Experience from the Delivery Desk – Part 4
Three Perspectives. One Mission: Better Project Delivery
One such lesson came from a project developed for a State Government. The mandate was clear—the AI-powered search utility had to be operational within 150 days.
The knowledge base consisted of Government Regulations, Acts, Amendments, Circulars, Rules and By-laws. The objective was to enable any citizen to ask questions without logging into the system. The AI would generate answers in simple, commonly understood language while maintaining complete transparency by providing references to the original Government documents.
The pilot implementation was encouraging. The response format, language and document references were reviewed and accepted. Development progressed as planned, end-user testing was completed, and by the 120th day the application was deployed for review by Government officers.
Then came an observation that changed the project.
A few officers suggested,
“The answer should be generated using the exact legal language mentioned in the Government Resolution. It will make the response more authentic.”
Technically, the application was functioning correctly.
But the definition of success had changed.
The objective was no longer to provide simplified answers. It had evolved into generating legally accurate responses while preserving the original wording.
Meeting this expectation required changes to the AI implementation and extended the project by another 30 days.
That experience taught me that AI projects introduce a modern delivery risk.
Unlike conventional software, success parameters themselves can evolve during implementation. As stakeholders experience the system, their expectations mature, and what was considered “successful” at the beginning may no longer be sufficient.
For AI initiatives, defining what success looks like is just as important as defining what the system should do.
Delivery Insight
In AI projects, requirements define functionality. Success parameters define acceptance. Both must be validated before development begins.
Have you experienced a project where the solution worked as expected, but stakeholder expectations evolved during implementation?
#ArtificialIntelligence #ProjectManagement #ProjectDelivery #GovTech #AIProjects #DeliveryExcellence #DigitalTransformation
Get in touch
Let's build
something precise.
Office
4, Deepashree Society, Opp. Reliance Trends Mall,
Erandawane, Pune – 411004