Committee publication · Correspondence · 4 June 2026

Letter from the Department of Science, Innovation and Technology relating to Treasury Minute: Government Services: identifying costs, 21 May 2026

From: Public Accounts Committee

Inquiry: Government services: Identifying costs and generating income

Summary

DSIT's Permanent Secretary responds to the PAC's April letter requesting clarification on recommendation 3b from the Government Services report. DSIT explains why it cannot produce a cross-government prioritised list of legacy systems based on existing 2024 data, citing incomplete coverage, lack of updates, and poor correlation between legacy severity metrics and actual inefficiencies. The letter outlines alternative approaches including the Digital Business Review process and pilots of AI code remediation tools.

Key findings

  • 2024 legacy systems dataset is incomplete, unevenly covers public sector bodies (excluding NHS, policing, local authorities), and has not been refreshed to reflect remediation or degradation since collection
  • Legacy severity framework does not reliably indicate where operational inefficiencies such as manual workarounds and duplicated staffing effort are greatest
  • DSIT will use Digital Business Review process to capture departmental modernisation priorities rather than produce centrally-mandated prioritised list
  • AI code remediation pilot with DEFRA completed 9 months of work in 4 weeks; DEFRA subsequently launched own Legacy Application Programme independently
  • Pilots expanded to four additional public sector organisations; Technology Modernisation Action Plan will improve data quality and evidence base over time, but implementation will exceed 6-month PAC recommendation timeframe

Tone

Procedural

Topics

public-financedigital-transformationgovernment-systemslegacy-infrastructure

Key actors

Emran Mian, Sir Geoffrey Clifton-Brown, Department for Science, Innovation and Technology (DSIT), Department for Environment, Food and Rural Affairs (DEFRA), Public Accounts Committee

Notable line

… found they were able to complete 9 months of work in just 4 weeks. Significantly, we identified that once the AI tools had been deployed …

Key Quotes

… it has not been refreshed to reflect remediation activity or further degradation since that period
Emran Mian · explaining limitations of the 2024 legacy systems dataset
Legacy severity, as measured in the framework, does not reliably indicate where inefficiencies such as manual workarounds, duplicated staffing effort, or process failures are greatest
Emran Mian · explaining why existing data cannot serve as basis for cross-government prioritisation
DSIT first piloted these tools in partnership with DEFRA and found they were able to complete 9 months of work in just 4 weeks
Emran Mian · highlighting success of AI code remediation tools trial
… once the AI tools had been deployed, understood and established in the working practices of DEFRA's teams, they were able to continue the work without further input from DSIT
Emran Mian · demonstrating sustainability and capacity-building through pilot programme
View original document →

Source · parliament.uk record ↗

Letter from the Department of Science, Innovation and Technology relating to Treasury Minute: Government Services: identifying costs, 21 May 2026 | Beyond The Vote | Beyond The Vote