Here come the coworking agents

Today: Anthropic and Salesforce roll out slightly different takes on the future of work with AI agents, Microsoft vows to limit the local impact of data-center construction as pushback starts to grow, and the latest funding rounds in enterprise tech.

Here come the coworking agents
Photo by Arlington Research / Unsplash

Welcome to Runtime! Today: Anthropic and Salesforce roll out slightly different takes on the future of work with AI agents, Microsoft vows to limit the local impact of data-center construction as pushback starts to grow, and the latest funding rounds in enterprise tech.

Please forward this email to a friend or colleague! If it was forwarded to you, sign up here to get Runtime each week, and if you value independent enterprise tech journalism, click the button below and become a Runtime supporter today.


Automatic for the people

One reason tech companies are so excited about the potential of AI agents to change the way we work is that AI agents have already changed the way tech workers work. But can they adapt the technology revolutionizing software development and IT operations management into tools that non-technical office workers actually want to use?

Anthropic and Salesforce took a stab at that goal this week, introducing new services designed to tackle a broader spectrum of work activity beyond coding. Anthropic's Claude Cowork debuted Monday as a research preview that uses some of Claude Code's capabilities to take on simple but boring work-management tasks.

  • After giving Claude Cowork access to your desktop, "it can re-organize your downloads by sorting and renaming each file, create a new spreadsheet with a list of expenses from a pile of screenshots, or produce a first draft of a report from your scattered notes," Anthropic said in a blog post.
  • It can also connect to external file-storage services like Microsoft 365 or Box, as well as widely used SaaS tools like Asana and Hubspot.
  • But Cowork is a little rough around the edges, according to ChatPRD founder and CEO Claire Vo: "The challenge with this sort of 'thin wrapper on Claude Code' UX is that it's not quite optimized for the non-technical, and too kneecapped for the [technical user]."
  • Anthropic was quick to acknowledge the limitations of Claude Cowork in its early stages: "By default, the main thing to know is that Claude can take potentially destructive actions (such as deleting local files) if it’s instructed to," it said in the blog post.

After previewing a new version of the venerable Slackbot last year, Salesforce made it generally available Tuesday within Slack. "What used to be your friendly neighborhood notification-helper is graduating to become your personal, context-aware AI agent for work," Slack said in a blog post.

  • The new Slackbot promises to help users prepare for meetings by finding and organizing information from across Slack channels into a single view, or by summarizing new reports or presentations.
  • The user interface for the new Slackbot is much simpler than what Claude Cowork users encounter, as they just chat with the agent the same way they'd chat with a human coworker about an upcoming meeting or project.
  • However, some of its initial capabilities are limited at launch; Slackbot can't book calendar meetings yet, according to VentureBeat, which also noted that Salesforce's recent decision to raise prices for developers using Salesforce APIs could have an impact on how Slack connects to third-party services.

The problem for companies like Anthropic and Salesforce is that Microsoft and Google's office productivity tools have been the daily driver for generations of office workers, and both companies are aggressively adding the same times of AI coworking tools into those core services. It's not clear how many people are actually using those AI tools, of course, but it's also not clear whether or not tools like Claude Cowork and the new Slackbot can entice them to change habits.

  • Salesforce co-founder and CTO Parker Harris "expects people to switch to Slack from Microsoft Teams and that companies might now start getting rid of ChatGPT subscriptions when they see what it can do," according to CNBC, but like a lot of things Salesforce executives imply, there's very little to back that up.
  • Anthropic's task will be even harder given that Claude Cowork is only available to Mac users who are paying it at least $100 a month.
  • But as we've noted many times over the last few years, true platform shifts tend to reward companies that can build something entirely new without having to placate their current customers, and somebody is going to figure this out.

Won't you be my neighbor

Towns and cities across the U.S. generally encouraged the internet-era data-center building boom over the last 20 years, even extending tax credits in hopes of bringing jobs into the rural areas with cheap land and cheap electricity favored by hyperscalers. But those days appear to be numbered, and Microsoft tried to get ahead of future opposition to its infrastructure building plans with a set of promises to work with local communities to minimize its impact.

In a blog post Tuesday, Microsoft President Brad Smith outlined a five-point plan to develop "Community-First AI Infrastructure — a commitment to do this work differently than some others and to do it responsibly." The highlight of that plan was a vow to avoid driving up local electrical costs by asking "utilities and public commissions to set our rates high enough to cover the electricity costs for our datacenters," but Microsoft also pledged to minimize water use and reject new tax breaks below current data-center rates in exchange for selecting a site.

The announcement comes after "at least 25 data center projects were canceled last year following local opposition in the United States," according to Heatmap News, a huge increase compared to previous years. "One enduring lesson is that successful infrastructure buildouts will only progress when communities feel that the gains outweigh the costs," Smith wrote in the blog post, but those communities might have a hard time hearing him.


Enterprise funding

Cyera raised $400 million in Series F funding, which values the data security company at $9 billion.

Torq scored $140 million in Series D funding for its take on the security operations center, which uses AI agents to triage and respond to security alerts.

Deepgram landed $130 million in Series C funding for its voice-centric AI agents, and announced that it had acquired OfOne, a startup working on similar technology for restaurants.

WitnessAI raised $58 million in new funding for its agentic AI security platform.

Protege scored $30 million in Series A funding as it builds out its data aggregation and labeling services.

Flip landed $20 million in Series A funding for its customer service agents, which are designed for specific industries.


The Runtime roundup

A significant amount of code and documentation was stolen from one of Target's development servers after what appears to be a credential leak, according to BleepingComputer.

Meta created a new "top-level initiative" called Meta Compute, which will be responsible for building "tens of gigawatts" of data-center capacity this decade, CEO Mark Zuckerberg said on Threads.


Thanks for reading — see you Thursday!

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to Runtime.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.