
Salesforce's AI for employees is now built around Agentforce, its platform for AI agents that answer questions and take actions inside Salesforce, alongside Einstein predictive AI and AI in Slack. Older names such as Einstein GPT, Einstein Copilot and Slack GPT have been retired or folded into Agentforce and Slack.
The productivity case is promising but still mostly self-reported. In Slack's Workforce Lab survey of 10,281 desk workers (January 2024), around 80% of those using AI said it was already improving their productivity. This guide explains what each Salesforce AI product does today, where it helps sales, service, marketing and development teams, and how to measure whether your employees actually use it.
Salesforce has renamed its AI products several times since 2023. Here is how the older names map to what Salesforce sells today.
| Older name | Current name | What it does today |
|---|---|---|
| Einstein GPT, Einstein Copilot | Agentforce | AI agents that answer questions, create action plans and carry out tasks in Salesforce. Salesforce describes Agentforce as "formerly Einstein Copilot". |
| Slack GPT | AI in Slack | Slackbot as a personal AI agent, channel and thread summaries, daily recaps, enterprise search and AI meeting notes. |
| Einstein for Developers, Agentforce for Developers | Agentforce Vibes | Agentic AI tools for Salesforce developers, with chat that can take actions and inline code completion for Apex and Lightning Web Components. |
| Einstein (predictive AI) | Einstein | Still embedded in the Salesforce Platform for insights, predictions and generated content in the flow of work. |
Salesforce AI is designed to take routine work off employees' plates, suggest next steps from CRM data, and speed up common workflows. Here is what that looks like by team, based on what Salesforce documents for each product.
Most of these figures and descriptions come from Salesforce and Slack themselves. Your own results will depend on which agents you deploy and how many employees use them, which is why measuring adoption matters.
Even the best AI tools deliver nothing if employees do not use them or do not trust them. In Slack's Workforce Lab survey, desk workers at companies with defined AI guidelines were nearly six times more likely to have tried AI tools, and 43% said they had received no guidance from their organization on using AI at work. Leaders and HR should give clear direction on which Salesforce AI features to use and for what, back it with role-specific AI training, and fold it into a wider AI adoption strategy.
One of the main challenges of AI adoption is data privacy and security. Many employees want to use generative AI but are unsure how to do so responsibly. Teach your workforce to verify AI outputs, to keep confidential information out of unapproved tools, and to understand where AI is unreliable. For agents that act on customer records, add human review checkpoints for AI-generated content, test agents before wide release, and set clear policies on what data agents can access.
Rolling out Agentforce or AI in Slack is not enough. Managers need to ask whether it is actually improving productivity and outcomes, and that takes data.
Worklytics helps companies see how work gets done, including how AI tools are used across the workforce, using data from the tools teams already use.
Worklytics is privacy-first. It analyzes metadata from work tools, never message or prompt content. Personal identifiers are pseudonymized before analysis, and results are shown only for groups above a minimum size.
Salesforce's AI has moved from assistants that draft text to agents that take action, under the Agentforce name, with AI in Slack and Agentforce Vibes alongside it. The tools can take routine work off sales, service, marketing and development teams, but the evidence so far is mostly self-reported or vendor-reported. Treat your rollout as something to measure: track which teams use Agentforce and AI in Slack, compare that with how their work changes, and invest more where the gains show up.
Agentforce is Salesforce's platform for building and running AI agents that answer questions and take actions on behalf of employees and customers. It includes prebuilt agents for service, sales, marketing and commerce, and Agent Builder for creating custom agents. Salesforce describes Agentforce as formerly Einstein Copilot.
Yes. Salesforce renamed Einstein Copilot to Agentforce, and earlier Einstein GPT features are now part of the Agentforce product line. Einstein remains the name for Salesforce's predictive AI built into the platform.
Slack GPT was the 2023 name for generative AI in Slack. Today Slack calls it AI in Slack, which includes Slackbot as a personal AI agent, channel and thread summaries, daily recaps, enterprise search and AI meeting notes.
Early evidence is positive but mostly self-reported. In Slack's Workforce Lab survey of 10,281 desk workers, around 80% of those using AI said it was improving their productivity. The best way to know for your company is to measure adoption by team and compare it with how work changes over time.
Start with Salesforce's built-in Agentforce analytics to see how agents perform. Then track adoption by team and role, and compare it with work outcomes such as cases handled or pipeline activity. Measure active users rather than licenses, so you can see where agents are not yet used.