LLaMandement — France's AI Tool for Processing Parliamentary Budget Amendments
France
LLaMandement is DGFiP's fine-tuned Llama 2 model that classifies and summarises amendments to France's annual Finance Bill. Officials report it …
Good practice
★ 67
★ 87
★ 53
★ 53
Consult — UK Government's AI Tool for Analysing Public Consultation Responses
United Kingdom · London · See the United Kingdom profile
Consult, built by the UK's i.AI unit with the Alan Turing Institute, uses generative AI to analyse tens of thousands of public consultation responses in hours. An evaluation found its tagging matched trained human reviewers (F1 0.79-0.82 vs 0.74 human-human agreement).
50,000+
Consultation responses analysed (Independent Water Commission call for views) (2025)
~2 hours
Analysis time (2025)
0.79-0.82
F1 score vs expert human coders
0.74
F1 score between two independent human reviewers (baseline)
~400x cheaper
Estimated cost reduction vs manual analysis
up to 75,000 official working days/year
Potential annual working-day savings if adopted government-wide
Details
- Maturity
- Pilot
- Promoter
- Incubator for Artificial Intelligence (i.AI), Department for Science, Innovation and Technology (DSIT), with the Alan Turing Institute
- Period
- 2024–present
- Keywords
- digital government, public consultation, natural language processing, regulatory policy
Context
The UK's Incubator for Artificial Intelligence (i.AI, DSIT), with the Department for Transport and the Alan Turing Institute, built 'Consult', a generative-AI tool that reads and thematically tags free-text public-consultation responses for policymakers.
Results
On the 2025 Independent Water Commission call for views, Consult analysed 50,000+ responses in about two hours; a published evaluation against expert human coders reported F1 scores of 0.79-0.82, higher than the 0.74 F1 measured between two independent human reviewers, and estimated roughly 400x cost savings versus manual analysis. Consult has been trialled on 26 live consultations, with three departments (Defra, DWP, Scottish Government) publishing their own evaluations, ahead of a wider rollout planned for summer 2026.
Implementation
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Data sources
Where this practice's information was retrieved from, and when.
- C-NAPSE web research — accessed 2026-07-01
- Incubator for Artificial Intelligence (i.AI), Department for Science, Innovation and Technology (DSIT), with the Alan Turing Institute — accessed 2026-07-01
- GOV.UK — AI Consultation Analysis Tool evaluation — accessed 2026-07-01
- AI Knowledge Hub — Consult — accessed 2026-07-01
- i.AI blog — Evaluating Consult — accessed 2026-07-01
- Nesta — How does the public feel about AI analysing consultation responses? — accessed 2026-07-01
- GOV.UK — Algorithmic Transparency Record: DSIT Consult — accessed 2026-07-01
Attachments
Similar practices you may find useful
vTaiwan — Taiwan's AI-assisted public deliberation platform for policy-making
Taiwan
AI-assisted civic deliberation platform run by Taiwan's PDIS, using open-source Polis AI to map consensus on ~30 policy issues since …
OpenEval — Finland's AI-Assisted Search Platform for Development Evaluation Evidence
Finland
Finland's Ministry for Foreign Affairs launched OpenEval, an Azure-based AI tool built with CGI that lets civil servants and the …
PCO AI Legislative Drafting Pilot — New Zealand's generative-AI trial for explanatory notes and consolidations
New Zealand
New Zealand's Parliamentary Counsel Office ran a six-month, five-partner R&D programme fine-tuning an 8-billion-parameter model on 200 examples of NZ …