Floating between LA, Lima, and San Miguel, while arguing with the algorithm. Notes from a studio in central Mexico, written for the kinds of businesses that keep neighborhoods alive.

Patterns in the reviews.

I've been reading local business reviews for thirteen years. At some point they start to blur together in ways that are more interesting than embarrassing. A patient describes a hygienist the way a diner describes a server: attentive, unhurried, made me feel like I wasn't being rushed out. A one-star review of a ramen bowl and a one-star review of an implant use almost identical language about feeling like a number.

The businesses are different. The anxiety the customer walks in with is not. Dental chairs and restaurant tables are both places where people make themselves slightly vulnerable, hand something over, and wait to see if they were right to trust. The whole transaction runs on that current.

Which means the marketing runs on it too. Not the ads, not the keywords, not the review count. The thing underneath all of it: whether the business has earned the right to be recommended. You can't manufacture that. You can only make it visible, or waste money pretending you don't need it.

At least I can enjoy playoff hockey now.

I remember watching Anze Kopitar go around Chris Pronger like he wasn't there and knowing immediately that something had changed. Pronger was the best defenseman in the world at the time, maybe the most physically imposing player in the league, and this kid from Slovenia just walked around him like it was a Tuesday. You don't forget moments like that. You file them away and spend the next two decades watching to see if you were right.

You were right.

The sweep hurt less than it should have, which is its own kind of verdict on the season. Colorado was always going to win that series. They were better built, better coached, and playing with a clarity the Kings couldn't match. The outcome was settled sometime in February, probably earlier. What the first round did was make it official.

The real autopsy starts with Holland, who replaced Blake and then spent his tenure not changing the coach when the evidence was already in. The roster had cap room, had promising players, had every reason to take a step forward, and instead took one back. Ceci and Dumoulin were bad signings in ways that were foreseeable. The team never gelled. One or two players outperformed all year. Everyone else regressed, or phoned it in, through a season that never found an identity.

Now it belongs to Quinton Byfield and Brandt Clarke. Byfield has the ability to be one of the best players in the league, but the window to prove it is getting shorter than it should be at his age. Clarke is becoming something real on the blue line. The question is whether the organization builds around them with any urgency or spends another two years being politely mediocre.

Kopitar deserved better than this exit.

Godspeed, life after Covid.

I've seen more concerts than I can count. I saw Godspeed You! Black Emperor in a barn in Pioneertown about seven years ago, pitched a tent in the corral afterward, and thought that was probably the definitive version of that experience. I was wrong.

January 23rd, 2026. Foro ARPA, Querétaro. I drove in from San Miguel not expecting much beyond the music, which would have been enough. What I didn't expect was to walk into a room that felt like 2019.

Reasonable beers. Friendly bartenders. A crowd that was there to listen, not document. Nobody had their phone out, or if they did I didn't notice, which amounts to the same thing. The room was full of people who had shown up to participate in something, which sounds like a low bar until you remember how many shows since things opened back up have felt like everyone was slightly somewhere else.

Godspeed gives you something to participate in. The music builds the way weather builds, slow and inevitable, apocalyptic and somehow always tilting toward hope at the last possible moment. You don't watch it so much as get caught in it. The crowd in Querétaro understood that. They leaned in instead of holding up screens.

I've been to a lot of shows since Covid. Most of them were fine. This one felt like something we had taken back.

What happens to local marketing in 2027.

By the end of 2027, the most successful local marketing agencies will have rebuilt their entire workflow around AI. Not as a tool layered on top of existing process, but as the organizing principle the process is built around. The agencies that treat AI as a productivity layer inside the old workflow will end up doing the same work for the same money on a faster clock, watching their margins compress while their competitors restructure entirely.

This happens in local marketing first, before B2B, before brand work, before anything else. The unit economics are tightest here. The willingness to experiment is highest. A dental practice paying $4,000 a month is not patient. They will reward whoever figures it out first.

Writing this down so I can be wrong on the record.

The thing about AI search nobody is talking about.

Dental patients are starting to use ChatGPT to translate their symptoms into vocabulary before they ever search Google. They describe what's bothering them in plain language, get a rough diagnosis or a set of terms back, and then search the term. By the time they hit a search engine, they're already inside a category they didn't choose for themselves.

Here's what that means for reviews. Reviews used to feed the website, which fed the search ranking, which brought in the next patient. That loop is breaking. Patients are now extracting their information from ChatGPT directly, which means the website is increasingly not the first touchpoint. But reviews still matter, maybe more than before, because they're one of the primary sources ChatGPT is pulling from when it builds its answer. The practice with 400 detailed reviews is training the model differently than the practice with 40 thin ones.

The search queries that bring patients in will increasingly be terms ChatGPT supplied, which means ranking for those phrasings is a different optimization problem than it was two years ago. I don't see anyone in the dental marketing space talking about this. I think they should be.

What I watch with growing unease.

Small business owners are letting AI write their replies to negative reviews. The replies are competent. Warm, appropriately apologetic, they offer to make things right. Indistinguishable from a thoughtful human reply. And in aggregate, they're draining the last bit of authentic voice out of local commerce.

We do this too. I want to be honest about that. AI helps us respond to negative reviews for clients and the replies are better than what most business owners would write at 11pm after a long day. I don't think that's wrong exactly.

What I can't stop thinking about is where it ends. The reviewer wrote their review as a person. They were angry, they had specifics, they were trying to be heard. Right now what they get back is a perfectly tuned piece of conflict-resolution copy. But the reviewer is increasingly using AI to write the review in the first place. Which means we're already closer than anyone wants to admit to a world where two language models are exchanging pleasantries as avatars of a business relationship that neither of them experienced. I notice that, and the noticing has not stopped.