Salesforce Agentforce sells on a conversation metric. The 2026 list price is two dollars per conversation. The Einstein 1 bundle layers the metric on top of every Sales Cloud and Service Cloud seat at varying entitlement levels.
The pricing is consumption based but the commitment is annual. The buyer side discipline is to forecast conversations conservatively, to commit minimally, and to keep the overage rate in the contract.
Agentforce is positioned as the agentic AI layer for Salesforce. The reality in 2026 is that adoption sits in the early third of the curve, with most enterprise deployments running pilot to production transition. Pricing locked at 2026 list will look generous in 2027 once usage matures.
Pair this guide with the Salesforce knowledge hub, the Salesforce advisory practice, the renewal negotiation guide, the renewal play book, the Salesforce SELA article, and the license utilization calculator before the next Salesforce renewal window.
Key Takeaways
What a CIO needs to know in 90 seconds
- Two dollars per conversation list. The 2026 Agentforce list price for a single agentic conversation.
- Conversation defined. A conversation is an end to end interaction with the Agentforce service, not a single API call.
- Bundled with Einstein 1. Sales Cloud and Service Cloud Einstein 1 bundles include conversation entitlements.
- Annual commitment, monthly burn. The conversation pool is committed annually and consumed across the year.
- Overage rates matter. The contract overage rate is the single most expensive line if usage grows.
- Forecast conservatively. Pilot to production multiples typically run at 3x to 10x the pilot volume.
- Multi year locks AI pricing. Three year terms can lock 2026 list rates against future Salesforce pricing actions.
How Agentforce is priced
Agentforce departs from the traditional Salesforce per seat model and adopts a per conversation metric. The intent is to align cost with value, since each conversation theoretically represents a unit of customer or agent work.
The conversation metric
- Definition. One conversation is one end to end interaction with an Agentforce agent.
- Boundary. A conversation closes when the user disengages, a topic switch occurs, or a defined idle threshold passes.
- Multi turn. Multiple turns within a single conversation count as one conversation, not multiple.
- Failed interactions. Conversations that fail mid stream still count, with adjustments per contract.
- API access. Direct API calls to Agentforce are metered against the same conversation pool.
The 2026 list price
| Tier | List price | Notes |
|---|
| Standard conversation | $2.00 | Out of the box agent flow |
| Enhanced conversation | $3.00 | Multi step, retrieval augmented |
| Agent Builder seat | $50/mo | Per builder, not per agent |
| Data Cloud add on | Variable | Required for some patterns |
| Connected app credits | Variable | For external system actions |
Bundle structure
Salesforce bundles Agentforce conversation entitlements into the Einstein 1 family. The bundle structure obscures the per conversation rate and makes apples to apples comparison harder. The buyer side discipline is to unpack the bundle and compute the effective per conversation rate.
Einstein 1 bundle entitlements
| Bundle | Included conversations per user per month | Effective rate per conversation |
|---|
| Sales Cloud Einstein 1 | 50 | Roughly $0.80 |
| Service Cloud Einstein 1 | 100 | Roughly $0.50 |
| Marketing Cloud Einstein 1 | 25 | Roughly $1.20 |
| Industry Cloud Einstein 1 | Variable by industry | $0.50 to $1.50 |
Three bundle rules
- Entitlements pool monthly. Unused conversations do not roll over.
- User to user transfer. Entitlements pool across users in the same org but not across clouds.
- Overage applies bundle wide. Once the bundled pool is exhausted, the contract overage rate applies.
Conversation math
The conversation forecast is the biggest commercial input. Forecasts that run too low produce overage. Forecasts that run too high produce wasted commitment. The buyer side discipline is to anchor the forecast in historical data and to leave headroom for the contract overage rate.
Annual conversations equals the user count times the average conversations per user per month times twelve. The average conversations per user per month varies dramatically by industry and by use case maturity.
Conversation benchmarks by use case
| Use case | Conversations per user per month | Notes |
|---|
| Sales rep coaching | 10 to 30 | Low frequency, high value |
| Service agent assist | 50 to 200 | High frequency, embedded in case flow |
| Customer self service | 0.5 to 3 per end user | Per end user not per seat |
| Marketing automation | 5 to 15 | Campaign frequency dependent |
| Internal HR or IT | 2 to 8 | Lower than customer facing |
Worked example
A typical service operation with five hundred service agents using Service Cloud Einstein 1 at one hundred conversations per agent per month consumes roughly six hundred thousand conversations per year. The bundle covers five hundred and sixty thousand of those, leaving a forty thousand conversation overage that lands at the contract rate.
The overage rate trap
The conversation overage rate is the single most expensive line in an Agentforce contract if usage grows faster than forecast. List overage runs at two times the bundled rate. Negotiated overage typically lands at one point one to one point three times the bundled rate.
The buyer side discipline is to size the bundled commitment realistically and to negotiate the overage rate aggressively. Underbundling and overpaying on overage is the most common mistake.
Negotiation levers
Six commercial levers move an Agentforce contract. The levers stack. The strongest contracts use four or five together.
Six negotiation levers
| Lever | Typical magnitude | Best used when |
|---|
| Discount on conversation list rate | 40 to 70% off list | Volume commitment is meaningful |
| Overage rate cap | 1.1x bundled rate | Usage growth uncertainty is high |
| Multi year price hold | 0% uplift for 36 months | Salesforce price actions are looming |
| Conversion of bundled cloud seats | Trade old credits for new | Existing Salesforce footprint is mature |
| True down clause | Right to reduce commitment | Pilot to production transition is uncertain |
| Co innovation pricing | Strategic discount for reference | Customer can become a Salesforce case study |
Lever stacking pattern
The strongest Agentforce contracts stack the conversation discount with the overage cap, the multi year price hold, and the true down clause. The stacking pattern preserves budget downside while capping budget upside.
Salesforce sells Agentforce as a strategic AI investment. The 2026 list rates lock customers into a metric that will repay Salesforce many times over once enterprise usage matures. The buyer side discipline is to negotiate the 2026 contract for the 2027 and 2028 reality, not for the pilot volumes that fit on a single screen.
Buyer side discipline
Five discipline rules separate well negotiated Agentforce contracts from poorly negotiated ones. Each rule is preventable with preparation.
Five rules
- Forecast in three scenarios. Pilot, mature, and stretch volumes. Commit to the mature, not the stretch.
- Negotiate the overage rate before the commitment. The overage rate is the asymmetric risk.
- Lock the conversation definition in the contract. The definition is moving and ambiguity favors the vendor.
- Get true down rights. Pilot to production transitions are uncertain.
- Bundle the close with the Salesforce renewal. Standalone Agentforce contracts close at list more often than bundled ones.
The right team
The Agentforce negotiation team carries a procurement lead, a finance partner, a Salesforce technical owner, an AI product owner, and an independent advisor. The independent advisor brings the conversation benchmark, the contract templates, and the overage discipline.
What to do next
The seven step checklist below is the buyer side starting position for any Salesforce Agentforce engagement.
- Inventory the Salesforce footprint. Sales, Service, Marketing, Industry, Data Cloud.
- Forecast the conversation volume. Pilot, mature, stretch.
- Decode the bundle math. Effective per conversation rate by bundle.
- Build the overage scenario. Conversation growth past commitment.
- Map the negotiation levers. Discount, overage cap, true down, price hold.
- Time the close to the renewal. Standalone Agentforce buys land worse than bundled.
- Engage an independent advisor. Salesforce led pricing tilts to the strategic frame.
Frequently asked questions
What exactly counts as one Agentforce conversation?
A conversation is one end to end interaction with an Agentforce agent from open to close. Multiple turns within a single user session count as one conversation. The conversation closes when the user disengages, when a topic switch crosses a defined threshold, or when an idle timer expires.
The exact thresholds vary by contract and are negotiable. The buyer side discipline is to lock the definition in the contract rather than to inherit a vendor side definition that can be tightened later.
How does the Einstein 1 bundle change the math?
The Einstein 1 bundles include monthly conversation entitlements per user. Sales Cloud Einstein 1 includes fifty conversations per user per month. Service Cloud Einstein 1 includes one hundred. The effective per conversation rate inside the bundle is significantly lower than the standalone list rate.
The buyer side discipline is to compute the effective rate, to plan usage within the bundle entitlement, and to negotiate the overage rate for any volume past the bundle.
What is a realistic discount on Agentforce list pricing?
Standalone Agentforce contracts negotiate at twenty to forty percent off list for meaningful annual commitments.
Bundled contracts attached to a Salesforce renewal regularly negotiate at forty to seventy percent off list, especially when the customer commits to a multi year term and accepts a co innovation reference clause. The lever is the renewal bundle and the volume commitment, not the Agentforce standalone conversation.
Should we forecast aggressively or conservatively?
Conservatively. The asymmetric risk on an Agentforce contract sits in the overage rate, not in the commitment level. A conservative forecast keeps the committed spend low and runs any usage growth into negotiated overage.
An aggressive forecast over commits and produces wasted spend if usage growth stalls. The discipline is to commit to the mature scenario, to negotiate the overage rate aggressively, and to revisit the commitment at the next renewal once actual usage data is available.
Can we negotiate true down rights?
Yes, particularly for pilot to production transitions where usage is uncertain. The true down clause typically allows a one time reduction in the annual conversation commitment at a defined milestone, often the end of year one.
Salesforce will negotiate the clause when the customer brings a credible pilot to production case. The clause is rare but available, and the negotiation magnitude on the commitment is meaningful.
How does Redress engage on Agentforce contracts?
Redress runs Agentforce engagements inside the Vendor Shield subscription and the Renewal Program. The work covers the conversation forecast, the bundle math, the overage scenario, the discount stack, the true down clause, and the contract close. Always buyer side, never Salesforce paid.
How Redress engages on Salesforce
Redress runs Salesforce Agentforce engagements inside the Vendor Shield subscription, the Renewal Program, the Benchmark Program, and the Software Spend Assessment.
Read the related benchmarking framework, management team, about us, locations, and contact pages.
