Salesforce AI Pricing: What It Actually Costs in 2026
A breakdown of how Salesforce prices its AI products — from per-conversation Agentforce fees to Einstein add-ons and Data Cloud credits.
Salesforce’s AI pricing is not straightforward. Products are bundled, metered, and add-on-ed in ways that make it genuinely difficult to forecast what you’ll actually pay. This post breaks down the current pricing structure so you can plan with real numbers.
The base: Salesforce edition pricing
Before any AI enters the picture, you’re paying per-user-per-month for your Salesforce edition. The most common starting points:
- Enterprise Edition: $165/user/month
- Unlimited Edition: $330/user/month
- Einstein 1 Edition (now called Unlimited Edition+): $500/user/month
The Einstein 1 / Unlimited+ tier is where Salesforce bundles in a baseline set of AI features — Einstein Copilot, generative AI in flows, and Data Cloud credits. If you’re on Enterprise or standard Unlimited, most AI capabilities are paid add-ons.
Agentforce pricing
Agentforce is Salesforce’s flagship AI product, and it uses a consumption-based model rather than per-seat licensing.
$2 per conversation. That’s the headline number. Each time an Agentforce agent engages with a customer or internal user and resolves a request, it counts as one conversation. Salesforce defines a conversation as a complete interaction — not per message, but per session.
What counts as a conversation depends on the channel and configuration:
- A service agent resolving a customer case via chat = 1 conversation
- A sales development agent qualifying an inbound lead via email = 1 conversation
- An internal agent answering an employee’s HR question = 1 conversation
Volume discounts
Salesforce offers negotiated rates at scale. Organizations running tens of thousands of conversations per month can typically get the per-conversation cost down to the $1.50–$1.80 range, though Salesforce doesn’t publish a formal volume discount schedule.
A poorly tuned agent that escalates most conversations costs more per resolved case than a human team. Configuration quality is the single biggest lever on Agentforce ROI.
The hidden variable: deflection vs. escalation
If an agent can’t resolve the issue and escalates to a human, it still counts as a conversation. This means your effective cost per resolution depends heavily on how well your agents are configured. A poorly tuned agent that escalates 60% of conversations is dramatically more expensive per resolved case than one that handles 90% autonomously.
Einstein Copilot and generative AI add-ons
If you’re not on the Unlimited+ tier, adding Einstein Copilot is a separate line item.
- Einstein for Sales: $75/user/month — includes Copilot for sales workflows, email generation, opportunity summaries, and call insights
- Einstein for Service: $75/user/month — includes Copilot for case management, knowledge article generation, and reply suggestions
- Einstein for Platform: $25/user/month — basic Copilot access for custom apps built on the Salesforce platform
These are on top of your base edition cost. A Sales Cloud Enterprise user adding Einstein for Sales goes from $165 to $240/user/month.
Data Cloud pricing
Data Cloud is essential for making AI features work well — it’s the unified data layer that gives models context. Pricing is credit-based:
- Data Cloud credits are consumed by data ingestion, identity resolution, segmentation, and activation
- Included credits vary by edition: Unlimited+ includes a meaningful baseline, Enterprise includes very little
- Additional credits are purchased in blocks, typically starting around $100K/year for meaningful volumes
The catch: if you’re running Agentforce at scale, your Data Cloud consumption goes up proportionally because agents pull from Data Cloud for context on every conversation.
Prompt Builder and Einstein Trust Layer
Prompt Builder is included with any Einstein add-on or the Unlimited+ edition. No separate cost, but it’s only useful if you’re paying for generative AI capabilities somewhere.
Einstein Trust Layer — the security and governance layer for AI interactions — is included with all Einstein AI products. Not a separate line item.
What a real deployment costs
Here’s a realistic scenario for a mid-market company with 200 Salesforce users:
| Component | Monthly cost |
|---|---|
| 200 users × Unlimited Edition ($330) | $66,000 |
| Einstein for Sales (50 reps × $75) | $3,750 |
| Einstein for Service (30 agents × $75) | $2,250 |
| Agentforce (~5,000 conversations × $2) | $10,000 |
| Data Cloud credits | ~$8,000 |
| Total | ~$90,000/mo |
Budget for the full stack, not just the headline rate. Data Cloud, integration work, and ongoing agent tuning are where the real costs hide.
That’s roughly $1.1M per year — a significant jump from a base Salesforce deployment. The Agentforce and Data Cloud lines are the variable costs that can swing significantly based on usage.
Negotiation leverage
A few things worth knowing when negotiating:
Multi-year commits. Salesforce typically offers 15–25% discounts on AI products for three-year agreements. The per-conversation Agentforce rate is the most negotiable line item.
Bundle pressure. Salesforce strongly prefers selling Unlimited+ over à la carte Einstein add-ons. If you’re buying multiple Einstein products, do the math — Unlimited+ at $500/user may be cheaper than Enterprise at $165 plus three $75 add-ons ($390 total, but without Data Cloud credits).
Pilot programs. Salesforce is still in aggressive adoption mode for Agentforce. Many organizations can negotiate free or deeply discounted pilot periods (3–6 months) for Agentforce deployments, especially for net-new use cases.
True-ups. Agentforce conversations are metered monthly. Unlike per-seat licenses, you can scale down without waiting for a renewal cycle — but only if your contract includes flexible consumption terms. Push for this.
The bottom line
Salesforce AI isn’t cheap, but the pricing model has a useful property: it’s tied to usage rather than seats for the most impactful products. An Agentforce agent that resolves 10,000 service cases per month at $2 each costs $20,000 — far less than the human agents it’s supplementing.
The risk is underestimating the supporting infrastructure costs. Data Cloud, integration work, prompt engineering, and ongoing tuning aren’t free. Budget for the full stack, not just the headline Agentforce rate.
If you’re evaluating Salesforce AI pricing for your organization, we can help model the total cost of ownership based on your actual usage patterns. Get in touch.