I walked into the Toronto Tech Week workshop expecting a room full of founders who were curious about AI marketing but had not yet started. What I found in front of 118 of them was something more interesting, and more instructive, than that.
Most of the people in the room had started. They had tried AI tools, experimented with ChatGPT for content, maybe set up a basic automation or two. What they had not done was build anything that worked consistently without constant manual attention. They had tools. They did not have systems. And they had strong opinions, many of them wrong, about why the gap existed.
These are the field notes from that session. What the room believed, what surprised me, and what the pattern of misconceptions reveals about where most founders are stuck.
Misconception One: AI Content Means Lower Quality, Full Stop
The most common belief in the room was that AI-generated content is categorically lower quality than human-written content, and that using it means accepting a quality trade-off. This belief was held not just by skeptics but by some of the founders who were already using AI tools. They described their AI content as "good enough" in a tone that made clear they had set their expectations low.
I have a particular reason to push back here. I was a copywriter before AI could write a coherent paragraph, which means I spent years fixing robotic copy by hand. When the models arrived, I already knew exactly what good copy sounds like and exactly where AI drafts go wrong. That is the whole trick. The model does not know what good sounds like. You do, or you should.
The reality is more nuanced, and it matters. AI-generated first drafts are often not good. They are generic, they hedge constantly, they lack the specificity and voice that makes content worth reading. But AI-assisted content, where a human provides the specific angle, the distinctive examples, and the final editorial pass, is frequently indistinguishable from fully human-written content and sometimes better, because the AI handles the structural scaffolding while the human focuses entirely on insight and voice.
The founders getting the best results had figured this out intuitively. They were not asking AI to write their content. They were asking AI to draft a structure they then filled with their actual thinking. The ones who were disappointed had asked AI to do the thinking too, and were correct that it failed.
Misconception Two: Automation Means Set It and Forget It
Several people in the room described wanting to "fully automate" their marketing: set up the tools, let them run, and check the results quarterly. When we probed what that meant in practice, they wanted email sequences to run without updates, social posts to publish without review, and newsletters to generate without a human editor in the loop.
This is not a realistic or effective use of AI marketing systems. The goal of automation is not to remove humans from the process. It is to remove humans from the low-judgment, repeatable steps so they can spend more time on the high-judgment steps. The email sequence needs quarterly tuning as your product and audience evolve. The social posts need a human review pass before they publish. The newsletter always needs a human edit.
What surprised me was how this misconception was driving real purchasing decisions. Founders were evaluating tools based on how much they could set and forget rather than how well they supported a human-in-the-loop workflow. That evaluation criteria is backwards. The best AI marketing tools are the ones that make the human review pass faster and more effective, not the ones that promise to eliminate it.
This is where I find the 3-Layer AI OS useful as a way of thinking. Context, then Skills, then Workflows. The set-it-and-forget-it fantasy skips the first two layers entirely. You cannot automate a workflow well until the AI brain underneath it actually knows your business, your voice, and your offer. Most founders try to buy Layer 3 without building Layer 1, then wonder why the output reads like everyone else's.
The founders getting the most from AI marketing are not the ones who took themselves out of the process. They are the ones who redesigned the process so their time goes only where human judgment is irreplaceable. That redesign starts with knowing what your current process actually looks like. Get a free audit to find out.
Misconception Three: More Tools Equals Better Marketing
I asked the room how many marketing tools they were actively using. The median answer was seven. The range was three to fourteen. When I asked how many of those tools were connected through automated flows rather than requiring manual handoffs between them, the median dropped to one.
Most of the room had a collection of tools, not a system. And the pattern was consistent: when one tool did not produce the results they expected, they added another tool rather than connecting the ones they had. The problem was diagnosed as a tool problem when it was actually a workflow problem.
This is one of the most expensive mistakes we see lean teams make. You do not need more tools. You need the tools you have connected in a sequence where the output of one step becomes the input of the next automatically. A three-tool stack with a well-designed flow will outperform a fourteen-tool stack where everything requires a manual handoff.
The exercise that clarified this for the room was simple. Trace one piece of content from creation to distribution and count how many times a human has to manually initiate a step. For most people in the room, the answer was between six and ten manual steps. For a team running a connected AI marketing system, the answer is two or three. That gap is where the leak is, and it is almost never a tool gap.
What Actually Surprised Me
I expected the quality skepticism. I expected some confusion about automation. What I did not expect was how few founders had thought about email nurture as a system at all.
Most of the room had a welcome email. Almost none of them had a nurture sequence beyond that. When I asked what happened after someone subscribed to their newsletter and received the welcome email, the most common answer was a version of "they get the next newsletter when it goes out."
That gap, the space between subscription and the next meaningful touchpoint, is where leads go cold. Platforms like HubSpot and Mailchimp have made behavioural automation accessible to teams without technical resources. The tools to close this gap are not expensive or complicated. The gap exists because founders have not thought about the subscriber journey beyond acquisition.
When we worked through a simple nurture architecture, a five-email sequence spaced over three weeks addressing the most common questions a new subscriber has before becoming a customer, several founders looked genuinely surprised that this was within reach. It was not complicated and it was not expensive. It was simply not something they had prioritised, because they were focused on generating new subscribers rather than converting the ones they already had. If that sounds familiar, it is the most common shape of a bandaid system I see.
The Pattern Underneath All of It
After the session, reflecting on the conversations, there is a pattern underneath all three of the misconceptions that I think is worth naming.
Most founders are thinking about AI marketing as a content production problem. How do we produce more content, faster, at lower cost? That framing leads to the tool-centric thinking, the quality anxiety, and the set-it-and-forget-it fantasy. When content production is the frame, more tools and less human involvement seem like the obvious solutions.
The founders who are getting the best results are thinking about AI marketing as a systems design problem. How do we build a set of connected flows that move our audience from awareness to trust to conversion without requiring constant manual intervention? That framing leads to the right questions: What are the steps in our conversion journey? Where are the manual bottlenecks? Which steps have low enough judgment requirements to be automated? Which need to stay human?
The systems design frame also produces better content. Designing a system forces you to articulate your conversion path, your audience's questions at each stage, and your positioning, which is exactly what makes AI assistance useful. You have something specific to feed it rather than asking it to write "a LinkedIn post about our product." That, in plain terms, is what a clean OS does for you. It gives the AI brain something real to work from.
What I Would Teach Differently Next Time
The diagnostic exercise at the start, eight questions about each founder's current marketing setup, was the most valuable fifteen minutes of the workshop. Next time, I would expand it to thirty minutes and use the results as the basis for small group conversations, letting founders with similar setups work through the same build-order questions together. Peer learning in those moments is something a presentation cannot replicate.
I would also spend more time on the email nurture gap. The AI-Assisted Newsletter Workflow and the Content Ops Calendar for Lean GTM Teams cover production and planning well, but the nurture architecture deserves its own dedicated session. It was the biggest gap in the room.
FAQ
What was the most surprising thing you learned from the Toronto Tech Week session? The near-universal absence of email nurture sequences beyond a single welcome email. The tools to build a basic five-email nurture sequence are free or nearly free, and the platforms, HubSpot and Mailchimp among them, are designed for exactly this use case. The gap was not about resources or technical complexity. It was about founders not having thought through the subscriber journey past the acquisition moment.
What misconception about AI marketing is most damaging for startups? The belief that AI can fully automate marketing without a meaningful human review layer. This misconception leads founders to build systems with insufficient quality control, which produces consistently mediocre content. The mediocrity then gets attributed to AI rather than to the design of the workflow.
Is the "more tools" problem specific to Toronto founders or is it widespread? It is widespread, a global startup behaviour rather than a Canadian one. What may be specific to Toronto is that the talent pool for growth marketers who understand both strategy and marketing technology is smaller than in larger US markets, meaning fewer people are available to diagnose workflow problems before they calcify.
What is the most important thing a founder can do this week based on these field notes? Map your subscriber journey. From the moment someone joins your list to the moment they become a customer, write down every touchpoint and who initiates it. Every step that requires a human to remember to do something is a gap. Every gap between touchpoints that is longer than a week in the first thirty days is a nurture problem. That map tells you exactly what to build first.
Will there be another Toronto Tech Week session or similar workshop? Yes. We are planning a follow-up session focused specifically on the email nurture gap and on building connected AI marketing flows from scratch. Details will be announced in the Digiwell newsletter first, so sign up if you want early access.
Read Next
- AI-Assisted Newsletter Workflow
- Content Ops Calendar for Lean GTM Teams
- The AI Marketing Stack We Build for Early-Stage SaaS Teams
Want Help Applying This?
If the gaps described in these field notes sound familiar, tools but no system, a welcome email but no nurture sequence, content production but no connected distribution flow, a free audit gives you a specific picture of what to fix and in what order.
So here is the question I left the room with, and I will leave it with you too. If you traced one subscriber through your own program today, from the moment they joined to the moment they bought, how many of those steps are actually a system, and how many are just you remembering to do them?