The agents will continue until revenue improves

Welcome to Runtime! Today: Salesforce continues its agentic AI push, Databricks secures one of the biggest funding rounds in tech history, and the rest of this week's enterprise funding.

(Was this email forwarded to you? Sign up here to get Runtime each week.)

Show of force

Like every enterprise software company, Salesforce put agentic AI at the heart of its product strategy in 2024 in hopes of unlocking the value that this industry has promised generative AI can deliver. It will take several months before we know if those efforts will pay off as companies build, integrate, and deploy AI agents, but that hasn't stopped Salesforce from declaring victory.

On Tuesday it unveiled Agentforce 2.0, a remarkably quick turnaround of a 2.0 version for a product that was just launched in September and became available in October. The new version isn't the great leap forward for Salesforce that such a cadence would imply, but introduces new agents for sales and marketing tasks and adds links into the bevy of tools Salesforce acquired over a decade-long spending spree, including Mutesoft, Tableau, and Slack.

Agentforce 2.0 does add new tasks for Sales Cloud customers that can help activate leads and provide feedback on conversations with prospects, according to Salesforce.
Mulesoft now allows customers to find APIs across Salesforce's properties as well as third-party services so they can link skills together across different workflows, and Tableau provides better analytics on agent performance.
Slack users can also ask questions of agents from within the platform and build agents that can send Slack messages.

Agentforce 2.0 relies on an improved version of Salesforce's "Atlas Reasoning Engine," named after an ancient Greek god condemned to hold up the heavens after taking the losing side of a war. Much like Atlas himself, Salesforce's agentic AI push is bearing a lot of weight as the company tries to jumpstart its revenue growth, which has slipped into single-digit percentages this year.

The company said Atlas is now capable of handling more complicated questions than the previous version, such as asking for advice on the best investment strategy rather than checking the status of a portfolio, in one example provided by Salesforce.
Customers can also now add metadata from across Salesforce's properties to unstructured data found using RAG (retrieval-augmented generation) techniques to provide additional context specific to their businesses.
"Results from our customers in field testing show it’s about twice as accurate than other techniques,” Salesforce's Adam Evans told SiliconAngle.

Salesforce CEO Marc Benioff was already crowing about the impact of Agentforce on his company's trajectory during its last earnings call, which covered the three-month period ending just a few days after the technology became generally available. On Tuesday, he told attendees at the Agentforce 2.0 launch event that the company plans to hire 2,000 new salespeople to push the technology.

Left unspoken, of course, was the fact that there are lots of companies looking at the rise of generative AI, coding assistants, and agents with thoughts of building their own internal enterprise software tools rather than spending money with companies like Salesforce.
And even if that proves too hard, every platform shift in enterprise technology has produced upstarts born around the new technology that moved faster than incumbent vendors to capitalize on those shifts.
If Salesforce and its other agentic AI antagonists hope to continue running the day-to-day operations of enterprises around the world, they'll need to prove that products like Agentforce will make it easy for companies to save money or improve productivity using generative AI.
That could be a challenge: "As AI use cases call for more and more queries into the core Salesforce platform, performance and availability may become an issue as some users are already citing that the more they leverage complex AI use cases, the number of round trips between the Atlas reasoning engine, other data systems, and Salesforce and data cloud are causing some performance issues," according to Constellation Research analyst Martin Schneider.

Show me the money

Databricks has been flirting with the public markets for the last several years during a period of huge growth thanks to demand for its data analytics tools, but the timing always seemed off amid war in Ukraine, a pullback in tech spending, and rising interest rates. It just ensured that a long-awaited IPO can wait a little longer.

In one of the largest funding rounds ever raised in Silicon Valley, Databricks secured $8.6 billion out of the $10 billion Series J round it expects to raise in due course, the company announced Tuesday. The round values the company at $62 billion, which exceeds the $56.4 billion that longtime rival Snowflake was worth following the close of the stock market on Tuesday.

The round is "non-dilutive," which means it allows Databricks employees to cash out of their holding much like an IPO would have done. The company also announced that it is on track to record $3 billion in annual revenue by the end of its fiscal fourth quarter in January and grew at a 60% clip during the third quarter.

Enterprise funding (non-Databricks version)

Crusoe raised $600 million in Series D funding and launched its GPU cloud service, which was built around data centers sited next to energy sources.

Liquid AI landed $250 million in Series A funding led by AMD Ventures, which will allow the model-research company to "to scale our compute infrastructure" presumably on AMD's chips.

Zest AI scored $200 million in new funding as it expands the reach of its AI lending software, which helps banks assess the creditworthiness of potential customers.

Sublime raised $60 million in Series B funding for its email security software, which helps companies when they've got to find a reason why their money's all gone.

Keepit scored $50 million in new funding as it builds out its data protection and backup software for SaaS applications.

Stigg landed $17.5 million in Series A funding for its billing software, which helps SaaS providers switch to new billing systems without wrecking their sales pipeline.

The Runtime roundup

Hackers were able to steal more than 390,000 login credentials from security professionals over the last year by tricking them into downloading compromised packages from GitHub and npm, according to Ars Technica.

AWS's Jeff Barr is stepping back after 20 years as the company's lead blogger, during which he introduced the public to thousands of new AWS services and upgrades.

Thanks for reading — see you Thursday!