Part of our team spent 21 and 22 July at DotDev, Shopify's developer event in Toronto. We had said beforehand that we would be there as part of building a US presence, so here is the follow-up.
Three things were worth the flight: what the workshops showed is now shippable, one uncomfortable session about AI discovery, and a series of conversations with other agency founders that were more useful than any talk.
[Add event-wide photo here]
Merchants are adding apps, not consolidating
The number that reframed the event: high-GMV shops now run 37 apps on average, three more than last year. Across all shops the average is seven, up two. Shopify paid developers $1.31 billion in 2025, a 24% increase.
Nobody is simplifying their stack. So the apps that last are the ones that reduce friction between tools rather than adding another dashboard. Shopify named three ways to do that: Sidekick extensions, admin intents, and rendering your app's data inside Shopify Analytics instead of your own reporting screen.
Shopify never said it outright, but reading between the slides, how deeply an app uses these integrations looks like it is becoming part of how apps get discovered. They are not just features, they are distribution.
Michal [surname], Digismoothie
The five workshops, and what is actually shippable
Each workshop came with a repository and a build path, so you left knowing the implementation cost rather than just that a feature exists. Status matters more than the feature list here.
Embedded app analytics. Production-ready. Write to order metafields, query with ShopifyQL, render metric cards in your own UI. We are looking at this as a second surface for Setpilot data.
Collection Sources. Production-ready in the 2026-07 API version, and you need that exact version because older ones filter these collections out. Your app writes a signal metafield and Shopify decides collection membership. Not available on Starter or Retail plans. The cleanest idea of the event.
Market-driven shipping. Preview only, GraphiQL, development stores. App-owned shipping profiles that price a whole catalog and coexist with merchant rates. Merchant rollout starts 1 October 2026, everyone by 1 July 2027.
This is the one I would put on every developer's radar, and not because of the feature. The merchant-owned deliveryProfile APIs are being deprecated under it. If you have a shipping integration in production today, that is a migration on your roadmap.
Josef Rousek, Digismoothie
Dynamic Discounts. Production-ready. Product, order, and shipping discount classes in one Shopify Function, wired to Flow and a theme block. Useful when a merchant has a repeatable rule spanning several discount types.
B2B pre-orders. Production-ready, Shopify Plus only. Pre-book without selling plans: delayed fulfillment, payment terms, charge on ship.
[Add photo from a workshop here]
The session that sent people back to their own stores
AI crawlers read core HTML, structured data, and server-rendered text. Anything injected client-side is invisible to them, which quietly disqualifies a lot of product pages.
Client-side FAQ and review widgets are basically invisible to AI crawlers, and almost every store has at least one. It is a fifteen-minute audit with a large blind spot behind it. Turn JavaScript off and look at your own product page, because that is roughly what the crawler sees.
Michal [surname], Digismoothie
The other half is your robots.txt, which was probably written before Claude-SearchBot, OAI-SearchBot, GPTBot and Storebot-Google existed. And the phrasing point we keep repeating internally: shoppers bring problems, not keywords, and specific language wins. “Glazed donut skin” gets found, “hydrated skin” does not.
What other agency founders are doing with AI
The hallway conversations were remarkably consistent. Everyone is investing in upskilling the people they already have, and nobody is using AI to sell different work. They are using it to do the same work with less of it.
Pricing is where that stops being theory.
If a task took ten hours before and now takes five plus tool costs, most agencies have settled somewhere around seven. Nobody bills either extreme. Client-facing, tokens get converted into hours: two hours plus $80 of tokens becomes three hours on the invoice.
Tomáš Pěnka, Digismoothie
Two other patterns stuck. Flat higher-tier plans beat token pricing at agency scale, so teams are giving key people the top plans rather than policing usage. And the agencies with the sharpest pitch specialize in two to four industries, then recycle each win into a case study for similar merchants in other countries, arriving with a designed answer to the client's biggest problems rather than a capability list.
The one that validated our own approach: Adrian from e-tante.de, a platinum partner in Germany, moved from a one to two week discovery to a five or six week process with multiple workshops, and says it works better. Time spent understanding the problem upfront is the cheapest time in the project.
[Add group or networking photo here]
The people, briefly
We talked partnership expansion with the LoyaltyLion founders, spent time with the Klaviyo team, and had a genuinely useful conversation with Bharati Balakrishnan, who leads Global Partnerships for B2B at Shopify, about the US B2B market. Thanks to Damir for arranging that one. Plenty of others, including our partner manager Marina Vavilova and Justin D. Harris, who leads Shopify's 100-person Sidekick team.
The closing party spread across two exhibition buildings with around 8,000 people, 2,000 partners and 6,000 Shopify employees. Biggest single-company event any of us had been to.
[Add closing-party photo here]
What went on our list
Audit server-rendered product data and robots.txt coverage, for us and for clients. Look at admin analytics as a distribution surface for our apps, not just a reporting feature. Check our shipping integrations against the market-driven shipping deprecation timeline.
If you want this kind of conversation closer to home, we host it too. We ran the Shopify CEE meetup at our Prague office earlier this year, and we will do it again.
