AP presents AI/ML-based GST admin model at National Co-ord Meet

VIJAYAWADA: The Commercial Taxes Department of Andhra Pradesh presented its Artificial Intelligence (AI) and Machine Learning (ML) driven GST administration model at the 6th National Co-ordination Session (NCM) held at Vigyan Bhawan in New Delhi on Friday.

Key points

• Average detection per case rose to 27.63 lakh from 3.08 lakh under manual processes.
• According to Babu A, intends are underway to expand AI risk-scoring to provisional refunds (RFD-04/RFD-06 processing), arrears recovery, and advance rulings.
• The system uses four data sources, 19 automated analytical reports and a 35-parameter risk matrix to screen cases and route high-value matters for extra scrutiny.
• Babu A remarked the department had developed an integrated AI framework covering case selection, return scrutiny, GST audits, inspections and litigation management.
• Babu A underscored a sharp contrast between AI-assisted results and traditional manual processes.

Chaired by the Revenue Secretary, Administration of India, the gathering brought together senior authorities from the GST Council Secretariat, the Central Board of Indirect Taxes and Customs (CBIC), and Commercial Tax Commissioners from nationwide.

Delivering the presentation titled "Use of AI/ML and Automation in GST Tax Administration From Case Selection to Litigation, " Chief Commissioner of State Tax Babu A outlined how the State has leveraged technology to enhance tax compliance and operational efficiency during a dedicated agenda session.

For context, the system uses four data sources, 19 automated analytical reports and a 35-parameter risk matrix to screen cases and route high-value matters for extra scrutiny. He highlighted the Legal-AI Officer Assistant, trained on the complete GST statutory corpus and regarding 22, 000 judicial rulings from courts and tribunals.

While ensuring human oversight through mandatory officer review and digital signatures, deployed throughout three litigation forums (the First Appellate Authority, GSTAT, and the HC via Online Legal Case Management System), the engine has assisted in over 13, 700 cases by drafting instructions, para-wise remarks.

Babu A underscored a sharp contrast between AI-assisted results and traditional manual processes. Within six months of implementation, AI-driven return scrutiny yielded revenue detection of Rs 743.43 crore, compared to Rs 365.75 crore recorded during a legacy period.

Average detection per case rose to 27.63 lakh from 3.08 lakh under manual processes. Babu A remarked automated systems strictly followed statutory workflows, ensuring traceability through BO Case ID, ARN and DIN generation. The Revenue Secretary praised AP's framework as a national model for AI-led tax administration, prompting delegations from seven states including Tamil Nadu, Bihar, Rajasthan, Chhattisgarh, Kerala, Assam, and Meghalaya to visit AP to study the architecture.

According to Babu A, intends are underway to expand AI risk-scoring to provisional refunds (RFD-04/RFD-06 processing), arrears recovery, and advance rulings. The Chief Commissioner submitted recommendations to the Union Administration and GSTN, requesting direct GSTN API integration, cross-border taxpayer visibility, and sustained peak-period API access while offering cooperation to share AP's technological framework with other states.

For now, AP presents AI/ML-based GST admin model at National Co remains the part of the story worth watching, and further updates are likely as more details are confirmed.