Most “AI-native” claims I hear are branding. A landing page update. A new badge on the careers page. Two prompt engineers who report to marketing and run a Slack channel.

An AI-native team is something more specific. It’s a team where the default way work gets done assumes a model is carrying a part of the load. Take the model away and the motion breaks. That’s the bar.

When I do a readiness assessment on a B2B GTM team, I’m looking for five things. None of them are about tools.

1. Workflow redesign, not workflow acceleration

Has any revenue-generating motion been rebuilt assuming AI does part of the work? Rebuilt, not sped up.

If your SDRs still run the same eight-step cadence and just draft faster, you’re accelerating. That’s not readiness.

Rebuilt looks like this: the sequence only fires on scored accounts, the first touch is grounded in a specific signal, the SDR spends the time they used to spend typing on research that was never getting done. The motion is structurally different, not just faster.

When I audit, I look for a motion where I can draw an “old” and “new” workflow diagram and the diagrams are genuinely different shapes. If I can’t, the team is accelerating.

2. One operator accountable for one motion’s outcome

Find the person at the company whose job is “AI” and look at how they’re measured.

If the AI leader is being measured on tool adoption, seat activation, training hours delivered, or prompts published, the organization is not ready. Those are activity metrics. They tell you nothing about whether the business has changed.

If the AI leader is being measured on a specific revenue number, in a specific motion, over a specific horizon, now we’re talking. That’s a person with real accountability.

The difference between “VP of AI” who owns tool rollout and “VP of AI” who owns conversion rate in the enterprise funnel is the difference between a readiness score of 1 and a readiness score of 2.

3. Data that a model can actually use

This part is unsexy and kills more readiness programs than anything else.

Answer a simple question. How many unique accounts are in your funnel right now, across every system? If that takes three people and a week to figure out, your data is not ready for a model to use.

It doesn’t matter how good the AI strategy is if the underlying data has duplicates, gaps, and contradictions. You’ll get confident-sounding outputs grounded in broken inputs. That’s worse than having no output at all.

I ask three questions here. Can you get a single source of truth for account-level data in under an hour? Do you have intent and signal data integrated with your CRM? Is the data refresh frequency fast enough that the model’s output is actually usable by the person getting it? If any of those answers are no, the data work has to come first.

4. Literacy that lives in the work

I’m not looking for certifications or training hours. I’m looking at whether a senior individual contributor can explain how they modified their own workflow in the last month.

If the answer is “I use ChatGPT to rewrite emails,” literacy is shallow.

If the answer is “I built a sequence that only fires when the score passes 70, pulls context from an account research folder, and drafts using a style guide I wrote for my segment,” literacy is deep. That’s the person you want running a team.

One practical test. Pull a senior SDR aside and ask them to walk through one change they made this quarter. Their specificity will tell you everything about the literacy layer at your company.

5. Permission to break the org chart

AI-native work crosses functions. Marketing, RevOps, sales, CS, sometimes product.

If your change initiatives still require an exec steering committee to approve every redesign, the org doesn’t have enough operating permission to actually move. Readiness includes the political latitude to change how work flows across functional boundaries.

I’ve watched well-funded programs fail at this step. The strategy was right. The data was clean. The literacy was real. Leadership wouldn’t let a RevOps leader rewrite how the SDR org operated. The program stalled in committee.

How to use the assessment

Score each of the five on a zero-to-two scale. Zero means absent. One means started. Two means this is how work actually gets done.

Most teams I assess land between 2 and 4. A score of 7 or higher is genuinely rare. That is not a failure. It’s a starting point.

The real value of the exercise is forcing honest disagreement in the leadership team about where the gaps are. When the CEO rates readiness as 8 and the RevOps leader rates it as 3, that gap is the most important data the company has.

Where teams usually fall down

Data readiness and workflow redesign. Those are the two categories that kill programs early.

Literacy gaps can be closed with training, cohort by cohort. Accountability can be reshaped with new metrics. Permission can be unlocked by a single decision from the CEO. Data and workflow, though, those are quarter-long problems, and you can’t sprint your way past them.

What comes next

A readiness assessment isn’t the deliverable. The deliverable is a ninety day plan, targeted at the lowest-scoring of the five areas, with one named owner and one measurable outcome.

If the assessment ends with a report and nothing else, the spend was wasted. The only point of the scoring is to force the conversation that produces the plan.

The answer to “is your team AI-native” is almost always “not yet.” The useful answer is to know exactly where you aren’t, and to have a single owner working on one thing for the next quarter to change it.