AI has a lot of agency owners re-checking the math, the margins, and the messy parts of the business they used to ignore. When budgets feel tighter (or when we can see that coming), “do more” stops sounding like a plan and starts sounding like risk. The better move often looks like simplifying instead of scaling, at least for a season.

In this post, we’re walking through how we can decide what to cut, what to keep, and what to package up into repeatable offers. We’ll also cover why AI fear is real, why it doesn’t always match what’s happening in our client work, and how to define success without tying it only to revenue.

Why agencies are rethinking everything as AI changes expectations

The first signal that something needs to change usually isn’t a headline about AI. It’s that moment when we look at our monthly software stack and realize we’ve slowly built a second business that exists only to pay subscriptions.

Kurt von Ahnen shared a very real version of that. He runs many of his plugin and software renewals through one credit card so nothing lapses. Then he periodically audits it, pays it down, and keeps moving. Two weeks before this conversation, that audit turned into a “big gulp” moment: how much are we paying, what are we actually using, and what is truly returning revenue through client work?

That audit forces a deeper choice. Not “should we learn AI,” but this tougher question: are we trying to be the expert in every new thing forever, or are we choosing the lane that already pays and fits our strengths?

Kurt framed it as a trade, not a loss. If we say “no” to one opportunity, we can say “yes” to the work that actually blesses the business and the team. That is hard, because it feels like we’re walking away from potential. Still, the alternative is death by a thousand tools, a thousand edge cases, and a thousand half-promises.

When we simplify, we’re not just cutting costs. We’re reducing the brainpower it takes to deliver consistently, especially when the market is noisy and everyone wants “AI added” to the scope with no clear definition.

If we can’t connect a tool, a service, or a new offering to real client outcomes, it’s probably not a “growth move.” It’s a distraction with a monthly bill.

The real fear of AI, and why it might not hurt you yet

Toby Cryns put words to something many of us feel but don’t always say: the fear that if we don’t invest the next eight weeks learning AI deeply, our agency will fail. At the same time, he also admitted something that matters even more, he’s not seeing AI disrupt his business right now.

That gap between fear and lived reality is the tension we’re all sitting in. We hear bold claims. Kurt referenced a conversation where Spencer Foreman (WP Launchify) talked about using an AI “friend” to code whatever he wants inside WordPress, to the point that plugins become optional. Whether or not we’re at that level, it makes us wonder if we’re behind, or if our current stack is a mistake.

Here’s the practical counterpoint we can’t ignore: even if AI can generate code fast, we still have to own the output. That means maintenance, compatibility checks, updates, and the responsibility when something breaks after the next WordPress release.

A simple way to compare the two approaches looks like this:

1. AI-built custom code: We pay for credits and time, then we supervise, test, and keep maintaining it as the ecosystem changes.
2. Proven plugins and platforms: We still vet them, but we offload long-term upkeep to teams who wake up every day thinking about that product.

This is why Kurt landed on a steady view: if we keep doing the work that drives revenue today, AI probably won’t erase that path overnight. He even pointed to a long-range truth many of us have seen: enterprise clients stick with platforms for decades. He’s helping organizations migrate off systems they bought 25 to 30 years ago. So it’s not unrealistic to think a well-built WordPress system could last 15 to 20 years in the real world, especially in enterprise training and internal learning where traffic doesn’t depend on search.

If we want a broader view of where agencies are putting AI to work (and where it can add complexity), we can compare notes with pieces like Cloudways’ overview of AI and automation for agencies. We don’t need to agree with every prediction to benefit from seeing patterns.

Client stories that prove the human touch still wins

Trends are interesting, but client behavior tells the truth faster than any headline. Toby shared a story that every agency can relate to because it’s not glamorous, it’s normal work.

A maintenance client emailed saying they’d had “a lot of downtime.” Since they also had a hosting contract, Toby started troubleshooting. After about an hour, he traced the issue down to DNS. A record was pointing to WP Engine. The client then confirmed they were hosted at WP Engine, which raised a bigger question: why did they still have hosting with The Mighty Mo?

It turned out they’d switched hosts two months earlier to a massive provider, and the problems started after the move. The “downtime” concern even had nuance. The uptime was around 99.8% over three months, roughly 30 hours total, and it was happening around midnight or 1:00 a.m. The client didn’t notice it in real time, but an uptime monitor did, and the alerts triggered the complaint.

The punchline matters: they’re likely coming back. Not because the big host “doesn’t use AI,” but because the experience didn’t meet expectations, and the relationship is human. When someone can email us, get a real answer, and feel understood, that relationship often beats “enterprise-grade” branding.

Kurt shared a different human-side boundary: saying “no” to self-proclaimed vibe coders who want him to build an open-ended product off vague ideas. He’s willing to do hourly work if the client understands what that means, but when someone tries to negotiate a “multi-million dollar product” out of fuzzy inputs, that’s a red flag.

At the same time, Kurt also admitted the other side of the issue: saying “yes” to too many small projects. Even when margins look fine, small projects don’t always carry enough revenue to sustain growth. Sometimes simplification isn’t only about cutting services. It’s about raising the floor on what we agree to take on.

The hidden cost of complexity (and why processes are the real product)

When we add services, we don’t just add billable tasks. We add complexity that has to live somewhere, in our brains, in our project management, and in our documentation.

Toby and Kurt took different paths here. Toby has a team and is learning to let the team build and own processes. His project manager taking over the design process is a good example of growth done the right way, because it moves ownership closer to the work.

Kurt described the solopreneur problem clearly. Much of his work is him: public speaking, coaching, corporate training. He can’t “send a trainer” in his place without changing the offer. So he’s starting to bring younger local talent into the business to offload what doesn’t require his face and voice. That’s not a trendy move, it’s a survival move.

They also named the thing many agencies feel but don’t price properly: the complexity tax. Even with a team, processes take time to develop, document, share, and keep current. In other words, complexity charges interest.

Kurt shared how he documents operations at Manana No Mas. He’ll put his son on a Zoom call and record task walkthroughs, then turn them into lessons inside a course. Those lessons cover how they do things, for example: setting up hosting packages, reserving domains, installing WordPress, and adding LifterLMS.

That idea connects well to what we see in established platforms. LifterLMS, for instance, has deep documentation across core features, add-ons, troubleshooting, known conflicts, and theme compatibility. It also integrates with widely used WordPress tools (page builders like Beaver Builder and Elementor, community tools like bbPress and BuddyPress, and commerce and payment options like Stripe, PayPal, and Authorize.Net). The point isn’t that we need every feature. The point is that mature tools reduce the “tribal knowledge” problem because guidance exists, and support patterns are known.

If we want our agency to be sellable, or simply easier to hire into, that same principle applies: what we do repeatedly should be teachable without us sitting in every chair.

How we stop being a Swiss Army knife (without killing sales)

Early on, saying yes to everything is normal. We’re learning. We’re building cash flow. Clients also shape our direction because we follow the money to stay alive.

The trap shows up when “yes” becomes our brand. We start doing websites, hosting, maintenance, SEO, paid ads, custom apps, brand strategy, CRM automation, and whatever else shows up in the inbox. The offer becomes a menu, and the menu becomes chaos.

Kurt gave a concrete example: SEO. His agency has done it, including using contractors, but he found it hard to meet expectations when results moved backward due to algorithm shifts and new search behavior (including AI summaries changing click patterns). If the client is paying real money and the story is hard to explain, the service can become a liability. So he has largely stopped selling SEO for the last year and a half to two years, because it doesn’t match his core promise: getting work done on time and under a known budget.

That ties into a second nuance: niche changes requirements. Kurt’s enterprise e-learning clients often don’t care about SEO at all because training demand comes from dealer networks, internal teams, or existing channels. Meanwhile, course creators selling to the public often need traffic, and the agency can get pulled back into broader business consulting.

So what replaces the Swiss Army knife approach?

Productized services plus a standard stack

Kurt described hosts that support “blueprints,” basically a way to replicate a known-good site package. Once a blueprint exists (food truck sites were his example), we can launch quickly and swap the key inputs: logo, images, menu, and core details.

Toby added the deeper truth: the process is the product. In pitch meetings, we don’t win trust by listing plugins. We win trust by explaining what week one looks like, what week two looks like, and what launch looks like. That predictability is what many buyers are purchasing, even when they think they’re buying “a website.”

A standard stack supports that promise. Kurt talked about becoming an expert in a small set of powerful plugins. That means fewer debates about which calendar tool to use, or which event system is best. The agency decides. The client gets results.

For a broader take on how some teams are thinking about automation and agency operations, we can compare viewpoints like E2M’s discussion of WordPress automation and MCP. We don’t need to adopt the same tooling to learn from the “reduce repetitive work” mindset.

Client boundaries: don’t overshare, don’t overpromise, don’t skip the onboarding

Simplifying isn’t only internal. It’s also about simplifying the client experience and protecting the business model.

Kurt made a strong point: when we give away too much detail about our tools and step-by-step methods, we can train clients to replace us. If we tell someone exactly how we migrate sites, then follow up two weeks later, we shouldn’t be shocked when they say, “Our internal person already did it.” That’s money gone, and it’s also a signal that we sold tactics instead of expertise.

Toby shared a similar frustration. He was hosting a compliance-heavy website that required server hardening and extra care. Another agency tried to stage and push updates but didn’t know how to get content from local to server. Toby ended up teaching them how to do a basic task, which felt odd because he was helping a competitor deliver on a client relationship he wasn’t paid to manage.

None of this means we should be secretive or rude. It means we should be intentional. Clients don’t need a shopping list of plugins. They need confidence that we’re accountable.

Kurt’s client handoff example shows how to do that with clarity. A client dealing with serious health issues needed to transfer ownership. The new party wanted a rebrand and a new URL. Kurt suggested reserving the new URL, then pointing the hosting package to it. He also required the new owner to choose a hosting package and attend an onboarding call before the transfer.

That’s a boundary with a purpose: we don’t transfer an account to someone we’ve never met, and we don’t become a Swiss Army knife during the handoff. We sell what we do, then we deliver it well.

Redefining success: leaner can still mean growth

Near the end of the conversation, Toby asked the question that sits underneath all the tactics: how do we redefine success when “bigger” stops being the only goal?

He mentioned tracking unusual but useful metrics, like credit card fees (more fees can mean more revenue moving through the business), or even simple life markers like vacation time. He’s also watching personal health more closely, gym time, yoga, and whether work supports family time.

Kurt’s answer was honest: his sense of success changes depending on the day. One Friday afternoon feels perfect, grill going, beverage in hand, dog in the yard. Another day looks like staring at credit card balances, unpaid invoices, and asking why he doesn’t just get a job.

His solution wasn’t hype. It was discipline: journaling and tracking KPIs so emotions don’t run the company. He also used a metaphor that hits home for many agency owners: the duck looks calm above water, but below the surface the feet are moving nonstop.

The final reminder was simple and blunt. At the end of our lives, the slideshow is mostly family, kids, hobbies, and memories. It probably won’t be a photo of us alone next to a keyboard.

Conclusion

If we want to stop being a Swiss Army knife agency, we have to choose what we’re willing to be known for. That starts with auditing our tools, cutting services that don’t fit our promise, and building offers around a repeatable process, not a grab bag of “whatever you need.” AI will keep changing how work gets done, but clients will still pay for clarity, trust, and outcomes.

If we’re simplifying this year, we should pick one metric that proves it’s working, whether that’s higher average project size, fewer tool subscriptions, more delivery consistency, or more time back with family. Then we can let that metric, not the fear cycle, guide the next decision.