THE POINT / WHAT CHANGES FOR YOU

From news to practice.

A highly discussed product can be withdrawn or change ownership. Before entrusting it with an important task, consider service continuity and the possibility of exporting data.

The future is no guarantee of success

Artificial intelligence fills announcements, industrial plans, and investor presentations. But the promise of a transformative technology does not protect individual companies from failure. Alongside growing projects, there are companies in crisis, devices withdrawn, and services that have lost their autonomy. Builder.ai, Humane AI Pin, and Artifact tell three different outcomes: they should not be confused but help to view the AI race with greater awareness.

Builder.ai: when financial crisis strikes

The Financial Times reported on 20 May 2025 Builder.ai entering insolvency procedures. The company operated in application development with an AI-based offer and had attracted leading investors. The case shows that notoriety, funding, and technological promise do not guarantee business continuity. It is a business crisis: not a demonstration that every technology proposed in the sector is destined to fail.

Financial Times — news of 20 May 2025 ↗

Humane AI Pin: a device can lose its service

In February 2025 Reuters reported the closure of activities related to the AI Pin wearable device by Humane and the sale of assets to HP. On 18 February HP announced an agreement to acquire expertise, intellectual property, and the Humane software platform. These are two aspects to distinguish: the consumer-targeted product was abandoned, whereas part of the technology and expertise found a different industrial path.

Reuters — closure of AI Pin activity ↗

Artifact: the app stops, the technology continues elsewhere

Artifact, the news discovery platform created by Instagram co-founders Kevin Systrom and Mike Krieger, followed a different path. Yahoo announced the acquisition on 2 April 2024, specifying that the service would not continue as an autonomous app and that the personalisation technology would be integrated into its products. It is therefore incorrect to present it as a bankruptcy: it marks the end of an independent experience, with technology continuation inside a different group.

Yahoo — acquisition of Artifact and end of autonomous app ↗

Three different outcomes, one common question

Insolvency, product withdrawal, and acquisition are not synonyms. Even these three cases do not allow calculation of how many AI projects fail: a systematic survey with a defined period and sample would be needed. They do offer a useful question for those who buy, invest in, or report innovation: beyond technical demonstration, is there a business capable of sustaining the service over time? It is the shift from promise to continuity that deserves attention.

A young sector in adoption, not a newly born science

AI has a research history spanning decades. Many of the mass uses of generative AI, the products that make them accessible, and the economic models sustaining them are still under development. In this phase, interfaces, subscriptions, agents, and devices are being experimented with. Some attempts find a market; others are reconsidered or abandoned. The success of a general technology and that of a single product are two distinct matters.

What to observe before relying on a new service

The first check concerns the problem solved: does the product offer a concrete advantage over alternatives? The second is the sustainability of the service: what costs and conditions accompany the offer? The third concerns dependence on the provider's infrastructure: what remains if the service changes or closes, and is it possible to export one's own materials? These are interpretative keys of the analysis, not causes automatically attributed to the three cases described.

Innovation also means selecting

Stories of failure do not erase the opportunities of artificial intelligence. They remind us that the journey proceeds through trial, verification and correction, and that not every experiment becomes a lasting product. For businesses and readers, understanding this uncertainty helps to evaluate announcements with balance: asking for observable results, distinguishing prototypes from established services and updating judgement when new evidence arrives.

Also read: the ten companies leading AI ↗

Sources and documents

Document date or reference period: Events of 2024–2025 · sources consulted on 23 October 2026.

Text prepared with AI assistance based on the indicated sources. Interpretive evaluations are distinct from data and statements from sources.

Report an update or correction →