AI Code Lab – Legacy Modernisation

Tender description

Pre-market engagement This project is closely related to another GDS project, already underway, to roll AICA tooling out to software development teams across government. A combined Request for Information (RFI)/ Expression of Interest (EOI) was live for that broader project 11/09/2025 to 23/09/2025. Responses largely agreed with the feasibility, scope, proposed timelines and delivery methods of the project, and emphasised the need for clear milestones throughout the delivery process. That intelligence has been reflected in the structure and content of this tender. Work done so far Between January and June 2025, DSIT, in collaboration with one strategic supplier and DEFRA, experimented on the use of AI - specifically GitHub Copilot - to modernise a legacy application. The experiment demonstrated that tasks which previously took days could be completed in hours (e.g. building interfaces, drafting migration scripts, and developing APIs were all accelerated with AI assistance). The journey was documented, best practices captured and reusable guidance created in the Defra AI Legacy Modernisation Playbook (https://defra.github.io/defra-ai-legacy-modernisation/) Which phase the project is in Not started Existing team Project sponsor: Chief Architect for the UK Government, GDS Project and Technical Leads: GDS Technology Management Team Address where the work will be done The work will be done mostly remotely. GDS is based at: Government Digital Service, 7th floor, The White Chapel Building, 10 Whitechapel High Street, London E1 8DX Working arrangements The supplier staff will be able to work remotely for the duration of the project. The supplier staff might be asked to work in our London office on an ad-hoc basis to maximise knowledge transfer and integrate with team development processes. Expenses will be capped at £5,000 for the duration of the contract and all travel and subsistence will require prior approval. Provide more information about your security requirements: Baseline Personnel Security Standard (BPSS) Provide more information about your security requirements: All must have BPSS as a minimum. SC might be required to work alongside teams working on sensitive services and assets. At least one individual within the supplier's team must hold SC clearance. Latest start date 2026-02-13 Enter the expected contract length: 4 months Extension period: 6 months Special terms and conditions All expenses must be pre-agreed between the parties and must comply with the Cabinet Office (CO) Travel and Subsistence (T&S) Policy. All vendors are obliged to provide sufficient guarantees to implement appropriate technical and organisational measures so that the processing meets the requirements of GDPR and ensures the protection of the rights of data subjects. For further information please see the Information Commissioner's Office website https://ico.org.uk/for-organisations/data-protection-reform/overview-of-the-gdpr/ Write the term or acronym: AICA: Artificial Intelligence Coding Assistants API: Application Programming Interface BPSS: Baseline Personnel Security Standard CO: Cabinet Office DEFRA: Department for Environment, Food & Rural Affairs DSIT: Department for Science, Innovation and Technology EOI: Expression of Interest GDS: Government Digital Service GDPR: General Data Protection Regulation KPI: Key Performance Indicator PIA: Privacy Impact Assessment RFI: Request for Information SC: Security Check SDLC: Software Development Lifecycle T&S: Travel and Subsistence Are you prepared to show your budget details?: Yes Indicative maximum: 350000 Provide further information: Excluding VAT Confirm if you require a contracted out service or supply of resource Contracted out service: the off-payroll rules do not apply Summary of work GDS requires a supplier to test and evaluate how AI tooling can accelerate the public sector’s remediation of its legacy estate, for example by using AI as a code explainer. This will involve running a series of pilots in central government departments, assessing the strengths and limitations of different AI coding assistant (AICA) tools in understanding and remediating legacy code. Licences for these tools will be procured and distributed either centrally via GDS or by individual departments and the supplier will not need to source these. The aim of the overall AI Code Lab – Legacy Modernisation is to: - Develop a reusable framework for applying AICA to modernise different classes of legacy IT. - Evaluate the financial and non-financial return and broader value of AI-enabled modernisation. - Generate credible evidence and reusable practices to support government-wide adoption. - Feed lessons learned into the legacy technology strategy for Government. This specific project aims to: - Demonstrate financial and operational benefits of AICA for legacy IT modernisation in the prioritised pilots and case studies. - Demonstrate ethical, secure, and compliant deployment of AICA for legacy IT modernisation. - Produce scalable insights and best practices for the use of AICA for legacy IT modernisation to be adopted across departments. - Define the success criteria/scorecard that can be used to evaluate AICA’s application for legacy modernisation in UK Public sector. - Evidence lessons learned to feed into the Government legacy technology strategy. - Provide end of phase reporting that can be used for government publications to support cross department knowledge sharing and generate wider interest. The primary users of AI tools for remediating legacy code bases in government are civil service software engineers and technical delivery teams working across departments such as DSIT, GDS, the Home Office, and other public sector bodies. Thes Where the supplied staff will work No specific location (for example they can work remotely) Who is the organisation using the products or services Government Digital Service (GDS) Why the work is being done Government departments rely on poorly documented, hard-to-maintain legacy software. There is no scalable, efficient approach to modernise these systems without disrupting essential services. AI Coding Assistants (AICA) may offer valuable capabilities in remediating government’s legacy software estate, however there is a lack of robust evidence as to where and how these tools could be most useful, and a lack of structured guidance for AI-enabled remediation activity. The business problem you need to solve The aim of the overall AI Code Lab – Legacy Modernisation is to: -Develop a reusable framework for applying AICA to modernise different classes of legacy IT. -Define the success criteria/scorecard that can be used to evaluate AICA’s application for legacy modernisation in UK Public sector. -Evaluate the financial and non-financial return and broader value of AI-enabled modernisation. -Generate credible evidence and reusable practices to support government-wide adoption. -Feed lessons learned into the legacy technology strategy for Government. This specific project aims to: -Demonstrate financial and operational benefits of AICA for legacy IT modernisation in the prioritised pilots and case studies. -Demonstrate ethical, secure, and compliant deployment of AICA for legacy IT modernisation. -Produce scalable insights and best practices for the use of AICA for legacy IT modernisation to be adopted across departments. -Evidence lessons learned to feed into the Government legacy technology strategy. -Provide end of phase reporting that can be used for government publications to support cross department knowledge sharing and generate wider interest. This evidence will then be used to build a business case for future phases of work. First user type: User Stories for Software Developers First user type: User Stories for IT Managers