Half of every company's 'AI wishlist' is just regular software they never bothered to build. Slapping AI on the request does not make the underlying problem more sophisticated. It makes it more expensive.
Honest question for agency owners. When was the last time a deal closed purely from inbound, no relationship, no warm intro, no prior conversation? I will wait. I think the wait says everything.
Everyone is racing to build the AI agency. Nobody wants to do the boring part first: deliver the service by hand until you actually understand where it breaks. You cannot automate a process you have never run.
The question that separates the agencies that make it from the ones that do not:
Can you draw a clear line between the intelligence work in your delivery (pattern recognition, rule application) and the judgement work (strategic taste built over years)?
If you cannot draw that line, you cannot protect the judgement and you cannot systematize the intelligence. You are just hoping the wave does not reach your shore.
Killed two prospects out of my cadence today. Not because they said no. Because I finally admitted to myself they were never going to say yes. The follow-up you are too scared to stop is just hope wearing a calendar reminder.
Spent the morning sending the same product invite link to six different people with a friendly 'no pressure, would love your feedback.' There is pressure. There is always pressure. We are founders, not monks.
@TheGeoMethod Buyers are very educated now and also most people in leadership roles have grown up in the tech age so they know how to find what they need.
Buyers show up to the sales call having already done 80% of the research. The job stopped being 'inform them' a decade ago. The job now is to understand their problem better than they do. That is consulting, not selling.
I sold a client on a system faster than they could actually use it. Demoed the shiny version while she sat there asking what to do when she opens the spreadsheet tomorrow. Moving fast is not the same as being helpful.
Software engineering is primarily intelligence work. That is why it hit the AI threshold first.
Agency work? Also primarily intelligence work. Just nobody wants to admit it because we have been calling it "strategy" for twenty years.
Your competitor in three years is not another agency with better AI tools.
Your competitor is an autopilot company that sold the same outcome directly to your client and does not need you in the middle.
A copilot puts AI in the hands of a professional. The professional still does the work.
An autopilot skips the professional and sells the outcome directly to the buyer.
Most agencies using AI right now are accidentally becoming copilot shops. They think they are winning. They are running in place.
The math most agency founders are ignoring:
For every $1 a company spends on software, roughly $6 goes to services.
You are the $6. The AI-native companies building autopilot businesses are coming for that budget. Not because your work is bad. Because your work is mostly pattern recognition and they can do it cheaper.
B2B marketing in 2026: 'we're an AI-first agency.' last year: 'data-driven.' year before: 'strategy-first.' the noun changes. the founder still does all the selling.
the moat in AI services is not the model. the model is rented. the moat is the context layer that knows your business well enough that switching models in the basement does not change the output upstairs.
keys to a great diagnostic engagement:
1) talk to the leadership
2) find the actual problem
3) deliver the answer
4) compress it from a month into a single afternoon because the client was never going to read the deck anyway
if you are a CEO and you are not doing your own outbound, you do not have a sales team. you have a list of people waiting for you to apologize for the silence.
opened a 67-minute meeting today by telling my partners i was extremely exhausted. we then proceeded to make 17 decisions. nobody mentioned the exhaustion again. this is also a decision.
every agency founder is convinced their sales process is broken. it's usually not. the founder is just the entire sales department and nobody told him that's the actual problem.