Claude Cowork 16 min read

Claude Cowork Live Artifacts: What Changed in 2026 and How Artifacts Work Now

Cowork live artifacts ended Aug 19, 2026 and Cowork merged into Claude. How artifacts work now: live data, Dashboards, sharing, storage, pricing and limits.

Shabnam Katoch

Shabnam Katoch

Growth Head

Claude Cowork Live Artifacts: What Changed in 2026 and How Artifacts Work Now

Live artifacts launched in April, stopped being created in August, and Cowork itself merged into Claude in September. Here's what still works, what replaced them, what it costs, and when you need monitoring instead of a dashboard. Updated November 3, 2026.

Claude live artifacts in 2026: old Cowork live artifacts next to the new Claude dashboards and docs that replaced them

The short answer. Claude Cowork launched live artifacts on April 20, 2026: persistent HTML dashboards that pulled fresh data from your connected apps each time you opened them. Since August 19, 2026 you can't create new ones. New artifacts use Claude's updated artifacts system instead: they're saved to your account, open on the web and can be shared. Then on September 16, 2026, Cowork and chat merged into one Claude, so there's no separate Cowork tab any more. Your old live artifacts still open and can still be shared, but you can't edit them in place. To change one, you republish it as a new artifact.

What you'd build as a live artifact today is a connected artifact or a Claude Dashboard. Connecting apps needs Pro ($20/month) or higher, and Dashboards is in beta on Pro, Max, Team and Enterprise.

The rest of this post covers the timeline, how artifacts work now, pricing, how long they last, and the one thing none of them do: tell you when something changes.

Claude artifacts timeline: every change that matters

If you're trying to work out "when did this start" or "when did that change", here it is in one place.

DateWhat happened
June 2024Claude artifacts first launched: code, documents and small apps in a panel beside the chat
April 20, 2026Live artifacts launch in Claude Cowork: dashboards and trackers connected to your apps that refresh when opened
July 2026Artifacts published from Claude Code can call MCP connectors, so pages show current data each time someone opens them
August 19, 2026Cowork stops creating live artifacts; new ones use the updated artifacts system
September 16, 2026Cowork and chat merge into one Claude; Claude Docs and Claude Slides arrive in beta
October 6, 2026New Cowork-style tasks on Pro and Max run in the cloud
October 8, 2026Claude Dashboards and Claude Motion launch in beta; Docs, Slides and Design leave beta on every plan, including Free
October 15, 2026Docs, Slides and Design turn on by default for Enterprise (except HIPAA-enabled organisations)
October 28, 2026Anthropic adds a gallery of ready-made legal artifacts to Claude for Legal
December 14, 2026The standalone Claude Design site closes; Design lives inside Claude conversations

Claude artifacts timeline from the June 2024 launch through live artifacts in April 2026, their retirement on August 19, the Cowork merge on September 16 and Dashboards in October 2026

Why live artifacts got so much attention

A user on the Duke Digital Media Community built a live dashboard in 30 minutes. No code. No engineering team. No BI tool subscription. She described what she wanted in Claude Cowork, and the dashboard started pulling YouTube video counts, RSS feed titles and calendar events every time she opened it.

A working dashboard from a prompt, refreshing with live data, built by someone with no technical background.

That's what made live artifacts a big deal. The idea survived the August and September changes; only the packaging changed. But there's a gap that hasn't changed at all, and it matters if you need monitoring that alerts you, not a dashboard you have to open.

What Live Artifacts Actually Are (the Clear Answer)

Live artifacts are artifacts made in Claude Cowork before August 19, 2026. Each one is a persistent, interactive HTML page (a tracker, dashboard, comparison tool or reference page) that pulled current data from your connected apps and files through MCP (Model Context Protocol) whenever you opened it.

Here's where they stand now:

  • They stay in your Artifacts view, marked with a "Cowork" label, and keep working.
  • You can open them and share them.
  • You can't edit them in place. To change one, open it, click Share and choose to republish it as a new artifact. If you'd already shared it, people with the old link see the new version, so no one needs a new link. After that you can edit it like any other artifact.
  • Organisations that use customer-managed encryption keys (CMEK), zero data retention (ZDR) or a HIPAA-ready setup keep using live artifacts, because the updated system isn't available for those configurations yet.

What artifacts are now. In Anthropic's current definition, an artifact is anything Claude makes that you'd put in front of someone: a design, a deck, a document, a dashboard or a small interactive tool. It opens beside the conversation, and you can edit it, come back to it and share it. Everything you make is saved to the Artifacts tab in the Claude sidebar.

Legacy chat artifacts (before Sep 16, 2026)Cowork live artifacts (before Aug 19, 2026)Artifacts now
Can you create new ones?NoNoYes
Live data from your appsNo (snapshot)Yes, on openYes, on load, on Pro and up
Edit in placeCan't create new; old ones keep workingNo, republish firstYes
Where they openWeb and appsDesktop appWeb, desktop, iOS and Android
SharingPublic "Publish" link on Free/Pro/Max/UltraWithin your organisationPrivate by default; link, email or organisation sharing depending on plan

The core distinction from the original post still holds: a snapshot shows what was true when Claude built it; a connected artifact shows what's true when you look at it.

Snapshot Claude artifact showing data from when it was built versus a connected artifact that fetches current data from Slack, calendar, CRM and warehouse connectors each time it loads

What You Can Build (Five Real Examples)

The same ideas that made live artifacts popular now map onto the new artifact types:

  1. Morning briefing page. "Build a page that shows my Slack mentions, today's calendar and open pull requests." A connected artifact fetches current data each time you open it.
  2. Competitor tracker. "Track my top five competitors: recent releases, blog posts and pricing changes." Ask Claude to build in a refresh button so you can re-pull without leaving the page.
  3. Project status board. "Show open tasks by project from Asana and Linear." Share it with your team, and each person sees it through their own connected accounts.
  4. Sales pipeline dashboard. "How does this month's pipeline by stage compare with last month?" This is what Claude Dashboards is built for: Claude writes a SQL query for each chart, runs it against your data source (Snowflake, BigQuery, Databricks, Redshift, ClickHouse or an app like Salesforce), and shows the query and when the data was last refreshed.
  5. Content calendar with results. A Claude Doc with a live chart of last week's engagement embedded in it. Docs can include charts that Claude refreshes when you ask.

Tip from Anthropic's own guidance: say what you want to know, not just what to chart. "How do signups this month by channel compare with last month?" gets you further than "a chart of signups".

For where Claude's agent features fit against an always-on agent, our Claude Code vs Cowork comparison and OpenClaw vs Claude Cowork go deeper.

How to build a live, connected artifact now (two paths)

You don't need Cowork, a "Live" toggle or a special tab any more. Both paths take a few minutes.

Before you start: artifacts need "Create and edit files" turned on under Settings > Capabilities (on Team and Enterprise, an owner controls this in Organization settings). Connecting apps needs a Pro plan or higher, and the apps you want to pull from should already be connected to Claude.

Path 1: Ask in any conversation

  1. Open a new conversation in Claude on the web, desktop or mobile.
  2. Describe what you want and name the apps and data Claude should use. For example: "Build me a dashboard of open GitHub issues for my repo, today's Slack mentions in #engineering and this week's calendar. Add a refresh button."
  3. The first time the artifact needs an app, Claude asks you to approve the apps and tools it will use. You can switch individual tools off, and that choice sticks.
  4. The artifact opens beside the chat and is saved to your Artifacts tab automatically.

Path 2: Start from the Artifacts tab

  1. Open the Artifacts tab in the Claude sidebar.
  2. Click New artifact, or pick a template: Docs, Slides, Design, Dashboards (beta) or Motion (beta, Team and Enterprise).
  3. Claude asks a few questions about your connectors and what you want to build, then builds it.

Changing it later

To change an artifact you own, paste its link into any conversation or ask Claude to find it, then describe the change. Each change saves a new version. In Docs, Slides and Design you can also edit directly on the canvas, and in a doc you can highlight text, choose "Edit with Claude" and describe the change. If something breaks, "Try fixing with Claude" attempts a repair.

Two ways to build a connected Claude artifact in 2026: ask in any conversation naming your apps, or start from the Artifacts tab with a Docs, Slides, Design, Dashboards or Motion template

How much do Claude artifacts cost?

Artifacts themselves are on every plan. What changes by plan is live data, Dashboards and Motion.

PlanPriceDocs, Slides, DesignConnected apps and stored dataDashboards (beta)Motion (beta)
Free$0YesNoNoNo
Pro$20/month, or $200/yearYesYesYesNo
Max$100/monthYesYesYesNo
Ultra$200/monthYesYesYesNo
Team$25/seat Standard ($20 annual), $125/seat Premium ($100 annual)YesYesYesYes, on by default
Enterprise$20/seat/month billed annually, plus usage at API ratesYes (on by default from Oct 15, 2026)YesOff until an owner enables itOff until an owner enables it

Two cost details people miss:

  • Everything counts against your plan's usage limits. Building and refreshing a Dashboard uses your Claude usage like any other work.
  • Artifacts that call Claude bill the viewer, not you. If an artifact uses Claude inside it, each person signs in with their own account and the usage counts against their own plan. There are no API keys and no cost to the creator.

If you're already hitting limits, our guide to the Claude rate limit covers what counts and how to stretch it.

How long do Claude artifacts last, and where are they stored?

Where they live. Every artifact you make is saved to the Artifacts tab in your Claude sidebar, tied to your account rather than the chat it came from. You can reopen it from any device you use Claude on.

Do they expire? Anthropic doesn't document an expiry for artifacts. Old chat artifacts made before September 16, 2026 and Cowork live artifacts made before August 19, 2026 both keep working.

Do they save data? Yes, on Pro and higher. An artifact can store data between sessions, either personally (only you see it) or shared (everyone using the artifact sees it). Storage is limited to 20 MB per artifact and holds text only. The person who builds the artifact decides which data goes where, so check before typing anything sensitive into someone else's tool.

What deletes data? For legacy chat artifacts, unpublishing permanently deletes everything stored in them, and you can't publish that artifact again afterwards.

Mobile. On iOS and Android you can ask for designs, decks and docs, open them in the Artifacts tab and edit them. Sharing settings are changed on the web or desktop.

Are Claude artifacts public? How sharing works

Every artifact starts private to you. What you can share depends on your plan:

PlanShare options
FreeNo link sharing for new artifacts; old chat artifacts can still be published publicly
Pro, Max, Ultra"Anyone with the link", plus email invites (beta)
TeamAnyone in your organisation; links outside it only if an owner allows external sharing
EnterpriseSame as Team, plus sharing with groups; outside invites off by default

A few rules that apply everywhere:

  • Viewers need a Claude account. The only exception is an old chat artifact that was published publicly.
  • Shared artifacts use the viewer's connections, not yours. If a teammate doesn't have access to a data source, that part of the page shows an error instead of your data. That's good for security and it's also why a shared dashboard can look different for each person.
  • Permissions: view, comment or edit. Treat an artifact from someone you don't know the way you'd treat a file from an unknown sender.

Three limitations of Claude Cowork live artifacts

Some of the original limitations went away with the new system. These three still hold for live artifacts and for artifacts in general.

Limitation 1: They show data; they don't watch it. A connected artifact fetches data when it loads, and it can refresh on an interval or through a refresh button if Claude builds one in. Dashboards show when each chart last refreshed. But no artifact sends you a notification when a number crosses a line. If you don't open the page for three days, you find out three days late.

Workaround: Claude's scheduled tasks (on paid plans) can run a recurring job in the cloud and hand you a report to review in Claude. For alerts that land in the chat app you already live in, you need an always-on agent; see the comparison below.

Limitation 2: Every viewer brings their own access and their own limits. Shared artifacts run on the viewer's connectors and the viewer's plan. A teammate on Free can open a shared page but can't pull live app data, and anyone who hasn't connected the right app sees errors. Team and Enterprise links stay inside the organisation unless an owner opens up external sharing.

Workaround: Agree on the connectors your team uses before you share, and keep critical numbers in a format everyone can reach.

Limitation 3: Old live artifacts are frozen, and some organisations are stuck on them. Live artifacts made before August 19, 2026 can't be edited in place, and organisations on CMEK, ZDR or HIPAA-ready setups can't move to the updated system yet. Dashboards and Motion are still in beta, so features and connectors may change.

Workaround: Republish old live artifacts as new artifacts before you need to change them. Treat anything built on a beta template as a working draft.

A Claude dashboard only shows a drop when you open it, while an always-on agent pushes an alert to your chat app when the threshold is crossed

Claude Cowork live artifacts vs BetterClaw monitoring

Claude has closed part of the gap since this post first ran: scheduled tasks now run in the cloud on paid plans, even when your laptop is closed. So the honest question isn't "can Claude do anything on a schedule". It's where the result goes, what it runs on, and what it costs you.

Claude artifacts and scheduled tasksBetterClaw agent
What you getInteractive dashboards, docs and pages; scheduled reports you review in ClaudeAn agent that watches your tools and messages you
Data refreshOn load, on an interval or refresh button, or via a scheduled taskScheduled checks and workflows, around the clock
AlertsNo threshold alerts from artifactsPush to Telegram, Slack, Discord or email when something changes
Where you see itInside ClaudeIn the chat apps you already use
ModelsClaude models, counted against your plan's usage limitsBring your own key for Claude, GPT, Gemini, OpenRouter or local models, with zero markup
ConnectorsYour connected Claude apps500+ connectors
TeamsShare artifacts; each viewer uses their own accessProjects: several agents working together alongside your team
PricePro $20/month and up for live dataStarter $19/month, Pro $49/month

Use Claude artifacts when: you want a visual dashboard, doc or deck you check yourself, you're already working in Claude, and the data is useful when you look at it.

Use BetterClaw when: you need to hear about it the moment it happens ("pipeline dropped below $50K, tell me on Telegram"), you need checks running while you sleep, or you don't want monitoring eating into your Claude usage.

Use both when: the dashboard is for your morning review and the alerts are for everything in between. The Claude artifact shows you the picture; the agent tells you when the picture changes.

Comparison of Claude artifacts and a BetterClaw agent across output, refresh, alerts, models and connectors

What Claude's artifacts mean for BI tools and monitoring

Live artifacts were Anthropic's first answer to "Claude should make things that last". The October launches made that answer much bigger: dashboards with visible queries, living docs with refreshable charts, decks that export to PowerPoint and Google Slides, and animations that export as MP4. Since their beta launch, people have created more than 45 million documents, presentations and designs with these tools.

That puts real pressure on the quick internal dashboard, the kind that used to need a BI seat, a Retool app or a developer's afternoon. A $20/month Claude plan now covers a lot of it.

But Anthropic is careful about the limits. It positions Dashboards for quick, exploratory questions, not as a governed reporting system, and lets you hand a dashboard off to Grafana, Hex, Mixpanel, Omni, PostHog, Sigma, Amplitude or Perplexity when a question needs deeper analysis, with Looker and Tableau listed as coming later.

And none of it changes the pull-versus-push problem. A dashboard, however live, waits for you to look at it. For monitoring that comes to you, you still need an always-on agent.

Where Claude artifacts fit: quick dashboards you look at, BI tools for governed reporting, and an always-on agent that sends alerts to your chat apps

This is the gap BetterClaw fills

BetterClaw is a no-code AI agent builder for always-on agents. Instead of a page you open, you get an agent that runs your checks on a schedule and messages you where you already work.

  • Alerts where you live: Telegram, Slack, Discord and email.
  • 500+ connectors to pull from your CRM, analytics, project tools and inbox.
  • Any model, your key: Claude, GPT, Gemini, OpenRouter or local models, with zero markup on inference, so monitoring doesn't eat your Claude plan.
  • Projects: several agents working together as a team, alongside your people.
  • Security by default: sandboxed execution, a kill switch, AES-256 encryption and ISO 27001 certification.

Plans start at $19/month for Starter, with Pro at $49/month for 5 agents ($39/month billed annually) and Business at $149/month for 25 agents. Every plan starts with a 7-day trial. Keep the dashboard in Claude; let the agent tell you when to open it.

Frequently Asked Questions

Does Claude Cowork still have live artifacts?

Not new ones. Claude Cowork launched live artifacts on April 20, 2026, and stopped creating them on August 19, 2026. Live artifacts made before that date still open and can be shared, but can't be edited in place; you republish one as a new artifact to change it. New artifacts use Claude's updated artifacts system, and on September 16, 2026 Cowork merged into Claude, so there's no separate Cowork tab.

What replaced Claude live artifacts?

Connected artifacts in the updated system, plus the Claude Dashboards template (beta on Pro, Max, Team and Enterprise). On Pro and higher, an artifact can pull live data from your connected apps each time it loads and refresh on an interval or with a refresh button. Dashboards write a SQL query per chart and show when each chart was last refreshed.

How do live artifacts differ from regular Claude artifacts?

Old chat artifacts were snapshots of the moment Claude built them. Live artifacts were persistent Cowork pages that refreshed from your apps when opened. Today's artifacts combine both: they're saved to your Artifacts tab, can connect to your apps on Pro and up, open on web, desktop and mobile, and can be shared.

Do Claude artifacts update automatically in the background?

No. Connected artifacts fetch data when they load and can refresh on an interval while open or when you press a refresh control, but they don't run or alert you in the background. Claude's scheduled tasks can run recurring jobs in the cloud on paid plans. For alerts pushed to Telegram, Slack or Discord, use an always-on agent like BetterClaw.

How much do Claude artifacts cost?

Artifacts are on every plan, including Free, and Docs, Slides and Design are available to all. Connected apps and stored data need Pro ($20/month or $200/year) or higher. Max is $100/month and Ultra $200/month. Team seats are $25 Standard or $125 Premium per month ($20 and $100 billed annually). Dashboards is in beta on Pro and up; Motion is in beta on Team and Enterprise. All of it counts against your plan's usage limits.

How long do Claude artifacts last? Do they expire?

Anthropic doesn't document an expiry. Artifacts are saved to the Artifacts tab on your account, and older chat artifacts and Cowork live artifacts keep working. The one way to lose data is unpublishing a legacy chat artifact, which permanently deletes its stored data.

Where are Claude artifacts stored, and can they save data?

In the Artifacts tab in your Claude sidebar, tied to your account. On Pro and higher, an artifact can store up to 20 MB of text between sessions, either private to each user or shared with everyone using it.

Are Claude artifacts public?

Not by default; every artifact starts private. On Pro, Max and Ultra you can share with anyone who has the link. On Team and Enterprise, sharing stays inside your organisation unless an owner allows external sharing. Viewers need a Claude account, except for older chat artifacts that were published publicly.

Can I share a connected artifact with my team?

Yes. Shared artifacts use each viewer's own connections, not yours, so teammates see live data only from apps they've connected and have access to. Anything they can't access shows an error instead of your data.

Which connectors work with Claude artifacts?

Any app you've connected to Claude, plus custom connectors, on web and desktop on Pro and higher. Anthropic's examples include Asana, Google Calendar, Slack and Salesforce. Dashboards also connect to Snowflake, BigQuery, Databricks, Amazon Redshift and ClickHouse. Connector tools that need approval for every action aren't available inside artifacts.

When did Claude artifacts come out?

The original artifacts feature launched in June 2024. Live artifacts arrived in Claude Cowork on April 20, 2026, the updated artifacts system replaced them on August 19, 2026, and Claude Dashboards and Motion launched in beta on October 8, 2026.

Can Claude Code create artifacts?

Yes, on every plan that includes Claude Code. It can publish a session's output as a live page at a private URL that updates as the session continues. On Pro, Max, Team and Enterprise, those pages can call MCP connectors each time someone opens them, so they show current data rather than a snapshot.

Want to skip the setup?

BetterClaw does this in 60 seconds. No Docker, no config files.

Start free
Tags:Claude Cowork live artifactsCowork real time dashboardClaude live artifactsCowork artifacts refreshClaude Cowork dashboardlive artifacts MCPClaude dashboardsno-code BIclaude artifacts pricingclaude artifacts storage
Shabnam Katoch

Written by

Shabnam Katoch

Growth Head

Shabnam leads growth at BetterClaw and has spent the past three years building organic distribution for AI-native products. She writes about no-code agent builders, content strategy, and what it actually takes to grow without an ads budget.

Share this article
Was this helpful?