Salesforce has expanded Einstein Copilot to every paid Sales Cloud, Service Cloud, and Platform tier, ending months of speculation about whether the AI assistant would remain an upsell-only feature. The change takes effect August 1 and was confirmed in an internal communications memo that circulated among Salesforce admins this week.
The announcement affects roughly 150,000 organisations running Salesforce on Professional, Enterprise, or Unlimited plans who have until now been locked out of Einstein features unless they purchased Copilot as a separate add-on — typically priced at $50 per user per month on top of existing licence costs. That add-on cost going away sounds significant. The more important question is what remains paywalled — and the answer is nearly everything mid-market teams actually want from AI.
What's actually included
The expanded tier covers three core Einstein Copilot capabilities: conversation summarisation, pipeline coaching, and deal risk alerts. Conversation summarisation automatically condenses call recordings and email threads into structured CRM entries. Pipeline coaching surfaces deal-specific recommendations based on historical win/loss patterns. Deal risk alerts flag stalled opportunities based on engagement drop-off and competitor activity signals.
What is not included in the expanded standard tier is the full Agentforce suite — autonomous AI agents that can initiate outreach, update records, and execute multi-step workflows without human prompts. Agentforce remains a separate product with its own pricing, and Salesforce has been careful to position it as a premium add-on distinct from the core Copilot capabilities now being standardised.
What Agentforce still can't do — and what it costs when it can
Agentforce is Salesforce's headline AI product, and it has been marketed aggressively since its launch. But the reality of what Agentforce delivers today versus what the demos suggest is materially different. Autonomous agents require significant configuration through Salesforce's Flow and Prompt Builder tools — setup that typically runs 80 to 200 hours of implementation time for even a basic outreach workflow. The agents do not come pre-trained on your business context; they must be taught what your pipeline stages mean, how your team qualifies deals, and what a good outreach sequence looks like.
The cost picture adds up quickly. Agentforce is priced at $2 per conversation for the standard tier, which sounds modest until you model it across a 50-person sales team generating 2,000 agent interactions per month — a figure that lands around $4,000 monthly on top of existing Sales Cloud licensing. Enterprise accounts with high interaction volumes can negotiate flat-rate contracts, but mid-market buyers rarely have that leverage. Salesforce's own pricing calculator, which the company does not prominently advertise, makes these costs clearer.
Beyond cost, Agentforce's AI reasoning capabilities are constrained to data that exists inside Salesforce. It cannot natively pull external signals — news events, intent data, company hiring trends — without additional data connections built through Data Cloud, itself another licensed product. Teams that need AI to reason across their full go-to-market context will find Agentforce's walled-garden approach a significant limitation.
Why now
The timing is not incidental. Salesforce has faced sustained pressure from HubSpot, which has aggressively bundled AI features into its core tiers without additional charge, and from a growing set of all-in-one platforms that include AI as a foundational layer rather than a premium option. Several mid-market deal losses in Q1 were attributed internally to pricing friction around AI features, according to two Salesforce channel partners who spoke with CRM Today.
Salesforce CEO Marc Benioff has publicly framed the Agentforce rollout as the company's most significant product bet in a decade. Standardising the lower-tier Copilot features is widely read as a defensive move to protect the installed base while Agentforce establishes its own market position. The risk for Salesforce is that broadening access to basic AI features does not address the deeper concern: that its pricing architecture makes comprehensive AI adoption economically inaccessible for the mid-market.
What RevOps teams should do now
For organisations already on paid Salesforce plans, the practical steps are straightforward. Einstein Copilot will need to be enabled by a Salesforce Admin — it does not activate automatically. The setup process involves enabling the Einstein platform in Setup, assigning Copilot permissions to relevant user profiles, and configuring which objects the AI has access to read and write.
RevOps leaders should also audit their existing Einstein feature permissions before August 1. Salesforce has reorganised several permission sets ahead of the rollout, and some existing configurations may conflict with the new default states. Salesforce's own help documentation on this was updated on June 28 and is the authoritative reference.
The deal risk alert feature requires at least 90 days of historical opportunity data to generate meaningful signals, which means newly onboarded teams will see limited utility in the first quarter. Teams with longer Salesforce tenures will see these features light up more or less immediately.
The competitive context — and the mid-market calculus
The move lands as the CRM AI feature war intensifies across all major platforms. Microsoft announced Copilot Sales Coach for Dynamics 365 last month — a direct conversation coaching tool that sits inside Teams and surfaces deal-specific guidance during live calls. HubSpot's Scout prospecting AI has been in closed beta since March and is expected to roll out broadly to Sales Hub Professional and Enterprise users this quarter.
The pattern is consistent: AI capabilities that were priced as premium differentiators 18 months ago are becoming table stakes, compressed downward by competitive pressure and the declining marginal cost of inference. For buyers, this is a genuine improvement. For vendors, it creates pressure to find new premium surface area — which is precisely why Agentforce autonomous agents are being positioned as the next frontier rather than folded into the standard tier.
For mid-market teams — typically those with 20 to 200 salespeople and budgets that do not accommodate Salesforce's layered licensing model — the math on Salesforce AI remains challenging even after this announcement. The basic Copilot features landing in the standard tier are useful, but they represent the floor of what competitive AI-assisted selling requires in 2026, not the ceiling. Teams actively evaluating whether Salesforce's AI roadmap justifies its total cost of ownership should compare it against platforms where AI is architecturally central rather than added on top. Independent platform reviews offer a useful reference point for understanding where AI-native approaches differ structurally from Salesforce's bolt-on model.
Salesforce has not responded to a request for comment on the specific features included in the August rollout. The information in this article is based on the admin memo and conversations with two certified Salesforce implementation partners.