Actuarial Scientist Behind KRA UshuruGPT System Reveals How AI Simplifies Tax Compliance

Elizabeth Andiko maintained an interest in technology long before artificial intelligence became a dominant global topic. She constantly imagined practical ways digital tools could make daily activities simpler for ordinary citizens.

That initial vision took concrete form when the Kenya Revenue Authority (KRA) introduced its conversational platform. The tax agency unveiled UshuruGPT at its annual summit in October 2025.

Andiko served as a primary technical advisor during the development of Kenya's inaugural Artificial Intelligence (AI) tax assistant. The system provides clear, step-by-step guidance without complex jargon.

Citizens interact with the platform using straightforward conversational language. This design eliminates traditional long physical queues, complex paperwork, and unnecessary anxiety for first-time filers.

The system translates complicated tax policy frameworks into clear digital responses. Users receive immediate answers to urgent tax-related questions through user-friendly channels.

System designers prioritised low-bandwidth parity from the start of development. Consequently, taxpayers can access guidance through popular messaging platforms like WhatsApp.

The architecture also supports Unstructured Supplementary Service Data (USSD) channels for simple feature phones. This approach ensures accessibility for users across diverse internet connectivity levels.

Engineers incorporated gender-responsive design principles directly into the platform interface. Special support features were also included to assist Persons with Disabilities (PWDs).

Multilingual interfaces were built into the software from inception. These additions make digital public services accessible to users with varied literacy backgrounds.

System integration allows users to log in with National Identification (ID) credentials. This removed the persistent requirement to memorize a Personal Identification Number (PIN).

Direct connectivity between national registration systems and tax databases reduces user frustration. Interoperable infrastructure allows state agencies to share data securely without placing burdens on citizens.

Project performance will be measured through public trust and usability data. Key metrics include reduced physical office visits, quicker dispute resolutions, and higher user satisfaction.

Andiko previously applied human-centered design to agricultural registry projects. Farmers verified crop inputs using offline Quick Response (QR) codes and mobile USSD codes.

That agricultural testing proved that public technology must focus on building trust. Operational success depends on clarity and ease of use, rather than complex backend features.

The deployment of UshuruGPT reflects a broader shift toward inclusive governance. Bringing automated assistance to everyday messaging tools makes official tax compliance direct and accessible.