Digital transformation tracker

Nothing, for almost everyone. Four university nodes could exchange data over leased lines.

Made possible: A network built to survive the loss of any node, which is why no one can switch the internet off today.

Computing stopped being something you requested from a department and became something on your own desk.

Made possible: Individual authorship of documents, models, and software, without asking permission from a mainframe operator.

Finding information stopped requiring you to know where it lived, or to own the right reference book.

Made possible: Publishing to a global audience with no printing press, no distributor, and no permission.

A shop stopped needing to be somewhere you could walk to during opening hours.

Made possible: Selection beyond what a local shelf could hold, and sellers reaching customers with no shopfront at all.

Checking a balance or moving money stopped requiring a queue and a counter.

Made possible: Financial admin at any hour, and the slow decline of the branch network that had defined retail banking.

The internet stopped being a place you went to and started being a thing that was simply on.

Made possible: Media, software delivery, and real-time services that a metered, occupied phone line could never carry.

Small sellers reached national customers without building distribution of their own.

Made possible: A one-person business with global reach, and a new dependence on the platform that grants it.

Laptops and VPNs made it possible to work away from the office, for those permitted to.

Made possible: The technical precondition for remote work, sitting unused for fifteen years because the culture said no.

The phone stopped being a phone. Access became continuous rather than a thing you sat down for.

Made possible: Location-aware, always-available services, and the first computing most of the world ever owned.

Starting something no longer required buying servers or waiting months for a data centre.

Made possible: A two-person company renting the same class of infrastructure as a bank, and paying by the hour.

Taxes, licences, and records became forms you submit rather than queues you join.

Made possible: Service at population scale without proportional staff, and a new class of exclusion for those without access.

Watching stopped being something that happened at a time somebody else chose.

Made possible: On-demand everything, the collapse of the broadcast schedule, and eventually the fragmentation of the audience.

Your history stopped living in a folder in one building.

Made possible: Care that follows the patient between providers, and a permanent argument about who may read it.

Paying stopped requiring a card, and eventually stopped requiring a wallet at all.

Made possible: Payments for people a card network never reached, and merchants with no terminal.

Meals, rides, and groceries arrived by app, and waiting became a tracked progress bar.

Made possible: Convenience at scale, alongside a labour model whose employment status is still being litigated worldwide.

Anyone could publish to a potential audience of millions with a phone and no gatekeeper.

Made possible: New livelihoods and new reach, plus a recommendation layer that now decides what most people see.

Consent, access, and erasure became rights with penalties attached rather than policies you hoped for.

Made possible: A template copied worldwide, and the first credible answer to 'what happens if they misuse my data'.

Source: Regulation (EU) 2016/679, applicable since 25 May 2018.

Offices closed and knowledge work moved home in weeks, using capability that had existed for years.

Made possible: Hiring detached from geography, and a permanent renegotiation of what an office is actually for. The clearest example that transformation is usually blocked by assumption rather than by technology.

Brazilian residents gained enforceable rights over their personal data, with a national authority behind them.

Made possible: Data protection anchored in Brazilian law rather than borrowed from abroad, alongside the Marco Civil's earlier guarantees.

Source: Lei 13.709/2018 (LGPD), in force from 2020; ANPD as supervisory authority.

Transfers between any two people or businesses settled in seconds, free for individuals, at any hour.

Made possible: A national instant-payment rail built by the central bank rather than by card networks, adopted at a speed that surprised nearly everyone and became a reference case internationally.

A consultation stopped requiring travel, a waiting room, and a day off work.

Made possible: Access for remote and immobile patients, following regulatory change that had been debated for a decade and moved in weeks.

Drafting, summarizing, translating, and explaining became things you could ask for in ordinary words.

Made possible: Software that responds to intent rather than to syntax, for people who never learned the syntax.

Providers of large general-purpose models took on documentation and transparency duties in the EU.

Made possible: Regulatory attention on the model itself, not only on the application built from it.

Source: Regulation (EU) 2024/1689, GPAI obligations applicable 2 August 2025.

Certain AI uses became illegal in the EU outright, with penalties up to 35 million euro or 7% of global turnover.

Made possible: The first binding line between AI that may be built and AI that may not, anywhere in the world.

Source: Regulation (EU) 2024/1689, prohibitions and AI-literacy obligations applicable 2 February 2025.

Projected: software that carries out multi-step work rather than waiting to be operated step by step.

Made possible: If it holds, routine process work shifts from being performed to being supervised.

Source: Gartner projects 40% of enterprise applications will integrate task-specific AI agents by the end of 2026, up from under 5% in 2025. Gartner also projects that over 40% of agentic AI projects could be abandoned by 2027 on governance and return-on-investment failures. Field surveys in 2026 report high adoption but only around one in ten deployments running in production.

Systems used for recruitment, credit scoring, education, law enforcement and similar face conformity assessment, registration, and human-oversight duties.

Made possible: The most operationally demanding tier of AI regulation, and the clearest test of whether the framework is workable.

Actively moving. The Digital Omnibus on AI reached provisional trilogue agreement on 7 May 2026 to defer Annex III obligations to 2 December 2027, and product-embedded Annex I obligations to 2 August 2028. That agreement is NOT yet law: until it is formally adopted and published in the Official Journal, 2 August 2026 remains the operative legal date and should be planned for.

Source: Regulation (EU) 2024/1689, Annex III high-risk obligations scheduled for 2 August 2026.

AI inside medical devices, machinery, vehicles and similar regulated products comes under the AI Act's high-risk regime.

Made possible: Regulatory coverage of AI that arrives inside a physical product rather than as software you choose to install.

Source: Regulation (EU) 2024/1689, Annex I obligations scheduled for 2 August 2027; the provisionally-agreed Digital Omnibus would move this to 2 August 2028 if adopted.

Projected: software evaluating vendors, comparing terms, and transacting on a company's behalf.

Made possible: If it holds, procurement cycles compress from weeks to hours, and vendor discovery stops being a human activity.

Source: Gartner projects that by 2028, 90% of business-to-business buying will be intermediated by AI agents, routing over 15 trillion US dollars of spend through agent exchanges.

Projected: infrastructure changes proposed and executed by software under human policy rather than human hands.

Made possible: If it holds, operations work moves toward defining guardrails and reviewing decisions instead of making changes.

Source: Gartner (Predicts 2026: AI Agents Will Transform IT Infrastructure and Operations, published 4 December 2025) projects adoption rising from under 5% in 2025 toward a majority by 2029, and human-in-the-loop involvement in IT operations workflows falling from 95% in 2025 to 40% by 2028.

Projected: robots, vehicles, and sensors producing trajectory data at a scale that dwarfs today's digital telemetry.

Made possible: If it holds, the data-protection and networking problems of the next decade are physical rather than web-shaped.

Source: Gartner data and analytics predictions (announced 11 March 2026) project that by 2029, AI agents will generate ten times more data from physical environments than all digital AI applications combined.