The Problem: 300 Systems That Do Not Talk
The UK government runs hundreds of incompatible IT systems. HMRC does not talk to DWP. The NHS does not talk to social services. Local councils run their own planning, waste, licensing, and council tax systems that are incompatible with each other and with central government. The result: a citizen fills in the same information dozens of times for different agencies, while government runs duplicate processes at vast cost, makes decisions based on incomplete information, and pays billions for IT projects that fail.
The solution is not another bespoke IT project costing billions that fails after a decade. It is what Estonia built in 2001 and has been running ever since: a secure data exchange layer connecting every government database, with a single citizen identity at the centre.
A. UK X-Road: The Estonian Architecture
Estonia's X-Road is a data exchange layer, not a central database. Each agency keeps its own data in its own system. X-Road provides a secure, audited pathway for agencies to query each other's data with citizen consent. When you log in to an Estonian government portal, your identity is verified once. The portal queries, with your consent, every agency that holds relevant data for your transaction. You fill nothing in. The data flows.
The results after 25 years of operation:
- 99% of all public services available online, 24 hours a day
- 2% of GDP saved annually in administrative costs
- Every Estonian citizen saves approximately 5 working days per year previously spent on government bureaucracy
- X-Road now connects 900 organisations and handles 1.8 billion queries per year in a country of 1.3 million
- The EU is now adopting cross-border X-Road for interoperability between member states
Forge builds UK X-Road: a single interoperability layer connecting HMRC, DWP, NHS, DVLA, Land Registry, Companies House, local councils, the courts, and border control. Every system can query every other system, securely, with an audited consent record, in real time. No more bespoke IT projects. No more forms asking for information the government already holds.
B. Government Admin: What Gets Simplified
| Area | Current waste | Forge fix | Annual saving |
|---|---|---|---|
| Benefit fraud and error | 9.5bn per year in overpayments | Digital ID auto-verifies income, address, and household composition in real time. Eliminates 80% of errors at source. | 4 to 6bn |
| DVLA, Land Registry, Companies House | Three separate agencies, paper-heavy processes, no shared identity | Single digital identity links vehicle ownership, property title, and company records. Standard transactions fully automated. | 300m |
| GP appointment booking | Millions of phone calls daily, receptionist gatekeeping | NHS App auto-allocation when slots open. AI triage routes urgent cases directly. Phone retained for those who need it. | 500m in admin time |
| Court scheduling | Manual case management, adjournments caused by administrative failure | AI-optimised scheduling. Digital case files. Remote hearings for minor matters and case management. | 200m |
| Council services | 300+ councils each running separate IT for planning, waste, licensing, and council tax | Single national platform for back-office functions. Local delivery, shared infrastructure. APIs rather than portals. | 1bn |
| Tax returns | 12 million+ self-assessment returns filed manually every year | Auto-calculated for 80% of taxpayers from payroll, bank, and dividend data already held. Simpler code means fewer errors. | 500m in HMRC admin |
| Life events (birth, death, marriage) | "Tell us once" exists but is poorly implemented. 12 agencies need separate notification. | Digital ID auto-cascades to all agencies. One registration updates tax, benefits, NHS, pensions, and council records simultaneously. | 100m |
| School admissions | Each council runs a separate process with different forms and different timelines | Single national portal. Apply to any school in England from one form. Published algorithm allocates based on statutory criteria. | 50m |
Total estimated annual savings from digital integration: 7 to 9 billion pounds. This is not speculative or future technology. It is database integration, API connections between existing systems, and a single identity layer. Estonia has been running this for 25 years. The architecture is published and freely available. The UK has the resources to build it. What has been missing is the institutional willingness to do it once, properly, rather than paying for 300 fragmented projects.
C. Sovereign Data: Staged Exit From Palantir
The UK has committed approximately 900 million pounds to Palantir, including 330 million for the NHS Federated Data Platform awarded without competitive tender and 240 million for the MoD. Switzerland conducted a full review and walked away from Palantir on sovereignty grounds, citing concerns about a US company holding access to critical national data infrastructure under US legal jurisdiction. The same concerns apply here.
But cancelling overnight would be reckless. Government and NHS systems now depend on these contracts. The staged approach:
- Maintain existing contracts until UK X-Road and domestically-controlled replacement systems are demonstrably operational and tested
- Run parallel UK-built alternatives during the transition period. Contract UK consortia to build the replacements while Palantir runs in parallel. No cliff-edge switchover.
- Exit at break clauses. The NHS FDP break clause is 2027. Forge exits at that point if the parallel system is ready. If not, extends only until it is.
- British-led consortium preference in replacement procurement. Build sovereign capability and UK skills rather than recreating the dependency with a different foreign vendor
- Sensitive data on UK-controlled infrastructure once alternatives are ready. Health records, defence data, financial regulation data, and immigration records must ultimately run on infrastructure that UK law can govern without reference to the US Cloud Act
D. Citizen Digital ID: Tell Us Once, Update Everywhere
Estonia issues a chip-based digital identity to every citizen at birth. It serves as proof of identity, the basis for digital signatures, access to all government services, and authentication for banking. The UK introduces a Citizen Digital ID through the existing digital infrastructure at HMRC and DVLA, which already holds identity data for the vast majority of adults.
- Voluntary at launch — because the infrastructure is not yet ready for compulsion. On day one the X-Road connections between HMRC, DWP, NHS, and DVLA are not yet complete. Mandating something the system cannot yet process creates public hostility without the benefit. The voluntary period is the build period. Estonia — the model for this — launched their ID card as voluntary in 2002. By 2007 it was effectively universal through genuine utility rather than compulsion. The same principle applies here.
- Incentivised take-up from launch. Anyone with a Digital ID has their benefits calculated automatically from real-time verified data, eliminating the risk of accidental overpayments they then cannot repay. Tax statements are pre-filled. NHS login is unified. The ID is genuinely useful. That drives take-up without mandating it.
- Why compulsion eventually matters. The fraud reduction only works at scale. If a significant minority opt out of Digital ID, those opt-outs are disproportionately the fraud cases. The voluntary period lets the honest majority sign up because it helps them. Compulsion then closes the gap that voluntary take-up leaves. Without eventual compulsion the verification system has a structural hole in it — specifically in benefit fraud, which is precisely where it matters most.
- Mandatory for benefits and voting by Year 2 to 3. Once the X-Road infrastructure is live and proven, Digital ID becomes mandatory for: accessing welfare benefits, registering to vote under the mandatory voting system (Section XII), and certain regulated financial activities. This is the sequencing: build the system, prove it works, then mandate it where the stakes are highest.
- The "papers please" objection answered. Opponents will frame mandatory Digital ID as authoritarian. The answer is the sequencing: a year or more of voluntary use demonstrates it is a useful service, not a surveillance tool. By the time it becomes compulsory for benefits, the majority of the population already has one and knows what it does and does not do. The compulsion argument lands very differently after demonstrated utility than it does on day one.
- Strong privacy protections by architecture. No single database of all citizen activity. Each agency holds its own data. The ID enables queries with consent; it does not create a surveillance record. Citizens can audit which agencies have queried their data and when
- Open standards. The ID uses open international standards (ISO/IEC 18013-5 for digital driving licences is already adopted). Interoperable with EU and partner country digital ID systems as those develop
E. Procurement Reform: Open Standards, No Lock-In
- Government IT contracts must use open standards and open-source software by default. No proprietary formats that create vendor lock-in
- Modular systems: individual components can be replaced, upgraded, or retendered without replacing the entire stack
- Every government digital project over 10 million pounds published with full technical specifications, APIs, and data models. Civil society, academics, and competing suppliers can audit and propose improvements
- The Cabinet Office Digital Service and GCHQ's National Cyber Security Centre provide central technical standards and security certification. Agencies cannot build non-compliant systems and claim they are too complex to integrate
F. Civil Service Attrition
As digital services eliminate administrative roles, headcount reduces through natural attrition (around 10% per year through retirement and voluntary departure), redeployment to front-line services, and voluntary redundancy. No compulsory layoffs. Target: 100,000 fewer administrative roles within the parliament, saving 4 to 5 billion pounds annually. Staff who remain receive higher salaries reflecting genuinely skilled roles rather than administrative volume roles that the digital system can perform faster and more accurately.
The principle: a citizen should interact with government once and have the data flow where it needs to go, with their consent, securely, with an audited record they can inspect. Government should not run 300 incompatible systems. Public money should not flow to foreign companies for sovereign services. None of this is technically difficult. It has been institutionally and politically difficult because each department guards its own systems, each outsourcing contract creates lock-in, and each IT project failure makes the next attempt harder to justify. Forge commits to the architecture, not to a single project. X-Road is infrastructure, not software. It connects existing systems. It does not replace them all at once. That is why Estonia built it in 3 years and has been running it for 25.