Conversation credit metric, Data Cloud overlay, Einstein 1 bundle, persona rollout framework, and the buyer side moves across the 2026 Agentforce renewal cycle.
25 minute call. No follow up sales pressure unless you ask for one. We will tell you what we would do next on your renewal, audit, or contract.
Agentforce is the Salesforce agentic AI platform launched in late 2024 and scaled through 2025 and 2026. The product is sold on a per conversation credit metric attached to a Service Cloud, Sales Cloud, or Data Cloud foundation. The economics depend on the conversation pool sizing, the Data Cloud overlay, and the persona segmentation, not on the brochure rate.
Agentforce is the Salesforce agentic AI platform. The product launched at Dreamforce 2024 and scaled through the 2025 spring, summer, and winter release cycles. By the 2026 spring release, Agentforce supports service agents, sales agents, marketing agents, commerce agents, and the wider custom agent framework.
This knowledge hub anchors the buyer side view across the Agentforce pricing metric, the Data Cloud overlay, the persona pilot framework, and the renewal moves that hold the conversation credit pool honest.
Read the related Salesforce knowledge hub, the Salesforce advisory practice, and the Salesforce renewal playbook for the wider 2026 framework.
Agentforce is the umbrella for the Salesforce agentic AI tooling. The platform includes pre built agents for service, sales, marketing, and commerce, plus the Agent Builder for custom agent development on top of the underlying Atlas reasoning engine.
The Atlas engine is Salesforce's own agent runtime. The platform also supports model selection across the major large language model providers, with the customer choosing the underlying model per agent.
The Atlas engine is the Salesforce native reasoning runtime that orchestrates the agent across data sources, large language models, and action surfaces. The engine handles the agent reasoning loop, the retrieval, the planning, and the action execution.
Atlas is a Salesforce proprietary runtime. The platform abstracts the underlying large language model choice, which lets the customer mix and match models across agents without rewriting the agent logic.
Agentforce is priced on a per conversation credit metric. A conversation is defined as a complete agent interaction from initiation through to resolution or handoff. The published 2026 rate sits at 2 USD per conversation for the Service and Sales personas.
The credit pool is typically attached to the underlying Service Cloud, Sales Cloud, or Data Cloud subscription. The pool is sized against a forecast monthly conversation volume, with overage above the pool billed at the higher pay as you go list rate.
The conversation credit pool typically carries a per pool discount against the standalone list rate. A larger annual pool unlocks a deeper discount. A multi year pool unlocks a further discount.
The trade off is the same as any prepaid credit pool. Overcommit and the customer pays for unused credits at the end of the term. Undercommit and the overage rate destroys the unit economics.
Agentforce 2026 commercial framework
| Component | Metric | 2026 list rate | Typical fit |
|---|---|---|---|
| Service conversation | Per resolved conversation | $2.00 | Service Cloud overlay |
| Sales conversation | Per resolved conversation | $2.00 | Sales Cloud overlay |
| Custom conversation | Per resolved conversation | $2.00 to $3.50 | Agent Builder use cases |
| Einstein 1 Service | Per user per month | $500 | Bundled Service plus Agentforce |
| Einstein 1 Sales | Per user per month | $500 | Bundled Sales plus Agentforce |
| Data Cloud credit | Per credit consumed | Published rate card | Required foundation |
Agentforce sits on a Data Cloud foundation. The customer data platform layer harmonizes the underlying customer, product, transaction, and content data into the unified profile that the agent reasons over.
Without Data Cloud, the agent has no context. The agent answer quality collapses, the hallucination rate climbs, and the rollout fails. Data Cloud is a precondition for Agentforce, not an optional add on.
The Service Cloud rollout typically lands on three personas. The Tier 1 agent assist case where Agentforce drafts the response and the human agent reviews. The Tier 2 case deflection where Agentforce resolves the case without human involvement. The Tier 3 supervisory case where Agentforce coordinates the multi step resolution.
The early enterprise rollouts of 2025 and 2026 anchored on the case deflection persona because the unit economics are clearest. A resolved case at 2 USD compares against a fully loaded human agent case at 8 to 15 USD.
The Agent Builder supports custom agents for use cases beyond the pre built service and sales personas. Common 2026 custom agent use cases include claims processing, IT service desk, HR self service, and partner onboarding.
Custom agents commonly run on the higher per conversation credit rate, reflecting the additional inference complexity. The unit economics scale with the volume of the underlying workload.
The single most common Agentforce trap is the over committed conversation credit pool. The Salesforce account team forecasts the persona rollout against an enthusiasm based volume. The customer signs the pool. The persona rollout takes longer than planned. The pool sits half consumed at year end.
The buyer side discipline is to right size the initial pool against a conservative pilot forecast, with a contractual expansion clause that triggers on documented adoption evidence.
Agentforce abstracts the underlying large language model choice. Some models are more expensive per inference than others. A custom agent on a premium model can multiply the per conversation cost beyond the published 2 USD rate.
The buyer side move is to monitor the model selection per agent and to require the underlying model cost transparency at renewal.
Agentforce is not a flat per seat product. It is a credit pool layered on a Data Cloud foundation. The pool sizing and the foundation right sizing are where the negotiation lives, not the per conversation rate.
The Agentforce rollout requires monthly conversation volume monitoring, persona adoption tracking, Data Cloud credit burn review, and model cost transparency.
Pull in the Salesforce advisory practice for the joint negotiation posture and the operating cadence that holds the Salesforce account team honest.
A 1,800 seat Service Cloud customer in financial services ran a 90 day Agentforce case deflection pilot across the Tier 1 retail banking inquiry persona. The pilot resolved 38 percent of inbound cases without human intervention at an average cost of 1.82 USD per resolved case.
The recommendation rolled Agentforce to the full Tier 1 retail banking persona of 700 seats with a conservatively sized conversation pool. The contractual expansion clause covered the Tier 2 corporate banking persona on documented adoption evidence. The first year commitment came in at 42 percent of the Salesforce account team proposal and the unit economics held through the year.
A 350 seat Sales Cloud customer in industrial distribution ran a 60 day Agentforce sales pilot on the inbound qualification persona. The pilot qualified 4,200 inbound leads at an average cost of 1.65 USD per qualified lead, against a manual cost of 9 USD per qualified lead.
The recommendation committed Agentforce on the inbound qualification persona with a conservative conversation pool, deferred the follow up automation persona to the next pilot cycle, and saved 64 percent against the Salesforce account team proposal that bundled all four sales personas at year one.
Agentforce is the Salesforce agentic AI platform launched in September 2024. The product includes pre built agents for service, sales, marketing, and commerce, plus the Agent Builder for custom agents on the Atlas reasoning engine.
Agentforce is priced on a per conversation credit metric, with the published 2026 rate at 2 USD per resolved conversation for the Service and Sales personas. Custom agents commonly run at 2 to 3.50 USD per conversation.
Yes. Data Cloud is the data foundation that Agentforce reasons over. Without Data Cloud, the agent answer quality collapses. Einstein 1 Service and Einstein 1 Sales include a starter Data Cloud allocation per user.
Einstein 1 Service is the bundled top tier edition that includes Service Cloud, Data Cloud, and a starter Agentforce conversation credit pool, priced at 500 USD per user per month list.
Atlas is the Salesforce proprietary reasoning runtime that orchestrates Agentforce across data sources, large language models, and action surfaces. The engine handles the agent reasoning loop, retrieval, planning, and action execution.
Yes. Agentforce abstracts the model choice. The customer can select among the supported major model providers per agent. The choice affects the per conversation cost depending on model premium.
The over committed conversation credit pool. The Salesforce account team forecasts an enthusiasm based volume, the customer signs the pool, the rollout takes longer than planned, and the pool sits half consumed at year end.
Run a documented persona pilot, capture the per conversation unit economics, compare against the manual baseline cost, and benchmark the overall commitment against the Salesforce knowledge hub reference rates.
Salesforce renewal posture, Agentforce conversation credit framework, Data Cloud overlay, persona pilot pricing, and the buyer side moves across the full Salesforce estate.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.
Salesforce wanted us to commit to a 4 million US dollar Agentforce conversation pool in year one. The independent pilot showed we would absorb 1.6 million. Redress structured a conservatively sized pool with an adoption gated expansion clause and the year one commit landed at 1.7 million with a documented path to the higher commit on real evidence.
500+ enterprise clients. 11 vendor practices. Industry recognized. One conversation can change what you pay for the next three years.
Agentforce conversation credit framework, Data Cloud overlay math, persona pilot pricing, and the wider Salesforce leverage signals across the practice.
