This is how you build your own dashboards

Claude now creates live artifacts like dashboards and trackers in Cowork that are directly connected to your apps and files. They update automatically when opened. The content remains, is saved in the Live Artifacts tab and can be accessed, edited and further developed at any time – including version history and independent of individual sessions. According to the announcement, the new tab is now available in Claude Cowork for all paid plans such as Pro, Max, Team and Enterprise.

The interface shows classic business metrics such as signups, activation and paid conversions, supplemented by diagrams and report modules. Claude Design has recently made it possible to create complete websites, one-pagers or presentations via prompt. Based on Opus 4.7, Claude not only generates layouts, but also directly usable prototypes that can be further developed in dialogue, via comments or via sliders.

This is how Claude’s live artifacts work

Claude’s live artifacts are content that does not disappear in the chat, but is saved as its own workspaces, for example as a document, code or dashboard. Unlike classic answers to prompts, they remain and can be reopened and edited later. They can also be connected to data, for example when a dashboard loads current figures from connected tools. As Felix Rieseberg, Engineering Lead for Claude at Anthropic, emphasizes on

Marketing, product and data teams in particular benefit from the fact that once created, dashboards, reports and evaluations continue to run automatically and remain up to date. Instead of manually collecting data or setting up dashboards again and again, live artifacts take on these often time-consuming tasks. This leaves more time for detailed analyzes and reporting with always up-to-date data. For the responsible teams, the applications can be shared and iteratively improved together, for example in campaign evaluations, product analyzes or internal reporting.

At the same time, new requirements arise. If external data sources are connected, errors can flow directly into the results, and when sensitive content is involved, the handling of data becomes more critical. Additionally, the need for control increases since automatically updated dashboards should not automatically be considered reliable. Errors in connected data sources or in the logic of the artifact can otherwise flow unnoticed into analyzes and decisions.

Massive computing capacity for Claude – thanks to Amazon

Permanently working AI systems such as live artifacts continuously tie up computing resources. In order to meet Claude’s increasing need for computing power for training and operation, Anthropic is expanding its cooperation with Amazon and securing up to five gigawatts of energy in the long term. As Anthropic reports, Amazon is investing another five billion US dollars, increasing the total stake to 13 billion. Anthropic is also committed to investing more than $100 billion in AWS technologies over the next decade. Additional capacities are to be made available in the short term, including new Trainium2 resources and a total of almost a gigawatt of Trainium2 and Trainium3 power by the end of 2026. In the future, the collaboration includes several chip generations from Trainium2 to Trainium4 as well as other specialized hardware from Amazon.

The mail order company entered into a similar agreement with OpenAI at the end of February 2026. As part of a financing round of around $110 billion, Amazon contributed around $50 billion, valuing OpenAI at around $730 billion. In addition to capital, access to cloud infrastructure also played a central role here.

The article comes from Larissa Ceccio from the OnlineMarketing.de editorial team and is published on as part of a content cooperation.