Einstein Setup
Einstein Setup is the guided Setup experience in Salesforce that walks administrators through enabling and configuring the org's Einstein AI features.
Definition
Einstein Setup is the guided Setup experience in Salesforce that walks administrators through enabling and configuring the org's Einstein AI features. It surfaces the available features in one place, recommends which to enable based on the org's data and usage patterns, and runs activation wizards for each. Where Einstein Platform is the organizing tree of per-feature pages, Einstein Setup is the curated tour that helps admins discover and turn on the right features without already knowing what they are.
Einstein Setup is most useful for new orgs or new admins joining an existing org. It shows which Einstein features the edition includes, which are enabled, which have outstanding setup steps, and which would benefit the org based on data volume and current configuration. Experienced admins who know their feature set tend to bypass Einstein Setup and go straight to specific Einstein Platform nodes; both surfaces are valid entry points and write to the same underlying configuration records.
Why Einstein Setup is the right starting point but the wrong end state
What Einstein Setup contains
Setup, Einstein, Einstein Setup. The page opens to a dashboard listing every Einstein feature the org's edition includes, grouped by category (Sales, Service, Commerce, Marketing, Industries, Agentforce, Trust Layer). Each feature shows status (Enabled, Not Enabled, Outstanding Setup), a brief description, and a Start Setup button that opens the activation wizard for that feature. The dashboard also surfaces personalized recommendations based on the org's data; an org with high case volume sees Case Classification recommended near the top.
The guided wizard pattern
Each feature has its own wizard accessed through Einstein Setup. The wizards are designed to be self-contained: a series of guided questions that produce the configuration. For example, the Case Classification wizard asks which case fields to classify, scopes the training data, and previews the model accuracy before deployment. The wizards are friendlier than the raw per-feature configuration pages in Einstein Platform but expose fewer advanced options. For most first-time setups, the wizard produces a working configuration; advanced tuning happens later through the Platform pages.
Recommendations and how they are sourced
Einstein Setup recommends features based on the org's data shape and usage. A Service Cloud org with 50,000 cases per month sees Case Classification and Article Recommendations recommended. A Sales Cloud org with active email integration sees Activity Capture and Opportunity Scoring recommended. A B2C Commerce Cloud org sees Product Recommendations recommended. The recommendations are heuristics, not magic; they catch the obvious feature matches and miss the org-specific judgment calls that come from knowing the workflow. Use them as a discovery aid, not as a deployment plan.
The relationship to Einstein Platform
Einstein Setup and Einstein Platform write to the same underlying configuration records. Enabling a feature in Setup updates the same metadata as enabling it in Platform. The difference is interface: Setup is a wizard-driven dashboard for discovery and first-time activation; Platform is a tree of per-feature pages for ongoing configuration. Most teams start in Setup and graduate to Platform as their familiarity grows. There is no migration step between the two; admins move freely between them based on what they need.
Permissions and who should see Einstein Setup
The Einstein Admin permission set grants access to the Einstein Setup dashboard and all the activation wizards. Per-feature permission sets grant narrower access. A reasonable governance pattern: one or two admins hold Einstein Admin and own the rollout planning; broader admins hold per-feature permissions for the features they own. Avoid handing Einstein Admin broadly; the dashboard exposes activation paths for features that span the whole org and unaudited enablement creates downstream surprises.
Activation order and dependency awareness
Einstein Setup does not enforce activation order, but some features perform much better when their dependencies are enabled first. Activity Capture before Opportunity Scoring. Knowledge curation before Article Recommendations. Trust Layer configuration before any LLM-driven feature. Einstein Setup surfaces dependency hints in some wizards but not all; admins should sketch the rollout sequence on paper before clicking through the wizards. Out-of-order activation produces features that look correctly configured but perform on thin data.
When to graduate from Setup to Platform
The natural graduation point is when admins need configuration options the wizard does not expose: tighter training data filters, advanced retraining schedules, per-channel configurations, custom prompt templates. The wizards are deliberately curated; not exposing every option keeps the first-time experience clean. Once an admin knows the feature well enough to tune the underlying parameters, they move to the Platform page for that feature. Both surfaces remain valid; Einstein Setup is still useful for discovering newly shipped features even on mature orgs.
How to use Einstein Setup as a discovery and activation tool
Einstein Setup is at its best in the first few weeks of an Einstein rollout and during quarterly discovery walks. Treat it as a discovery surface and an activation wizard, not as a long-term configuration interface. The advanced work happens in Einstein Platform; the introduction happens in Einstein Setup.
- Open Einstein Setup and review the recommendations
Setup, Einstein, Einstein Setup. Skim the recommendations and the dashboard. Note which features are enabled, not enabled, or have outstanding setup.
- Sketch the activation order on paper
Dependencies matter. Activity Capture before scoring features. Knowledge before article recommendations. Trust Layer before LLM features. Sketch the order before clicking any wizards.
- Activate one feature at a time through the wizards
Pick one feature. Run the wizard end to end. Let it complete and validate. Resist the urge to activate three features in parallel; per-feature focus avoids confused debugging when one wizard's choices affect another.
- Validate each activation before moving on
After each wizard, open the feature's configuration in Einstein Platform. Confirm the settings look right. Check that the feature is producing output (predictions, suggestions, signals).
- Graduate to Einstein Platform once familiar
For ongoing tuning, the Platform per-feature pages expose more options than the Setup wizards. Move to Platform once you know the feature well enough to want the extra knobs.
- Re-visit Einstein Setup quarterly
Salesforce ships new Einstein features per release, often without big announcements. The quarterly walk through Einstein Setup catches new capability you would otherwise miss.
- Document activations in your Einstein feature inventory
Each activated feature should have an owner and a review date in your inventory document. Activation through a wizard does not absolve you of operational ownership.
Drives by edition and license. Features the edition does not include are hidden or grayed out.
Heuristics based on data shape and usage. Useful as a discovery aid; not a substitute for judgment.
Curated subset of the underlying configuration. More limited than Einstein Platform, intentionally.
The activation wizards respect existing Trust Layer settings. Configure Trust Layer first.
Surfaced in some wizards but not all. Sketch dependencies on paper rather than relying on the UI to enforce them.
- Einstein Setup does not enforce activation order. Activating Opportunity Scoring before Activity Capture produces scoring on thin activity data and predictable disappointment.
- Wizards expose fewer options than the Platform pages. Some advanced tuning is not accessible from Setup at all.
- Recommendations are heuristics, not personalized strategy. Use them as a discovery aid and apply judgment before activating.
- Activation does not transfer ownership. Every activated feature needs an owner and a review cadence; the wizard does not enforce that discipline.
- Einstein Setup and Einstein Platform write to the same records. Changes made in either place are visible in the other; do not assume changes are isolated.
Trust & references
Cross-checked against the following references.
- Salesforce Einstein overviewSalesforce
- Einstein Setup referenceSalesforce
Straight from the source - Salesforce's reference material on Einstein Setup.
- Einstein SetupSalesforce Help
- Einstein Platform OverviewSalesforce Help
- Einstein Trust LayerSalesforce Help
Hands-on resources to go deeper on Einstein Setup.
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What does Einstein Setup need to work effectively?
Q2. How does the Einstein Trust Layer relate to Einstein Setup?
Q3. What technology powers Einstein Setup in Salesforce?
Discussion
Loading discussion…