Free tool
Agentforce Readiness Score
Most AI agent pilots don't fail on the technology. They fail on a job nobody named, data the agent couldn't trust, boundaries nobody signed and a bill nobody modelled. Twelve questions tell you which of those is waiting for you.
The job
One task, one owner, one measureHave you named one job for the agent, and the person who owns that work today?
Can you say how that job is measured today: handling time, resolution rate, conversion, cost?
Is the job worth it: enough volume to matter, small enough to finish in six weeks?
The data
What the agent reads, and whether it can trust itDoes the data the agent needs live in Salesforce, or somewhere it can reach cleanly?
Do you know who may see what, so the agent can't surface data a person shouldn't?
Is the knowledge the agent will answer from written down, current and owned?
The guardrails
What it may do, who checks, who owns itHas someone with authority signed off what the agent may do alone, and what it must hand to a person?
Can you test the agent before and after each change, with real examples and the answers you'd accept?
Is there a named owner for the agent once it's live: tuning, monitoring, incidents?
The money and the people
Cost at volume, adoption, a stop ruleDo you know what the agent will cost to run at full volume, in credits or conversations?
Do the people whose work changes know about it, and want it?
Is there a decision point agreed in advance: what result means scale, and what means stop?
Your read
0 of 12 answered
Start with any four answers
Your score, the weak spots and the first moves show here and sharpen with every answer.
4 more answers and your first read appears here.
How the score works
Twelve questions in four groups. Each answer scores 0 to 3, and "don't know" scores 0 because not knowing is a risk in itself. Each group is the share of its points you scored, and the overall score is the average of the four groups, so each counts equally. You see a first read after four answers; the full picture needs all twelve.
Ready to pilot 80 to 100
You know the job, the data and the rules. A six-week pilot should prove value or kill the idea cheaply.
Nearly there 60 to 79
The job is clear enough to start. Fix the gaps below in the first two weeks, not after launch.
Not yet 40 to 59
An agent built now would stall at the pilot. Two or three basics first, then go.
Start with the basics 0 to 39
Pick one job and get its data and owner straight. That's a fortnight of work, and it decides everything after.
It's a self-read from one person's view at one point in time. It won't replace an independent look, but it will tell you where to start, and it's a useful thing to put in front of a team before a planning session.
Frequently asked questions
Is this only for Salesforce Agentforce?
The questions are about the job, the data, the guardrails and the money, so they work for any agent platform. The data questions assume Salesforce because that's where most of the people who use this keep their customer data.
Why does 'don't know' score zero?
Because in an agent pilot, not knowing is the exposure. If you don't know what it will cost at volume or who may see what, that's the first thing to find out, so the tool puts it at the top of your list.
Do you sell Agentforce?
No. We're a Provisional Salesforce consulting partner and we lead programmes and pilots; we don't resell licences and nobody pays us for recommending anything. If the honest answer is 'not yet', the tool says so.
What happens with my answers?
Nothing leaves your browser. The score is worked out on the page. 'Copy link' puts your answers in the link itself, so only the people you send it to can see them.