Salesforce News lays out a vision of AI that turns the “app UI” into an agent-led interface

Salesforce is making a case that the interface layer of business software is changing, not just the models behind it. In a June 29 story, the company frames a future where AI agents handle the navigation and execution of work inside the tools people already live in, reducing the need to open separate apps or drill through dashboards.

The story points back to Apple’s 1987 “Knowledge Navigator,” a short film that imagined a professor working at a desk with an agent that could switch between voice and tablet-like visuals without prompts. The author uses the contrast to argue that today’s work still often follows an old pattern, where software is treated as destinations you visit one by one, whether that means checking a dashboard or running a spreadsheet.

From there, Salesforce describes a different interaction model: agentic AI embedded into everyday collaboration and workflow tools. It says work can increasingly happen “within the tool you’re already using,” where a user can ask for analysis, request a chart, or generate output without opening separate programs for each step.

The company also cites a set of examples meant to show where agents may run. It mentions open-source agents OpenClaw and Hermes for executing instructions in messaging contexts like Slack, WhatsApp, or Telegram. It further references Claude Code as an approach that can generate an interface itself, arguing that a system that builds interfaces on demand can adapt what it shows to a user’s needs and then discard it when the task is complete.

Salesforce’s core claim is that the shift changes what software interfaces are for. Instead of standing alone as the place where work gets done, apps become capabilities that surface through agents, conversations, and workflows. In that framing, “menus and dashboards slide into the background,” while the user’s interface becomes the agent conversation or the workspace where the agent delivers completed actions or a decision requiring human response.

The story illustrates this idea with workplace scenarios. It describes a manager approving an expense report inside a chat window rather than opening an expense application, and an analyst pulling sales data into a presentation without logging into a dashboard that stores it. It adds that as AI takes on more execution, human attention shifts toward supervision, review, and governance instead of data gathering and step-by-step navigation.

Salesforce also uses a customer example to ground the shift toward speech-based interaction. It says water treatment inspectors at a Salesforce customer, Workdry Group, previously spent two to four hours after plant visits writing up compliance surveys, and that Salesforce’s “Voice to Form” routes spoken details into the correct fields. The story claims the job can now be finished on site in about 20 minutes, even while inspectors are wearing safety gloves and working in a noisy environment.

Beyond voice, the story broadens into what Salesforce’s chief scientist calls “ambient intelligence,” describing software that watches the work continuously, understands context, and reaches users before they go looking. It argues the interface may become less about where a person goes and more about how the software meets them, whether through a phone or smartwatch, and through AI agents that can coordinate tasks. In Salesforce’s own platform terms, it points to “Headless 360,” saying it lets an agent, with permission, reach into the Salesforce platform to act on a customer’s full history without requiring a person to log into the usual entry points. The company co-founder Parker Harris is quoted as saying, “Why should you ever log into Salesforce again? Maybe you never will. Maybe you will go into Slack.”