MACH X 2026: Play, Plan, and The Coming Agentic Dam Burst

Play or Plan?

Steve Tannock, the VP of Engineering at TELUS Digital (the IT Services arm of TELUS, the giant Canadian Telco), opened the floodgates of AI to every employee. Unlimited tokens and a vast array of models are made available to everyone, and not just for work purposes. Employees are encouraged to play and learn, and build on AI, including silly or personal projects. Play is how you learn. This has led to massive adoption of AI without mandates, and delivered a ton of valuable, high ROI, work solutions.

In the very next keynote session Chrissy Munroe, Senior Director at Loblaw Digital (the Digital arm of Loblaw Companies, Canada’s largest retailer specializing in supermarkets, but also including pharmacies, banking, and more), laid out the importance of carefully establishing business problems, KPIs, ROI, and going through a comprehensive review process before getting approval for building.

Two completely different approaches to AI enablement, and both with very valid reasoning behind them. And a perfect example of where the industry is at with regards to how to effectively leverage AI: We don’t have a clear universal answer!

The MACH X 2026 Conference in Toronto brought together a fantastic mix of retailers, telcos, digital agencies, composable vendors, and the wonderful MACH Alliance team, in the striking and historic Design Exchange building. The building used to house the Toronto Stock Exchange back in the 1930s, and is a unique mix of Art Deco, Streamlined Moderne, and Stripped Classical architectures. A very fitting venue for a conference focused on evolving architectures, and billions of dollars of ARR.

The attendees were a mix of technologists, sales, marketing, and strategy leaders, all of whom have been very focused on AI and what AI means to their businesses for the past few years. One of the key differences between most conferences and MACH X, is the willingness to share openly what worked, what didn’t, and pool our knowledge. This is especially important when dealing with an emergent and evolving catalyst like AI. When we all share our knowledge, ask honest questions, and communicate, everyone walks away better off.

While I don’t think we know the best way to enable AI within your company, balancing costs and well thought out ROI versus encouraging experimentation and play, I did come away from my time at MACH X with many ideas, inspirations, and important takeaways. Here are a few big ones:

Agentic Commerce: The Dam Will Burst!

One of the speakers on the panel “Towards a Fully Enabled Enterprise”, Scott Adel, AVP of AI Activation at Canadian Tire Corporation, explained why we need to prepare for Agentic Commerce NOW. While Agentic Commerce has been a hot topic for a little while now, to date actual adoption numbers have been small and some early solutions have either failed to gain traction or been rolled back entirely. This has led to a lot of people holding off on investing in enabling Agentic Commerce, quite understandably.

However, the extremely eloquent point which was made during this panel session is that the use of AI tools and solutions has spread like wildfire, for work and personal use, across virtually all demographics. People exposed to ChatGPT at work are now using it on their phones and at home to replace Google, IMDB, and Reddit (for information retrieval). Non-technical people, young and old, are using AI tools and solutions to create images, videos, stories, applications, automations, and 100s of other things. And based on this, it’s only a matter of time, and probably not a long time, before the dam preventing Agentic Commerce bursts, and you have hundreds of millions of consumers (and companies) suddenly ready to transact through Agents. The businesses and brands which are ready will capture the lion’s share of this flood of spending.

So NOW is the time to assess your company’s readiness for Agentic Commerce, and likely to invest in improving your catalog data quality, and preparing your systems for Agents (AEO, GEO, UCP, ACP, APP, and more).

On the vendor side, MACH Alliance’s new Agent Ready certification is very technical and practical validation of your solutions. Thankfully not a pay to play rubber stamp, but an in depth review that should carry real weight in the market.

Optimize for Answers, not Traffic

The importance of AEO and GEO came up over and over again. AI Chat apps, like ChatGPT, Claude, and Gemini have already dramatically changed how people, both individual consumers as well as B2B purchasers, search for answers, brands, products, and solutions. It’s no longer enough to focus only on SEO and Google result ranking, if you aren’t cited by the LLM or one of the 2-5 brands or products it mentions in its response, you aren’t getting any of that traffic or brand recognition.

On the upside the fundamentals for strong AEO and GEO should largely align with your SEO needs, and a focus on data quality, so putting the work in has positive impacts in other areas, not just AEO and GEO.

There is currently a critical measurement gap to be aware of. Traditional analytics don’t see the majority of AI referred traffic, as many chat apps intentionally don’t pass referral headers or other tracking data for privacy reasons. There’s a new generation of tools to help address that gap, although today a lot of it is guesswork. It’s also early days, so there aren’t a lot of bulletproof papers on the strong ROI of investing in AEO and GEO. All that said, failing to invest in AEO and GEO will absolutely hurt you, and it shouldn’t cost that much to identify and bridge the gaps you may have today.

McFadyen offers an AEO/GEO Optimization Service, and we have a free AI Readiness tool which can help identify the AEO and GEO gaps your eCommerce site has today. We also recently spoke on an MDM Webinar on the details of AEO and GEO for Distributors. You can watch the OnDemand webinar on MDM’s website.

Architect for Change

Another recurring theme was “Build it so change is free” as Isaac Smith from Klue put it. The AI landscape is changing so rapidly, with new models, new tools, pricing changes, and unfortunate provider outages, that when you’re architecting your systems, it’s critical to design for resilience and future-proofing.

Specifically, if possible, abstract away from individual providers or models. Expose generic functions internally to your services and solutions, and allow your AI API layer to handle things like failover, model changes, automatic model selection based on price and response time optimizations, etc.. This layer is also where you need to put observability, logging, tracing, security, and validations. This is one of the big differences between a simple AI POC and a real enterprise ready AI solution.

It’s also a great example of the MACH ethos applied to AI. Composable AI LLMs.

Tidbits

The MACH Alliance led Agentic Hackathons created some impressive solutions, most of which have already gone into production with great measurable effect. Here is one which is fully open sourced: https://github.com/machalliance/solution-studio-product-questions

On-site chat bots should often be designed with multiple use-case specific backends, and a simple orchestrator routing messages appropriately. Think product discovery vs order support vs FAQ content.

AI, and LLMs, are often used when they are the wrong tool for the job. ML, Expert Systems, and normal deterministic programming can often be better, cheaper, and faster ways to solve problems. Don’t let the hype blind your solution design.

Closing Thoughts

One thing that really stood out for me at MACH X, was how many people were talking about actual deployed production systems, and not just more hype and smoke. It’s great to see AI based solutions and Agentic services getting built, deployed, and seeing real world results. People sharing the lessons learned, and the real wins, of what they have actually built and deployed, means so much more than analyst theories and solution sales hype.

One thing I heard over and over, from at least a dozen different people was, “If you aren’t fully embracing AI in your role, in 12 months you’ll be replaced by someone who is”. It’s not meant as a threat, just a realistic view of how things are changing.

If you’re working through any of these topics of questions at your organization, we’d love to compare notes.