TL;DR
A year ago, I was writing about the AI readiness gap in marketing teams. Now I am writing about the next one. Agentic readiness is a distinct capability from AI visibility readiness, and most financial institutions are not measuring it. This post lays out a six-dimension framework for agentic readiness, explains why each dimension matters, and offers a quick way to score your institution against it before someone else does it for you.
Back in March, I wrote about the AI readiness gap and how most marketing teams were not ready to compete in 2026. That post has aged well in some ways and poorly in others. It has aged well because the gap was real and is still real. It has aged poorly because the benchmark for readiness has already moved.
The new benchmark is not whether your team can use AI tools. It is whether your institution can be used by AI agents. Those are very different questions.
Why visibility readiness and agentic readiness are not the same thing
When I work with a credit union or community bank on AI visibility, we measure five dimensions. Technical health. SEO readiness. GEO and AEO health. ADA compliance. External reputation. A good score means an AI is likely to recommend you in an answer.
That is not the same as a good score for agentic transactions. An AI agent deciding whether to open an account on behalf of a member is asking a completely different set of questions.
Can I verify this human is who they say they are using your identity layer. Can I complete the account opening without running into a CAPTCHA or a workflow that assumes a human is at a keyboard. Can I read your current rates, fees, and eligibility criteria from a machine-readable source without parsing a PDF. Can I reach a support contact if something goes wrong in the middle of the transaction. Can I hand off cleanly to a human when the situation requires it.
Visibility readiness gets you recommended. Agentic readiness gets you transacted with. Most institutions are decent at the first and terrible at the second. Which is fine, for now. Not for long.
The six dimensions of agentic readiness
Based on the work I have been doing with financial institutions on this topic, here is the framework I use. It is six dimensions, each scored from zero to one hundred, with a weighted total that produces an overall agentic readiness index.
API completeness. Can an agent complete a full customer journey through an API without hitting a manual handoff. Account opening. Funds transfer. Loan application. Appointment booking. Each one gets scored independently.
Policy readability. Are your fees, rates, eligibility criteria, and terms published in a machine-readable format. If your only source of truth is a PDF or a rendered webpage with no structured data, your policy readability score is painfully low.
Identity interoperability. Does your identity and authentication stack support delegated authority from a consumer AI agent. Can a member authorize an agent to act on their behalf in a way your systems can verify and audit. Most institutions today answer no, but do not know it.
Agent directive clarity. Do you have an llms.txt file. Does it actually say something useful. Does it distinguish between agents you welcome, agents you allow with conditions, and agents you block. If your answer is “what is an llms.txt,” that is your score.
Audit and recovery. When an agent takes an action, can you trace it end to end. When an agent takes the wrong action, can you recover cleanly. When a member disputes an agent-initiated transaction, can you resolve it with confidence. This is where compliance and agentic capability meet.
Human handoff quality. When an agent reaches the limit of what it should do without human review, how clean is the handoff. Does it route to the right team with the full context. Does it leave a trail the human can pick up without making the member start over. This is the dimension most institutions ignore, and it is the one that will bite hardest.
Score each dimension. Weight them based on your institution’s priorities. Sum them up. That is your agentic readiness index.
What the scores actually look like in the field
For the institutions I have informally scored this year, the ranges are not encouraging.
The average agentic readiness index for mid-sized credit unions is somewhere in the low thirties. The top decile is in the low sixties. The bottom decile is under fifteen. Community banks, on average, score slightly higher, mostly because their technology partners have been more aggressive on API maturity. But the gap between the best and the average is still enormous.
The most interesting pattern is that institutions with high AI visibility scores do not necessarily have high agentic readiness scores. I have seen credit unions with GEO scores in the seventy-fifth percentile and agentic readiness scores in the twentieth percentile. The capabilities are related but not the same. The work you did on discoverability does not automatically get you ready for agents.
What to do in the next ninety days
You do not need to nail all six dimensions at once. But you do need to know where you stand.
Start with API completeness. Map every customer journey your institution supports. For each one, document whether an agent could complete it end to end without a human handoff. Be honest. The answer is almost always no. That map is your roadmap.
Follow with policy readability. Audit your website for structured data on your core products. Fees, rates, eligibility, restrictions. If it is not machine-readable, it is invisible to agents. This is low-effort work with high leverage.
Stand up an llms.txt file. It takes less than a day. It does not solve your agentic readiness problem, but it signals to agents and to your own leadership that you are paying attention.
Revisit your audit and handoff architecture. Ask your compliance team what they would want to see in an agent-initiated transaction that goes wrong. Ask your member service team what they would need to pick up where an agent left off. Those conversations are where the real work starts.
What comes next
Three things are going to happen in the next twelve months that make agentic readiness urgent, not optional.
First, the big AI platforms are going to publish their own versions of an agentic readiness scorecard. They will call it something else. It will be prettier than mine. But the institutions that rank poorly will find out publicly, which is not how most financial services leaders prefer to find out anything.
Second, the first wave of consumer-facing agents for financial services will go live. They will not be perfect. But they will start routing real transactions to the institutions they can transact with cleanly, and routing around the ones they cannot. That traffic will not come back.
Third, and this one is the most uncomfortable, the insurance and compliance industries will build their own agentic readiness requirements and start baking them into vendor reviews and renewal conversations. When your cyber insurance carrier starts asking about your agentic posture, the conversation is over.
The bottom line
AI visibility readiness was the last readiness gap. Agentic readiness is the next one. The institutions that start measuring and closing this gap now will be in a meaningfully different position this time next year than the ones that wait for the conversation to become impossible to avoid.
A year from now, I am going to write the follow-up to this piece. I suspect it will be about the institutions that moved early and the ones that did not. If you are reading this, you still get to choose which side of that piece you end up on.
Kevin Farley writes about AI readiness, agentic systems, and growth strategy for financial services. Read more on the blog.