In a podcast interview, Anthropic technical staff member Thariq Shihipar recommended using Claude’s Artifacts to manage long-term projects and collaborate with AI agents. He outlined a vision for sharing work plans and the activities of multiple agents on screen, while noting that the underlying infrastructure is still being developed.
Share the big picture of the work, not just requests in chat
For Artifacts, introduced in 2024, Shihipar proposed using the feature as a dashboard for tracking project status. The idea is to do more than send instructions to AI one at a time: it is to manage what is being worked on and what the requirements are in a shared workspace.
One example he gave was managing tasks with a Kanban board, organizing them as cards. He described a system in which task information is stored in a database that multiple Claude instances can access. An AI agent is an AI that carries out work in response to instructions. When work is divided among multiple agents, a shared view of the overall situation is needed alongside individual conversations.
The proposal shifts the focus from how to write prompts to how to share the status of work. For tasks whose requirements extend to diagrams, code or data structures, the idea is that it can be useful to make changes while viewing the work itself, rather than trying to handle everything through text exchanges.
How would chat and the workspace be used differently?
However, adding another point of interaction also means deciding how to use each one. In the interview, co-host SWYX asked how users should distinguish between feedback given through Artifacts and instructions sent to Claude in chat. The alternative, he suggested, was to communicate everything through chat, which would be simpler.
Shihipar envisioned a workspace where users could comment on an evolving work plan and check the work of multiple agents. He described Artifacts as a front end for the infrastructure that supports AI work.
The interface envisioned here is not just for viewing finished work. It is a place to give feedback on plans in progress and adjust the direction of the work. The proposed division of roles is that chat serves as the entry point for requests and discussion, while Artifacts is where users engage with the work itself. However, this division has not been announced as a finalized specification.
A vision that separates where AI thinks from where it acts
On the technical side, Shihipar outlined an approach that separates the interface users see, the place where AI makes decisions and the environment where work is actually performed. The vision is for cloud-based AI to coordinate tasks, carry them out as needed on a user’s computer or in a remote execution environment, and display their status in Artifacts.
For example, even when multiple helper agents are working on separate tasks and communicating with one another, users would be able to track their activity from a single interface. Separating AI decision-making from work performed on a local computer could also allow the system to operate without depending on the computer’s power status.
This is a future vision outlined by Shihipar, not a claim that all these features are already available in Artifacts. He acknowledged that the underlying infrastructure still needs to be built, and the interview did not provide a release timeline or development schedule for the overall concept.
Comments