | down 55% |
| Analyst throughput | † | up 35% |
| Deployment time for new AI use cases | † | down 40% |
† Measured against the prior level of the same measure. The source publishes the size of the movement, not the figure it moved from.
Document-processing time across supported workflows
Measured against the client's own prior process, indexed to 100. The source publishes the size of the movement, not the absolute figure it moved from.
Analyst throughput
Measured against the client's own prior process, indexed to 100. The source publishes the size of the movement, not the absolute figure it moved from.
Deployment time for new AI use cases
Measured against the client's own prior process, indexed to 100. The source publishes the size of the movement, not the absolute figure it moved from.
Figures are drawn from the practice's own delivery records for the engagement named, measured against the process that preceded it. They have not been through third-party audit, and none is presented as an average across clients.
What it turned on
The leverage was platform-level: every new use case inherited approved connectors, vector stores, model endpoints, prompt patterns, monitoring, and security controls, while use-case-specific evaluation and access rules stopped that reuse becoming a shortcut around tax, customs, or cybersecurity governance.
- Service line
- Workflow Copilots & Agents
The platform story is the strong one, and it is the argument for doing the foundations once. Lead with the structure, not the assistant.
Start here
Which of the four is yours?
Tell us the documents and the monthly volume and we will send the two closest records, with the proof behind each and the honest note where the match is partial.
You get a reply within one working day, from the engineer who would do the work - not a sales sequence.