What changed
Vercel now supports running Anthropic’s Claude Managed Agents through its Chat SDK, a combination that lets you deploy a working chatbot, like a research assistant that lives in a Slack channel, without writing your own server to manage the conversation.
The two pieces come from different companies. Claude Managed Agents is Anthropic’s own hosted service: instead of you writing the code that loops the AI model through steps, calls tools, and remembers what happened earlier in a conversation, Anthropic runs that loop for you on its own servers. Chat SDK is Vercel’s toolkit for wiring an AI agent into a chat interface, including the plumbing needed to make it work inside Slack, Microsoft Teams, Google Chat, Discord, WhatsApp, and other messaging apps, described as “30+ platforms,” without rewriting the whole thing for each one.
Put together, Vercel says you get a chatbot with one persistent conversation per Slack thread, replies that stream in as the model writes them (rather than appearing all at once when it’s done), and a live activity feed that shows which tools the agent is calling and what it’s searching for while it works. Because Claude Managed Agents already stores the conversation on Anthropic’s side, Vercel says you don’t need to run your own database to keep track of chat history. Vercel’s own path to try it is a one-click template: deploy a Slack research agent from a prebuilt project.
Why it matters
The part Vercel’s changelog doesn’t spell out is what “Anthropic stores the conversation for you” actually means for your data, and that’s the detail worth checking before you point this at anything sensitive.
Claude Managed Agents is currently in public beta, and Anthropic’s own documentation is explicit that it’s “stateful by design”: sessions are long-running, resume automatically after a pause, and store your conversation history, sandbox activity, and outputs on Anthropic’s servers so features like Vercel’s “no database to run” setup work at all. Because of that persistent storage, Anthropic’s docs say Managed Agents is not currently eligible for Zero Data Retention, an option some Anthropic API customers can turn on to stop their data from being retained after a request completes, and it doesn’t currently qualify for a HIPAA Business Associate Agreement either, the paperwork healthcare-related businesses need before sending protected health data to a vendor.
None of that makes this unsafe to use. It means the tradeoff that makes the one-click setup convenient (Anthropic keeping the conversation state so you don’t have to) is the same tradeoff that keeps this specific feature outside the standard data-retention opt-out for now. If the Slack bot you’re building will only ever discuss public documentation or non-sensitive research topics, that’s a non-issue. If you’re picturing it summarizing internal customer conversations, financial figures, or anything under a compliance obligation, that’s worth ruling in or out before you connect it to a real Slack workspace, not after.
You’ll also need your own Claude API key to run this at all. Access is enabled by default on Anthropic API accounts, but Claude API usage is billed by the token (the small chunks of text the model reads and writes, which is how most AI companies measure usage), not something bundled free into a Vercel plan. Neither Vercel’s changelog nor Anthropic’s public documentation for Managed Agents lists a specific price for running a session, so don’t assume a number until you check your own Anthropic billing dashboard once it’s running.
Who should care
This is aimed at builders who already have some code comfort, since setting it up means starting from a template and editing a small file of code called a handler, the piece that connects the agent to Slack or whichever app you’re routing it through, even though the one-click deploy button removes most of the manual setup. It’s most useful if you want an AI assistant living inside a tool your team already uses daily, like Slack, instead of a separate chat window nobody remembers to open. It matters less if you’re happy with a standalone chatbot experience, or if the conversations you’d route through it involve information you can’t put in front of a third-party AI vendor’s servers without a data-handling review first.
What builders should do next
Before deploying anything real, read Anthropic’s data retention documentation for Managed Agents so you know exactly what “stateful by design” covers for your account, and decide what topics are off-limits for the bot based on that, not on assumption.
Then test the template itself with a bounded task rather than a vague “try it and see.” Deploy the one-click Slack research agent, and give it a specific, checkable research job, for example: “find the current pricing pages for three named competitors and summarize the differences.” Run that same task through it three times.
Check two things each time. First, whether every fact in its summary traces back to a source you can verify in the live activity feed, since an agent that browses the web can still misread or misattribute what it finds. Second, how long each run takes from your message to a finished reply, since Vercel’s streaming and activity feed are supposed to make a slower research task feel more transparent while it works, not necessarily faster. If it’s accurate and the wait feels reasonable for what it’s doing, it’s ready for real use on non-sensitive topics. If it invents sources or the pace feels like a black box despite the activity feed, that’s a sign to dig into the setup before you hand it a task that matters.
End of article