Salesforce AI Readiness Checklist: 20 Questions to Answer Before Agentforce
Most Salesforce AI projects do not fail during the build. Instead, they fail after launch, when someone asks an agent a simple question and gets a confident answer that turns out to be wrong.
Trace it back, and the issue usually did not start with AI.
AI simply exposed the shaky Salesforce foundation that was already there: weak data, unclear ownership, outdated permissions, or a process nobody had fully defined.
AI readiness usually starts with the bigger roadmap. Where can AI help the business? Which processes are worth improving? What data needs to be trusted? What governance needs to be in place before teams start experimenting?
But once the strategy is clear, the next question gets more practical: Is your Salesforce org actually ready for Agentforce to touch this process?
That is what we built this checklist to answer.
Use it before you buy licenses, launch a pilot, connect external data sources, or give an agent access to customer, sales, service, or operational records.
Salesforce AI Readiness: Is Your Organization Ready for Agentforce?
This checklist includes 20 questions across five readiness areas: strategy, data, security and governance, people, and pilot planning. Score one point for every question you can confidently answer “yes” to today, not for the items you plan to fix later.
Readiness Area 1: Strategy
Before Agentforce touches any of your workflows, the business case needs to be clear. If the team cannot name the process AI should improve, define the baseline, and explain why AI is the right fit, the project is likely to drift before it delivers value.
If this is true about your organization, you should be able to say yes to these:
Have you named the specific business process Agentforce should improve?
Do you have a baseline performance number for the process today?
Have you confirmed that AI is a better fit than configuration, automation, or process redesign?
Readiness Area 2: Data
Agentforce will only be as useful as the data it can access. If your data is inaccurate, outdated, incomplete, duplicated, or poorly governed, AI may amplify these problems instead of solving them.
This includes Salesforce records, external sources, documents, knowledge articles, transcripts, and any other content the agent may rely on.
Before moving forward, your team should be able to say yes to these statements:
Do you regularly review the accuracy, completeness, and reliability of the data Agentforce may use?
Have you audited duplicate records in the last 12 months and created a process for managing duplicates going forward?
Does every object, field, and data source Agentforce may rely on have a clear owner?
Do you know which Salesforce data Agentforce should and should not use?
Do you know which external data sources Agentforce may read from, and have you reviewed their accuracy and reliability?
Do users trust the data in the records and systems they work from?
Readiness Area 3: Security and Governance
AI changes the access question. It is no longer only about who can see the data. It is also about what the agent can reach, what it is allowed to use, and how the organization will monitor that behavior over time.
Before Agentforce is given access, your team should be able to say yes to these:
Have you reviewed permissions and sharing rules with AI access in mind?
Do you know exactly what data an agent could reach today?
Do you have a written policy for what AI can and cannot access?
Do you have an audit trail, and is someone responsible for reviewing it?
Readiness Area 4: People
AI projects do not succeed on technology alone. Someone needs to own the initiative, validate outputs, handle issues when something goes wrong, and make sure the wider team understands why the change is happening.
Before launch, your team should be able to say yes to these statements:
Does the initiative have a named owner with real authority?
Is someone responsible for validating output before anyone acts on it?
Is there a working path for reporting a wrong answer?
Has leadership clearly explained to the team why this work matters?
Readiness Area 5: Pilot
A pilot is a controlled first test before a wider rollout. For Agentforce, that means choosing one specific workflow, a limited group of users, clear agent boundaries, and measurable success criteria before giving the agent broader access.
The pilot should be small enough to manage and specific enough to measure. If the scope, users, agent limits, success metrics, or decision date are vague, the pilot can become open-ended experimentation instead of a real business test.
Before starting the pilot, your team should be able to say yes to these:
Do you have a defined pilot scope and user group?
Have you defined what the agent will and will not be allowed to do during the pilot?
Have you agreed on success metrics and a decision date before launch?
What Your Score Means
16–20: Ready to Pilot
Your foundation is in good shape. You likely have enough clarity to begin a controlled Agentforce pilot. Spend the remaining effort on governance depth, user validation, and the decision point at the end of the pilot. That is where many AI programs start to drift.
9–15: Foundation Work First
Start with your lowest-scoring section and fix it before buying more licenses or expanding the use case. Data and permissions are the two areas most likely to create visible failures. If you need to choose where to begin, start there.
Under 9: Do Not Start with AI Yet
Start with a Salesforce health check. It will help you improve reporting accuracy, user trust, audit readiness, and operational control whether or not you ever deploy an agent.
Not Sure Where to Start with Agentforce?
Download Kander®’s Agentforce KickStart to see how your team can start small, validate the right use case, and scale what works.
DOWNLOAD AGENTFORCE KICKSTARTThree Readiness Gaps Teams Often Miss
1. Data Living Outside Salesforce
Agents can be grounded in call transcripts, PDFs, support logs, knowledge articles, shared drives, and other unstructured sources.
Most readiness reviews only audit CRM records. But if an agent can read a shared drive nobody has cleaned since 2019, that content is now part of your answer quality and part of your risk.
Before connecting Agentforce to external content, make a source inventory. List every non-Salesforce source the agent can read, who owns it, when it was last reviewed, and whether outdated or sensitive content should be removed, restricted, or excluded.
2. Prompt and Instruction Drift
As teams tune instructions over time, agent behavior can move away from what was originally tested and approved. Nothing looks broken, but the answers can quietly get worse, and there may be no obvious error message.
Version your instructions, review them on a schedule, and make clear who signs off on changes.
3. Fields with No Real Owner
Ownership is one of the questions teams often score themselves highest on and get wrong.
“The admin owns it” is not ownership. Ownership means there is a named person who decides what belongs in that field, notices when it stops being filled in, and can explain why the number changed.
Where This Fits in Your AI Readiness Roadmap
This checklist is not a full AI roadmap. It is a readiness check before Agentforce touches your Salesforce data, users, workflows, and customer-facing processes.
If your score is low, the next step is not to rush into AI, but to sequence the foundation work: clarify the process, clean up the data, review permissions, assign ownership, define governance, and then pilot Agentforce in a controlled use case.
As Pallavi Agarwal, Founder & CEO of Kander®, puts it:
“With AI, you still have to crawl before you walk. You still need to clarify the process, strengthen the Salesforce foundation, and then introduce Agentforce where it can safely reduce manual work.”
Start Agentforce with a Stronger Foundation
Thinking about Agentforce? Kander® can help you pressure-test your Salesforce foundation, spot the risks your team may be missing, and decide what needs to happen before AI enters the workflow.
