Salesforce's Dreamforce pitch: let agents replace the UI
Two stories from today worth your time: Salesforce arguing that its own interface is the thing standing between customers and its data, and Huawei's chairman saying out loud what his compute business is aiming at.
Salesforce opened Dreamforce by launching AIforce, a layer that pushes the company's data, permissions and business logic out of its own UI and into whatever AI tool an employee already has open. The idea, per Salesforce president Patrick Stokes, is that the value was never in the interface: it was in the platform holding the way customers encode their business, and the interface mostly got in the way. AIforce ships in three pieces — Claudeforce, which packs Salesforce into a prebuilt MCP server inside Claude with 37 prebuilt sales skills, now in beta; Slackforce, which adds Slack CRM and a Slackbot that reasons over Salesforce context without a dashboard; and Agentforce Coworker, an AI teammate inside Lightning that Salesforce says 100,000 users switched on in its first 35 days. The company also claims zero data retention: business data answers the question at hand and is not kept by the model provider.
Why it matters: every enterprise vendor now says "AI replaces the UI," and what that translates to is "our data model replaces our UI." If the store of record is the moat, the interface stops being the lock-in — agents read across hundreds of records and pull in connected systems, and the CRM becomes plumbing that any surface can call. That is a real simplification for buyers, and just as real a claim on their loyalty. It is also why the accompanying AgentExchange partner list matters more than the keynote: Anthropic, AWS, Google, Lovable and Vercel building on Salesforce's exposed toolbox is Salesforce betting its distribution on other companies' interfaces. Marc Benioff's framing — "AI is creating an interface revolution" — is a product strategy with an escape hatch built in.
Huawei supervisory board chairman Guo Ping told new employees that in ICT and computing, the company's goal is to become Nvidia — and that getting there does not require Huawei to own a frontier model. The remarks surfaced Tuesday in a roughly 10,000-character forum transcript published on Huawei's internal Heart & Soul community. Guo's argument is that Huawei's advantage is not a competing large model but support: making sure any major model in the world runs efficiently on Ascend, Kunpeng, supernodes and clusters. The strategy is differentiated by business, he said — device and intelligent-driving units should have their own models, since open releases only ship weights and data lags behind, while Huawei Cloud must have a competitive in-house model to stay relevant. He also conceded the constraint that shapes all of it: Huawei is limited on advanced process nodes, so it leans on integrated design and manufacturing, with the Ascend 950 series narrowing the gap through architecture and, in his words, a chance to catch up later.
What to watch: whether Ascend 950 shipments arrive on the schedule the memo implies, and whether Chinese labs keep their models portable across Huawei and Nvidia silicon now that Huawei is explicitly selling itself as the neutral substrate.
Huawei wants to be the neutral layer under everyone's model — does a vendor that competes with its own customers' models ever get to be neutral? Tell us in the comments.
