train in your lane
AI Readiness Report
Prepared for Flint Group leadership · June 2026
you're already
in the top 1%.
23 of your leaders told us how they actually use AI. Here is what they said, where the gaps are, and what we are doing about it on Tuesday.
the topline
this is not an AI-curious team. it's a team of daily users.
83%
reach for AI several times a day. Another 13% a few times a week.
100%
use it to research or summarize long material. The single most universal habit.
96%
have used ChatGPT. 78% Claude, 65% Copilot. Most use three or more.
78%
rate their own confidence a 4 or 5 out of 5. Average: 4.1.
If you open ChatGPT every day, you are already in the top 1% of AI users on the planet. This team is there. That changes the job on Tuesday - we are not teaching them to use AI. We are teaching them what they have been getting away with not knowing.
the finding that matters
confident daily users,
running on the wrong model.
We asked everyone, in their own words, what they think actually happens when they type something in and the tool answers. The answers were the most important thing in this whole survey.
65%
described a search engine
Nearly two-thirds said the tool is "searching," "scanning," "scraping," or "combing the internet" to find an answer. That is what Google does. It is not what these tools do.
9%
described what's actually happening
Two people got near the truth: the model predicts the most likely next words from what it was trained on. It is not looking anything up. It does not "know" it is right.
"It uses info it's been trained on to predict the most likely answer. It doesn't actually understand it."
- a VP, Flint Group · one of only two answers that named the real mechanism
"I basically think it works like a better google search. It reaches out to the web, gathers data and presents it back."
- a business development lead
"A super duper quantum computer is generating an answer based on every other answer ever created on the internet."
- a VP, AdFarm
Why this matters: if you think it is searching, you trust it like a search engine. That is exactly the instinct that gets people burned by a confident, wrong answer - and exactly what we fix first.
the vocabulary gap
the words are ahead
of the mental model.
96% correctly named a confident, made-up answer as a "hallucination." The vocabulary is there. But could they explain each idea to a coworker? Here it gets thinner.
"large language model"
70%
"Training data" is the weakest concept in the room - and it is the one that explains everything else. 57% could not confidently explain where the answer comes from. That is the same 65% who think it is searching the internet, wearing a different hat.
the confidence trap
they know what it's good at.
they also believe a few myths.
Right instincts - strong agreement on the real strengths:
summarizing a long doc
91%
The myths - things AI is not reliably good at, but some marked as strengths:
perfect memory of chats
22%
52% of the room marked at least one of these as something AI is "genuinely good at." Each one is a classic place people get a clean, confident, wrong answer.
the risky middle
everyone's using it.
not everyone knows the line.
We gave them a real scenario: a client sends a contract with their full customer list and asks for a summary. And we asked, flat out, how clear they are on what is safe to put into an AI tool.
3.4/5
average self-rated clarity on what is safe vs. risky to enter. Not low enough to scare them, not high enough to trust.
43%
rated their own clarity a 3 or below. Nearly half are guessing at the line every time they paste something in.
30%
did not pick the safe move on the client-data scenario - including two who said, plainly, "I'm not sure what's safe here."
"They are popping up so fast I worry about someone making a decision that puts us at risk."
- a director of IT
"Honestly, I'm not sure what's safe here."
- said by two separate leaders
in their own words
what they asked us to cover.
Every leader told us the one thing they want to walk out able to do. Six themes came up again and again. These are not our talking points. They are theirs.
01where it's strong vs. where it's not
Repeated request for a clear map of where AI is genuinely good right now and where it falls down.
"I'd love a breakdown of where AI is currently very strong, and where it is not at all."
02what's safe to put in
The single most common worry. They want one clear company line on client data and confidential work.
"Knowing what the agreed-upon safe use of AI is within the organization."
03agents, and where they fit
Strong appetite for agentic AI in real workflows - and an honest read on what is real now vs. hype.
"Better understanding of how and where we can use agentic AI in our workflows."
04prompting that works the first time
People are spending too long wrestling answers into shape. They want fewer reworks.
"I spend too much time prompting. Get it closer to what I need the first time I ask."
05leadership, adoption & ROI
How do you push AI adoption against billable-hour pressure, and where does it actually pay off?
"What makes sense to use AI for vs. keep doing yourself - and how we push it while pushing billable work."
06augment thinking, protect the craft
The creative leaders want AI to sharpen human ideas, not replace them. A real, recurring concern.
"I want AI to challenge human ideas or point out blind spots, rather than the other way around."
the use cases, in your world
six plays your team
runs every week.
The class is not theory. It is six use cases from the cowork method, each rebuilt as a workspace that already holds your files, your voice, and your context. Below, the plays from Tuesday's deck translated into agency work - and lined up against what your team already told us they do in AI. Same work. Far less of it.
PLAY 01
brief to concept directions
Drop the client brief and the brand's past work into the workspace. Get three on-strategy directions to react to, instead of a blank page.
91% already brainstorm with AI · 100% summarize
PLAY 02
copy & email in the client's voice
Teach a workspace the account's tone once. Every draft sounds like the client, not like a robot reading a brochure - and you stop re-explaining voice.
91% write with AI, but only 52% trust it on tone
PLAY 03
inbox & client-comms triage
Clear the pile with replies already drafted in your voice, flagged for what actually needs a human. The deck's worked example: a full inbox in about 25 minutes.
57% already use AI for email
≈ 3 hrs → ~25 min
PLAY 04
call prep & recap notes
Walk into the client call briefed on the account. Walk out with the recap and next steps already written, not a blank notes doc you'll fill in Friday.
100% use AI to research / summarize
≈ 4 hrs → ~40 min
PLAY 05
content engine + personal branding
Social, newsletters, and leadership LinkedIn from one workspace that holds the voice. The blank-box stare, gone.
35% use it for social; creatives want leverage without losing craft
≈ 2 hrs → ~20 min
PLAY 06
reporting & analytics cleanup
Turn the dashboard export into the client-ready, cross-discipline insight. The quarterly cleanup nobody gets to becomes a 10-minute Friday habit.
70% already use AI for data / reporting
quarterly-or-never → 10 min
~10 hrs
back on the calendar, per person, per week. Roughly a day. This is the exact ROI question your finance and client leaders asked - and the honest answer to "how do we push AI against billable pressure."
tuesday, mapped to your data
the class, lined up
against your survey.
We took the cowork class you're delivering and put every segment next to these 23 responses. Nothing here is generic. Each block is in the room because your team asked for it or the data shows they need it. The method runs in one order, every time:
E · education - plain language first
A · application - laptops open, you try it
T · transformation - intuition that sticks after we leave
what cowork actually is
a workspace per part of the business - not another chat window.
because the survey showed
100% use AI to research and summarize, but everyone re-explains who they are every time. We start by turning that into a teammate that already knows.
how it works + where you trip
prediction, not search. why it hallucinates. the three myths to drop.
because the survey showed
65% described a search engine and 52% marked a myth as a strength. This is the deck's "where you trip yourselves" segment, built from your own answers.
your #1 ask
security & safe use
what's safe to paste, what never is, and one clear rule for client data.
because the survey showed
clarity averaged 3.4/5, 43% rated themselves a 3 or below, and 30% mishandled the client-list scenario. Ingrid trains your team to the line, so nobody guesses with client data again.
the six plays, hands-on
brief matching, voice, inbox, call recaps, content, reporting.
because the survey showed
each play maps to what your team already uses AI for - writing 91%, research 100%, data 70%, email 57%. We make each one a workspace, not a one-off.
four prompts you keep
templates that get it right the first time, in your voice.
because the survey showed
"I spend too much time prompting" came up again and again. Your team leaves with prompts that work on Monday, not theory.
agents & what's next
what agentic AI can really do now - and what's still hype.
because the survey showed
agents were the most-requested advanced topic across the leadership team. An honest read for the people deciding where to invest.
the math & the adoption case
~10 hrs a week back, and how to lead the change.
because the survey showed
your finance and client leaders asked the ROI question directly. We put numbers to it and give leadership a way to push adoption that respects billable work.
the bottom line
your team is already in the top 1%.
now we close the gap between using it and understanding it.
They use it daily and they trust it. The one thing most of them are missing is the simplest one: what it is actually doing. That is a half-day fix - and it is exactly what Tuesday is for.