AIforce: Salesforce’s New AI Interface Layer for Slack, Claude, and Lightning

Dreamforce ’26 just wrapped up, and if you’ve been anywhere near LinkedIn this week, you’ve seen one phrase everywhere: AIforce Salesforce. It was one of the biggest announcements to come out of the event, so let’s slow down and go through what it actually is, what it isn’t, and why Salesforce people should care without the keynote hype.

Everything in this article is based on what Salesforce has officially published, and we’ve linked the original sources at the end of each section so you can verify it yourself instead of just taking our word for it.

What Is AIforce, in Simple Words?

Here’s the simplest way to put it: AIforce is not a new app. You won’t log into “AIforce” the way you log into Salesforce.

Instead, think of it as a layer that sits on top of your existing Salesforce org like your data, your workflows, your permissions, your automation rules and makes all of it usable from wherever you already work. That could be Claude, Slack, or the normal Salesforce Lightning screen you use every day.

So instead of opening Salesforce to check an account, updating a record, or running a report, you could just ask for it inside Slack or Claude, and get an answer that respects the same permissions and rules your org already has. Salesforce describes it as turning “the Salesforce you already have” into a live, intelligent interface, rather than asking you to adopt a separate product.

Official source: AIforce announcement, salesforce.com

A Quick Timeline, Because This Didn’t Appear Out of Nowhere

If you’re wondering why this feels like it came together fast, here’s the actual sequence:

• August 26, 2026 — Salesforce and Anthropic announced an expanded partnership under the name Claudeforce, bringing Claude’s reasoning together with Salesforce’s data, workflows, and governance. This was the first public use of the “AIforce” name, described at the time as the technology powering Salesforce in Claude.
• September 15, 2026 — At Dreamforce, Salesforce widened the scope. AIforce was introduced as the full interface layer, launching alongside two more surfaces: Slackforce and Agentforce Coworker.

So AIforce didn’t launch as one big-bang product . It started with the Claude partnership and grew into a three-surface layer by Dreamforce.

Official source: Salesforce and Anthropic announce Claudeforce, salesforce.com press release

The Three Things AIforce Launched With

AIforce isn’t one single product. It launched with three separate pieces, and each one targets a different place people already spend their day.

Claudeforce — Salesforce Inside Claude

Claudeforce connects Claude directly to your Salesforce data through a prebuilt setup, so a sales rep or admin can ask Claude about an account, a deal, or a case without switching tabs. It ships with 37 ready-made sales skills ,things like prospecting help and pipeline hygiene checks are so common sales questions work right away instead of needing custom setup.

There’s also a separate piece aimed at developers: a Salesforce Development plug-in for Claude Code, with more than 40 skills built in, meant to help with actual Salesforce development work like writing Apex, building Flows, and similar tasks directly inside Claude Code.

Salesforce has also said Claudeforce was piloted with a small group of companies, including Deloitte and GitLab, before the wider Dreamforce announcement.

Also read out How to Connect Claude to Salesforce Using MCP Servers: The Complete Setup Guide

Official source: Claudeforce press release, salesforce.com

Slackforce — Salesforce Inside Slack

If your team basically lives in Slack, this is the piece that matters most to you. Slackforce brings Salesforce context into Slack conversations and lets your team act on it without leaving the chat. A few specific things it includes:

• Surfaces — live, interactive views built from Salesforce and Slack data that a whole channel can look at and act on together, instead of one person screen-sharing a Salesforce tab.
• Slackbot — a reasoning assistant that can answer questions using both Slack conversation history and Salesforce data together.
• Slack CRM — the ability to create or update Salesforce records straight from a Slack message, instead of opening Salesforce separately.

Salesforce has also pointed out that Slack’s own internal team already uses Claude-powered Slackbot heavily, reporting a large jump in internal productivity hours from it which is part of why they’re confident enough to ship this externally.

Official source: AIforce product page, salesforce.com

Agentforce Coworker — An AI Teammate Inside Lightning (and Beyond)

This one doesn’t ask you to change tools at all at least, not at first. Agentforce Coworker lives right inside the normal Salesforce Lightning screen you already use. It can look across a customer’s accounts, past activity, and history, and call on other Agentforce agents your company has already built all while sticking to the exact same permissions that already apply to you.

What’s interesting is where this is headed next: Salesforce has said Agentforce Coworker is planned to expand beyond Salesforce itself, into Slack, Microsoft Teams, Claude, ChatGPT, and a dedicated desktop app. So the same AI teammate that starts inside your Salesforce org is meant to eventually follow you into whichever tool you’re actually working in that day.

Salesforce’s own numbers here: Agentforce Coworker reportedly had 100,000 users activate it within its first 35 days of early access.

Official source: Agentforce Coworker product page, salesforce.com · Salesforce Admins blog on Agentforce Coworker

Why Salesforce Is Doing This

The short version: most people don’t want to open Salesforce to get a Salesforce answer. They want the answer wherever they already are like in a chat app, in an AI assistant, in whatever tab is already open.

AIforce is Salesforce’s answer to that. Instead of trying to pull everyone back into one interface, it pushes Salesforce’s data and logic out to the interfaces people already prefer. Salesforce calls it a “live interface layer” because the same underlying data updates in real time no matter which surface you’re looking at it from , a change made in Slack shows up instantly if you check Lightning right after, and vice versa.

Under the hood, this is powered by something called the Headless Toolkit an open way for outside tools (like Claude or Slack) to talk to Salesforce properly, using APIs and MCP (Model Context Protocol) connections instead of a custom one-off integration for every single tool. Salesforce has described the Headless Toolkit as the shared foundation behind both AIforce and its broader partner ecosystem, AgentExchange.

Official source: AgentExchange ecosystem FAQ, salesforce.com

What About Data Safety?

This is usually the first question anyone with a security or admin background asks, and it’s a fair one. Salesforce says AIforce follows Zero Data Retention (ZDR) meaning the AI models involved don’t keep your company’s data afterward, and it isn’t used to train those models. Admins also don’t need to set up a separate permission system for this, it’s built to respect whatever sharing rules and permissions your org already has, so a user still can’t see through AIforce what they couldn’t already see in Salesforce directly.

Is This Replacing Agentforce?

No and this is worth clearing up, because the names sound similar. Agentforce (the AI agents that handle tasks like case resolution or lead qualification) is still its own separate thing, and Salesforce has confirmed it isn’t folding into AIforce. AIforce is about where you access Salesforce and its agents from, not a replacement for the agents themselves. In fact, Salesforce frames AIforce as one layer inside a bigger picture it calls the Agentic Enterprise architecture, which also includes Agentforce, Data 360, and Customer 360 as separate but connected pieces. Agentforce Coworker is simply one of the three ways AIforce shows up.

What This Means If You Work With Salesforce

If you’re an admin, developer, or consultant, this is worth watching closely over the next few months. A few practical things to keep in mind:

• Claudeforce (Salesforce in Claude) is in beta for all customers as of the Dreamforce announcement. Agentforce Coworker is also in beta, expanding to more surfaces over time.
• No public pricing has been announced yet for AIforce as a whole , this is something to watch for, since it will affect how quickly orgs actually adopt it.
• If your org relies heavily on Slack already, Slackforce is likely to matter the most for your day-to-day work.
• Permission and sharing rule design just became more important, not less since the same access rules now control what people see across three different tools instead of one. If your org’s sharing model has gaps today, those gaps will now show up in more places at once.
• This is a good concept to be ready to explain in interviews soon, especially for admin and consultant roles — expect “what is AIforce” to start showing up as a screening question over the next few hiring cycles.

Want to Stay Ahead on Salesforce AI Updates?

👉 Book a mentorship session for free: Click here

More Salesforce and Agentforce content at triggerhours.com