There are two contractual routes to Salesforce’s agent platform: upgrade the whole estate to Agentforce 1 Edition — Unlimited plus unmetered agent assistance, an annual Flex Credit allowance, Data Cloud capacity, Tableau Next and Slack in one fee — or stay on Unlimited and attach the Agentforce add-on and bought credits to the populations that need them. The edition pays when you would consume most of that bundle across every licensed user; the add-on path pays when the AI need is real but concentrated. The complication neither datasheet leads with: part of the bill is consumption-metered either way.
Published 2 April 2026 · Last reviewed 23 April 2026
Agentforce 1 Edition arrived in 2025 as the successor to Einstein 1 Edition and sits at the head of the editions ladder. Structurally it is Unlimited with the AI shelf pre-loaded: unmetered agent assistance for licensed users, an annual organisation-level allowance of Flex Credits for autonomous agent actions, Data Cloud capacity measured in its own credits, Tableau Next analytics and Slack entitlements — one per-user fee, one SKU, the whole estate repriced.
The add-on route reaches the same platform from below. An Unlimited (or Enterprise) estate attaches the Agentforce add-on — per-user agent capability with its own included entitlements — and buys Flex Credits transactionally or in committed pools as consumption demands. Nothing else moves: Data Cloud, analytics and Slack remain separate decisions, bought if needed and not if not.
The platform at the end of both routes is the same. What differs is the contract shape — which populations are repriced, which line items are visible, and where the consumption meter sits. Those three differences are the whole decision, and they are negotiating questions as much as architectural ones.
This guide compares licensing structures as published by the vendor; it is general information, not legal or licensing advice for your situation, and it names no firms. Bundle contents and credit mechanics shift — verify current inclusions against your order form and the vendor’s edition documentation. The firm directory lists Salesforce-capable advisors with balanced pros and cons, listed, not ranked.
| DIMENSION | UNLIMITED + ADD-ONS | AGENTFORCE 1 EDITION |
|---|---|---|
| Agent capability | Agentforce add-on per user, on the populations that need it | Included for every licensed user; unmetered employee-facing assistance |
| Consumption (Flex Credits) | Bought transactionally or as committed pools; sized to demand | Annual organisation-level allowance included; overage bought on top |
| Data Cloud | Separate purchase, if and when needed | Capacity bundled in, on its own credit meter |
| Analytics & Slack | Separate decisions, separate lines | Tableau Next and Slack entitlements included |
| Who gets repriced | Only the add-on populations | Every licensed user, whether or not they touch an agent |
| Cost shape | Several visible lines; each one droppable at renewal | One large line; unit economics folded out of sight |
| Exit & renewal posture | Granular levers; add-ons and credit pools can shrink or go | Edition downgrade only at renewal, against commercial friction |
Read the last three rows together and the pattern from every bundle decision on this ladder repeats: the edition buys consolidation and breadth at the price of granularity and reversibility. That trade can be worth making — but it has a price, and the price should be computed from your consumption data, not conceded to the demo.
The seat-based instinct — count users, multiply, negotiate the rate — covers only part of an Agentforce deal. Autonomous and customer-facing agent actions draw down Flex Credits per action, which makes that slice of the bill usage-shaped: it scales with what the agents do, not with how many people you employ. The edition includes an annual allowance; the add-on path buys pools outright. Either way, someone in your organisation now owns a burn-rate forecast.
This is the genuinely new discipline in the 2026 Salesforce estate, and it looks more like cloud FinOps than like classic licensing: estimate actions per workflow, watch consumption against the allowance, alert before exhaustion, and negotiate what overage costs before signature rather than at the moment of need. An included allowance that looks generous in month one can be half-consumed by a single high-volume service workflow — and an exhausted pool mid-term is the weakest possible position from which to buy more.
The forecasting asymmetry favours the vendor on both routes: it has fleet-wide consumption telemetry, you have a pilot. Whoever holds your own measured numbers holds the deal — standing work of exactly the shape a Salesforce licensing advisory engagement exists to do.
Unlimited plus the targeted add-on fits the concentrated case — which is most estates in the early agent years. Agent capability deployed to service and one or two sales functions, credits sized to measured pilot consumption, Data Cloud bought only if a real workload exists. Every line visible, every line droppable, and the populations that gain nothing from agents stay at their current price.
Agentforce 1 Edition fits the estate that would buy the bundle anyway: agents planned across the licensed population rather than a corner of it, genuine Data Cloud workloads, real intended use of the analytics and Slack entitlements, and forecast credit consumption that approaches the included allowance. For that estate, the edition consolidates five negotiations into one and removes per-population add-on administration.
The test is the same arithmetic that decides Enterprise vs Unlimited, at a larger scale: build the counterfactual invoice. Price the add-on path for the populations that genuinely need agents, add credits at your measured burn rate, add only the bundle components you would actually buy — then set that against the edition uplift across every user. Aspirational Data Cloud workloads and analytics nobody has asked for belong on the vendor’s side of the ledger, not yours.
The upgrade pitch now leads every Salesforce renewal. The standing offer is consolidation — move to the edition and the add-ons disappear into one number, often with a first-term discount that makes the comparison look closed. The number to hold against it is the term-two price of the same bundle, after the discount expires and the estate has absorbed the entitlements. Salesforce exposure is contractual rather than audit-shaped: no mid-term true-down, uplift at renewal unless capped, auto-renewal on notice windows. An edition upgrade raises the baseline all of that machinery operates on.
Credit terms are negotiable in ways seat prices are not yet. Overage rates, allowance rollover, the credit cost of specific action types, consumption reporting cadence — these are young contract terms without settled market norms, which cuts both ways: the vendor’s paper defaults are untested, and informed buyers have room. Negotiate the overage rate and rollover before signature, while you still have alternatives.
Mixed structures exist even where the pitch says otherwise. Large estates run editions side by side, and agent capability can sit on one population while the rest of the estate stays put. Whether that structure is offered unprompted is another matter — the negotiation belongs to whoever is holding the consumption data, and the renewal calendar in the when-to-engage guide says when that work has to start.
Repricing everyone for a use case that touches a few. The edition charges every licensed user for agent capability that one service team consumes. Price the targeted add-on first; upgrade the estate only when the arithmetic, not the roadmap, says so.
Reading “unmetered” as “unlimited.” Employee-facing assistance is unmetered; autonomous agent actions burn Flex Credits, and the included allowance is finite. Model the burn rate from pilot data before trusting any allowance to last the year.
Counting bundle components you will not deploy. Data Cloud capacity without a Data Cloud workload, analytics without an analytics migration, Slack already licensed elsewhere — inclusion is not value. The counterfactual invoice contains what you would have bought, nothing else.
Signing consumption terms blind. Overage pricing, rollover and reporting cadence left at vendor defaults become the expensive surprise of year two. They are negotiable at signature and rarely negotiable at exhaustion.
Taking the bundle valuation from the parties paid by it. The account team and the implementation partner both do better when the estate moves up the ladder. Apply the independence test to every adviser in the room before the renewal cycle, not after the signature.
Agentforce 1 Edition is the tier above Unlimited on the Salesforce editions ladder, introduced in 2025 as the successor to Einstein 1 Edition. It bundles the Unlimited feature set with unmetered agent assistance for licensed users, an annual allowance of Flex Credits for consumption-based agent actions, Data Cloud capacity, Tableau Next analytics and Slack into a single per-user subscription.
Employee-facing agent assistance tied to licensed users is sold as unmetered within the edition, but autonomous and customer-facing agent actions draw down Flex Credits, and the edition’s annual credit allowance is finite. Once the allowance is consumed, further usage means buying more credits. The edition changes where the meter sits; it does not remove the meter.
Yes. The Agentforce add-on attaches agent capability to existing editions on a per-user fee, and consumption-based agent actions can be bought as Flex Credits without any edition change. The add-on path buys the agents without the surrounding bundle — no included Data Cloud capacity, analytics or Slack — which is exactly the trade the arithmetic has to price.
Flex Credits are the consumption currency Salesforce uses to meter agent actions — each action draws a defined number of credits from a purchased or bundled pool. They make part of the AI bill usage-shaped rather than seat-shaped, which means forecasting consumption, monitoring burn rate and negotiating overage terms are now part of any Salesforce AI deal.
When the estate would genuinely consume the bundle: broad agent deployment across the licensed population, real Data Cloud workloads, planned use of the included analytics and Slack entitlements, and forecast credit consumption near the included allowance. If agent use is concentrated in one function, Unlimited plus the targeted add-on and bought credits is usually the tighter structure — and it keeps each line droppable at renewal.
No. This is a directory, not a ranking. Firms are listed alphabetically with balanced pros and cons. Independence is shown as a pro and reseller, Big-Four or vendor-side-audit ties as a con, both stated as factual trade-offs for you to weigh.
Modelling credit burn from your own pilot data, pricing the add-on counterfactual against the edition uplift, and negotiating overage terms while you still hold alternatives — that is precisely the work of a Salesforce licensing advisory engagement, and of renewal negotiation when the pitch arrives attached to your renewal. The Salesforce hub maps the vendor’s wider licensing world, and the directory lists every firm covering Salesforce — with balanced pros and cons, listed, not ranked. Choosing between them? Start with how to choose a Salesforce licensing partner.
The same arithmetic, one rung down →
The criteria, questions and red flags →
Every firm for this work →
The uplift event, prepared for →
Who your advisor really works for →
Every field guide on the site →
Tell us the vendor, the service you need and where things stand, and we will route your brief to firms that genuinely cover that combination. The directory and matching are free for buyers, no vendor ever sees your brief, and we add no markup.
Our weekly dispatch on vendor audit programs, regional developments and one buyer move. Subscribe to The Licensing Radar.