Salesforce takes on hyperscalers with Agentforce 2dx updates

Salesforce’s updates to its agentic AI offering — Agentforce — released this week could give the CRM software provider an edge over its enterprise application rivals and hyperscalers including AWS, Google, IBM, Service Now and Microsoft.

Currently, hyperscalers and enterprise software providers combined dominate 60% of the agentic AI market and Salesforce has a significant presence within this dominant group, according to Vershita Srivastava, practice director at Everest Group.

The updates, grouped under the name Agentforce 2dx, introduce new agent orchestration, development, testing, and deployment capabilities, which are expected to help Salesforce leapfrog over its rivals in ways that are analogous to OpenAI’s early leadership in the space of large language models (LLMs), said Arnal Dayaratna, vice president of research at IDC.

Agentforce API to increase extensibility of the platform

One of the key updates of the new version — the Agentforce API — is expected to increase the extensibility of the Agentforce platform and has the potential to streamline the integration of agentic technologies into digital solutions in ways that consolidate Salesforce’s early leadership in the agentic AI space, according to analysts.

“Without an API you need to think that all of the AI agentic capabilities are locked into the Agentforce platform. The API allows enterprises to build apps and agents with whatever they want,” Jason Andersen, principal analyst at Moor Insights & Strategy, explained.

API can be seen as a strategy to drive adoption by eliminating constraints on where it can be used, according to The Futurum Group’s CIO practice lead Dion Hinchcliffe.

This should make Agentforce omnipresent, easily accessible, and allow enterprises to use it for a variety of use cases, Hinchcliffe explained.

However, Andersen pointed out that building out an API to increase extensibility of an offering is not new and many rivals are already offering APIs for their agents, such as Google and Microsoft.

But Hinchcliffe believes that Salesforce has an edge over rivals as it is leveraging its deep CRM expertise, its customers’ sales, service, and marketing data, and business logic integration to drive differentiation.

“Without integration, AI agents are starved of contextual data that they need in order to act,” Hinchcliffe explained, adding that rivals are likely to respond by improving their own integrations, but Salesforce’s strength and deep bench in business logic, integration, and process automation will be difficult to match quickly.

Other updates, such as making features, such as the Topic Center, MuleSoft integrations, Tableau Semantics, and Slack integrations, generally available are expected to help enterprises cut down on the time and complexity of building a new custom agent, integrating it into a workflow, improving agent responses, and deploying them on Slack.

Wooing developers to increase agent adoption

As part of Agentforce 2dx, Salesforce has introduced a series of updates that are aimed at wooing developers by giving them more control to build, test, and deploy agents.

Some of the key updates includes Agentforce support for VS Code, a DX inspector, the Interaction Explorer, and a Salesforce Developer Edition that packages an environment for developers to test out agents.

While the support for VS Code and the Developer Edition itself is expected to woo professional developers, tools such as the DX inspector and Interaction Explorer can be used to understand an agent’s behavior pre and post deployment.

Expanding on Salesforce’s strategy behind launching the Developer Edition, analysts believe that the CRM software provider is trying to lower the entry barrier for developers, who are often considered as decision makers of what gets adopted in an enterprise.

“By offering a free environment complete with access to Data Cloud, a dedicated data space, and a quota of LLM generations per hour, Salesforce is encouraging hands-on experimentation,” said Cameron Marsh, senior analyst at Nucleus Research.

However, Moor Insights & Strategy’s Andersen sees the move as a page out of cloud service providers’ playbook, who offer low or no cost methods to try out AI with the aim of converting an enterprise to a high-paying customer.

In addition, Andersen believes that these updates are a “bold move,” at least in the software-as-a-service (SaaS) market.

“What we have been seeing with agentic AI is that several platforms are just augmenting the platform experience. By extending the capabilities of Agentforce, Salesforce is suggesting that it can become a viable competitor to Azure, AWS, and Google Cloud, who also offer developer-centric tools for building agents,” Andersen said.

As part of the latest updates, Salesforce is also offering a low code tool — Testing Center — to help enterprises test agents before being deployed.

Analysts also believe that these tools and updates will help enterprises achieve scale when deploying and managing agents.

“One of the biggest problems with bringing agents into public deployment, especially for a customer base, is in simulating and testing the hundreds or thousands of potential interactions. With these capabilities, companies no longer have to worry about doing this testing and ongoing management on a manual basis or building a custom tool or process,” said Hyoun Park, chief analyst at Amalgam Insights.

Salesforce claims that it has closed around 5,000 deals with its Agentforce offering and claims that its customers, such as The Adecco Group, Engine, OpenTable, Oregon Humane Society, Precina, and Vivint have realized some immediate value post adoption.