Claudeforce is coming. The teams that win will be the ones that are ready.
An agent inherits everything you've written down — and nothing you haven't.
We've been running something like Claudeforce since Feb 2026. Here's what ready looks like.
What is Claudeforce readiness?
Claudeforce readiness is how prepared your organization is to hand real work to an AI agent in your CRM. It has five dimensions: data an agent can trust, know-how that is actually written down, integration that reaches beyond one system, rules for what an agent may do, and defined work to hand it.
None of that is about the software. An agent inherits everything you have written down, and nothing you haven't, so the work of getting ready is the work your company never got around to. Companies grow. People leave. Whoever knew why the discount rule exists took it with them. Process that lives in a good rep's habits works right up until you try to hand it to something that has no habits.
That is why readiness is a question about your organization rather than about any product. The answer is the same whether the agent works on your CRM, your documents, or your help desk: it can only act on what you have made explicit. The teams who spend the next few months writing down how they actually work will be ready for whatever they decide to point at it. The teams treating readiness as something that begins on a launch date will begin behind.
The five dimensions of Claudeforce readiness
Dimension 1
Data Readiness — Can an agent trust your records?
Humans route around bad data; agents amplify it. Duplicates, stale pipeline, empty fields, the shadow spreadsheet.
Ask yourself
- If two people pulled the same pipeline report today, would they get the same number?
- How many of your open opportunities haven't been touched in ninety days?
Dimension 2
Knowledge Readiness — How much of your know-how is written down?
Methodology, pricing rules, playbooks and decision rationale live in heads, Slack threads and tribal memory — because companies grow, people leave, and nobody's job was to write it down. AI can only use what you've captured.
Ask yourself
- If your best rep left tomorrow, what would be hardest to reconstruct?
- Where would someone new look up your discount rules, and would what they found be current?
Dimension 3
Integration Readiness — Can agents act beyond one system?
Real work crosses systems: quote to invoice, order to fulfillment. Governed integration rather than point-to-point connectors and CSV exports.
Ask yourself
- Name a process that crosses two systems. Who moves the data between them?
- How many spreadsheets sit between a signed deal and an invoice?
Dimension 4
Governance Readiness — Who is the agent allowed to be?
Your permissions, your action boundaries, your approval gates — policies most orgs haven't written, because they've never had a non-human colleague.
Ask yourself
- Who decides what a new system is allowed to see, and is that written down anywhere?
- How many people can delete a record that matters?
Dimension 5
Workflow Readiness — Is there defined work for an agent to take?
If process lives only in behavior, an agent has nothing to run. Defined stages, clean handoffs, verifiable outputs, and a team postured to direct and review agent work. AI accelerates your people, never replaces them.
Ask yourself
- Pick a task somebody does every week. Could you describe its steps to a new hire without improvising?
- When that task is done wrong, how do you find out?
The five levels of readiness
The quiz scores 5 questions out of 10 and places you on one of five levels.
Agent-Blind
0–2
Little written down and records nobody trusts. An agent would have almost nothing reliable to work from.
Agent-Curious
3–4
The intent is there and the groundwork isn't. AI gets discussed more than it gets used.
Agent-Assisted
5–6
AI helps individuals. Work still stops at every handoff between a person and a system.
Agent-Ready
7–8
The foundation holds. One or two gaps stand between here and agents on live work.
Agent-Operated
9–10
Trusted records, documented process, governed integration, clear rules. Agents can be handed real work.
- Agent-Blind0–2
- Agent-Curious3–4
- Agent-Assisted5–6
- Agent-Ready7–8
- Agent-Operated9–10
Newer to Claudeforce?
Two short explainers — on the term itself, and on how it differs from Agentforce.
What is Claudeforce?
Separates what Salesforce has announced from what the market is assuming, so you know which parts are safe to plan around.
Claudeforce vs Agentforce
Where the two overlap, where they genuinely differ, and which one your team is actually asking about.
Frequently asked questions
What are the five pillars of AI readiness?
Five dimensions decide whether an organization can hand work to an AI agent: data the agent can trust, knowledge that is actually documented, integration that reaches across systems, governance defining what the agent may do, and workflow defined clearly enough to hand over. Weakness in any one caps what the others deliver.
Is my org ready for Claudeforce?
Take the five-question readiness score. It returns a level from Agent-Blind to Agent-Operated plus the dimension holding you back most. It takes under a minute and needs no data access, because it asks five questions about how your organization runs rather than auditing your systems.
How do you assess AI readiness?
Honestly, by looking at work rather than asking about it. Self-report tells you what people believe. The gaps show up when somebody tries to hand a real, recurring task to an agent and finds that the rule deciding it was never written down. Start with one task.
What is agent readiness?
Agent readiness is whether there is work an AI agent could actually take. People fill gaps by instinct, knowing which records to distrust and who to ask. An agent has no instinct. It needs records that are right, rules that are written, and a task with a defined shape.
Won't connecting Salesforce to Claude handle this?
Connect Salesforce to Claude and the agent knows your records. It still doesn't know your company, because the pricing logic, the qualification criteria and the reasons behind decisions were never written down anywhere it could read. Readiness includes deciding where the rest of your company's context lives and how it is governed.
How long does the readiness score take?
Under a minute. Five questions, one per dimension, no data access required. You get a readiness level, the dimension holding you back most, and a one-line read on each of the five. Leave an email and a written five-dimension profile comes back within one business day.
What happens after I submit the readiness score?
Your profile is drafted by an AI agent working from your answers and reviewed by a person before it reaches you. That's how we think agent work should run. It arrives within one business day and covers where you stand on each dimension and what to fix first.
Do we need to fix everything before agents are useful?
No. Readiness isn't a gate you pass. Most organizations get real value one dimension at a time, and the recurring task whose rules finally got written down is the one an agent can take. Fix the dimension holding you back most and the next move gets obvious.
Not sure where you land?
5 questions across the five dimensions. You'll get your level and your weakest dimension.