# A Future History, November 2026 to Easter 2028

> A dated chronology of how personal agents learn to borrow any screen, from the first voice devices to a universal interface protocol by Easter 2028.

Part of [Any Agent, Any Screen](https://blind-ant.com/pieces/any-agent-any-screen/index.md), a Piece.

*Design fiction, written 18 November 2026.*

## November 2026: The Agents Are Already Here

Nobody calls it the beginning.

Personal agents are appearing everywhere. Dots, Duet, Muse, Instinct and dozens of others can undertake tasks, remember intentions, monitor events, negotiate with services and return with results.

They are not yet the universal assistants science fiction promised. They inhabit applications, browsers, smartphones and cloud services. They have different permissions, different capabilities and different relationships with their users.

But something important has changed. People are beginning to delegate work rather than operate software.

The interface is still a screen. The interaction is increasingly a conversation.

## January 2027: The Agent Becomes Continuous

The first significant change is not a new device. It is persistence.

Personal agents begin operating continuously, maintaining context between conversations, anticipating routine needs and coordinating work across services.

Instead of opening an application to accomplish something, people tell their agent what they want and allow it to find a way.

The limitations become increasingly obvious. Agents can operate complex services, but they still spend considerable effort navigating interfaces designed for human beings.

A growing number of organisations begin exposing their services directly to agents.

The first Domain Experts begin to take shape.

## April 2027: The Voice Device

The first generation of dedicated personal agent devices reaches early adopters.

Small, unobtrusive and designed around speech, they are not conventional smartphones. Their principal function is to provide a permanent conversational connection to a personal agent.

They have no significant visual interface.

For tasks requiring detailed information, the agent transfers the interaction to a companion application on a phone, tablet or computer.

The arrangement works, but feels temporary.

The agent is increasingly independent of applications, yet still needs one to show its work.

## June 2027: Websites Begin to Withdraw

Major service providers increasingly distinguish between human presentation and machine interaction.

Banks, retailers, transport operators and public services begin providing structured information and transactional capabilities to authorised agents.

The website remains, but it is no longer the only entrance.

A Domain Expert can explain its organisation's services, provide authoritative information and carry out permitted transactions without requiring the visitor to navigate a conventional website.

Personal agents begin working directly with these Domain Experts.

The significance is not immediately apparent to most people. They simply notice that their agents have become considerably more useful.

## September 2027: The Display Problem

The popularity of voice-first devices creates an unexpected demand.

People are happy to talk to their agents, but they frequently want to see what the agent is discussing.

A holiday itinerary, a house purchase, a complex financial decision or a collection of photographs is often better presented visually.

The companion application has become a bottleneck.

Manufacturers experiment with transferring agent-generated interfaces between devices. Initially these systems operate only within individual ecosystems.

A television, tablet or laptop can temporarily display an interface created by an agent elsewhere.

The concept acquires a name: the Borrowed Screen.

## November 2027: The First Screen Protocols

Several competing approaches emerge.

Devices announce their display capabilities, supported interaction methods and conditions of use.

An authorised agent can request temporary access, render an interface and release the display when finished.

The early implementations are proprietary.

A household can share screens between its own devices, but a visiting agent cannot necessarily use them.

Hotels, airlines and vehicle manufacturers take an interest. A display could provide personalised services without requiring every organisation to build and maintain its own collection of applications.

The commercial incentive for interoperability becomes difficult to ignore.

## January 2028: Public Screens Join the Network

The first commercial deployments of temporary screen borrowing appear outside the home.

Seat-back displays, hotel televisions, meeting-room screens and selected public terminals become available to visiting agents.

A traveller sits down, grants their personal agent permission to use the nearby display, and receives an interface assembled specifically for their journey.

When they leave, the session ends.

Not every public screen is available for personal use.

Large information displays, advertising installations and transport control screens announce highly restricted permissions. Some can be accessed only by their operators or by authenticated emergency services.

Security researchers demonstrate several attempts to exploit the emerging protocols.

The problem is no longer simply whether an agent can reach a screen. It is whether the agent has the authority to use it.

## February 2028: The Shotgun Incident

During a major sporting event, attackers exploit a weakness in a commercial display network.

For eleven minutes, hundreds of screens in several public venues display an unauthorised video message.

The attack is quickly nicknamed a shotgun broadcast.

The incident reveals the dangers of an infrastructure designed to make displays discoverable and remotely accessible.

An international argument begins over authentication, emergency overrides, public-screen governance and the right to interrupt an existing display session.

Manufacturers accelerate work on hardware-enforced permissions and independently verifiable agent identities.

The incident does not stop the adoption of borrowed screens.

It makes the need for a trustworthy universal protocol impossible to dismiss.

## March 2028: Any Agent, Any Screen

A coalition of device manufacturers, agent developers, transport operators and service providers publishes the first broadly interoperable universal interface protocol.

A compatible display can announce its capabilities and access conditions to any compatible agent.

The agent can request a session without the screen needing to contain its application.

Private screens remain under the control of their owners. Public screens operate according to the permissions established by their operators and the temporary rights of their users.

Agents generate interfaces for the person, task and device involved.

The same agent might speak through a small personal device, present a complex visual interface on a hotel television, and use a seat-back display during a journey.

The interface has become independent of the machine.

## Easter 2028: The Second Generation

The second generation of personal agent devices arrives.

Like the first, these devices are built primarily around conversation. But they no longer require a proprietary companion application for every visual interaction.

They can discover, negotiate with and borrow compatible screens wherever their owners are authorised to use them.

At the same time, the Domain Expert model is replacing conventional websites as the primary way agents interact with organisations.

The personal agent manages intention.

The Domain Expert provides knowledge and capability.

The universal interface protocol connects the interaction to the physical world.

Websites and applications have not disappeared. Billions remain in operation, and many people still use them.

But their position at the centre of personal computing has begun to change.

A person no longer needs to know which application contains a service, or which device can display it.

They ask their agent.

And the world provides the interface.

---

*18 November 2026. A speculative chronology, not a record of future events.*


## More about this

- This is one Object of [Any Agent, Any Screen](https://blind-ant.com/pieces/any-agent-any-screen/index.md), a Piece at Blind Ant.
- [Accessibility, the Same Eighteen Months](https://blind-ant.com/pieces/any-agent-any-screen/accessibility/index.md): The future history of the Borrowed Screen retold through Maya Quinn, a woman blind since birth, from the first personal agents to Easter 2028.

## About the author

Blind Ant is written and run by Charlotte Joanne. [About Blind Ant](https://blind-ant.com/about/index.md) says more.
