Salesforce AI readiness assessment
A structured review of data, process, governance and licence readiness before you invest in Agentforce agents or Einstein features.
Most disappointing AI pilots fail for ordinary reasons: missing data, unclear processes, nobody owning the agent, or a licence that doesn't cover the feature. Our Salesforce AI readiness assessment checks these before you build. You get a clear view of what is ready, what isn't and which use case to start with.
If you want a quick self-check first, try the data and personalisation readiness tool. When you are ready to build, see Agentforce and AI automation.
What we assess
| Area | What we check |
|---|---|
| Data | Completeness, duplicates, ownership, and whether the records a feature needs actually exist |
| Knowledge and grounding | Whether articles, policies and product data are current enough to ground answers |
| Process | Which tasks are defined well enough to hand to an agent, and where a person must stay involved |
| Governance | Trust Layer settings, data masking, audit trail, permissions such as Manage AI Agents, and change approval |
| Licences | Which editions, add-ons and consumption models apply to each use case |
| Platform | Whether Agentforce and Einstein are enabled in Setup, Data 360 is provisioned where needed, and your team is ready for the new Agentforce Builder |
| Use cases | A ranked shortlist with risks, data gaps and a recommended first pilot |
How we work
- 01
Discovery
Interview process owners, list candidate use cases and gather access to the relevant objects and settings.
- 02
Design
Agree scoring criteria for each area and the evidence we'll collect.
- 03
Build
Profile the data, review settings, permissions and licences, and map each use case to its requirements.
- 04
Test
Check findings with data and process owners so nothing in the report surprises them.
- 05
Launch
Present the report, scores, risks and a phased roadmap.
- 06
Support
Help you prepare the first pilot, or hand over to your own team.
Data requirements are specific
Salesforce documents concrete thresholds for some features, and these make good readiness tests. Einstein Forecasting needs at least 12 months of opportunity history, with Amount filled in on at least 80% of open opportunities, and runs in production orgs only. Objective-based recommenders in Salesforce Personalization need at least three recorded engagement activities. Send Time Optimization in Marketing Cloud uses 90 days of engagement data.
Licences and commercial models
Salesforce says that "every Salesforce customer can get started with Agentforce for free, for any use case, with Salesforce Foundations". Some features need a Foundations add-on, and Salesforce also sells through Flex Credits, resolutions or per-user licensing. In the Sales Cloud editions introduced in September 2026, Max includes "our full suite of AI". Salesforce's Agentforce ROI calculator is "for illustrative purposes only" and excludes implementation, licence and Data 360 credit costs. We list the licence questions to raise with your account team and don't quote prices.
Governance and platform
Since the week of 13 July 2026, new agents can only be created in the new Agentforce Builder, which uses Agent Script. The Einstein Trust Layer provides grounding, data masking, toxicity detection and an audit trail. We check that someone owns each of these settings.
Common mistakes
- Starting with the most impressive use case rather than the best-prepared one.
- Assuming knowledge articles are accurate because they exist.
- Leaving governance until after launch. See Agentforce testing and governance.
Related reading
Responsible AI
How we approach AI work
→Data 360 readiness checklist
Prepare your data foundation
→Security and data handling
How we treat your data during an assessment
→Questions we hear
What do we get at the end?
Do we need Data 360 for Agentforce?
Will you need access to production data?
Can we act on the report without you?
Sources
Check your AI readiness
Tell us which AI use cases you're considering. We'll suggest the scope of an assessment that fits.
