There is a recurring moment in the life of a growing business when the organization begins to feel heavier than it looks on paper. Revenue may still be rising. Headcount may still be modest. Yet decisions take longer, handoffs multiply, and simple questions—Where do things stand? What just changed? Who needs to act?—require more coordination than they once did. In recent years, tools like n8n and a widening family of AI-enabled workflows have become the answer to this weight. Not as a silver bullet, but as connective tissue: a way to move information between systems, interpret it, and trigger responses with less human mediation. Having been around long enough to see a few technological sea changes, I tend to be skeptical of anything sounding magical or transformational. The real work getting a buisiness to scale lies in product-market fit and the day to day execution that delivers for customers. So I always look through the lens of whether or not something is worth it. There was a time that every company had to have an iPhone app and every company had to be on a slew of social media platforms, but few made real money in any of those places. It is worth stating that these tools are transformational, but are they worth the time and effort when the setup can be significant and the returns are unkown. This is worth putting on "paper." These peoducts and the promise they provde is considerable and exciting. For many founders and operators, the friction is where they have issues: learning curves, brittle automations, and a feeling that people at the company are spending more time adapting a technology than they save by moving to the more "efficient" system. Rather than declare a conclusion, I want to think through this question the way good operators historically have: by observing what is happening, recognizing patterns, forming a working hypothesis, and then iterating on the approach while asking if the hypothesis is directionally correct.

The first observation is simple and familiar. As businesses grow, the strain on management collects in several areas, but chief among them is financial control, strategy and data collection and presentation. Companies that raise money should be able to clearly tell their investors what is happening every month. Companies that hire more salespeople need to accurately track sales bonuses, and those that are adding employees in general need ways to onboard them and provide basic employer functions to them. Data lives in systems that were adopted at different times for different reasons. Context lives in inboxes, chat threads, and the heads of long-tenured employees. Exceptions—the things that actually matter—are buried inside routine activity. I have been in closing processes where the CEOs are emailing past employees for signed documents that may or may not be on those employees personal laptops. None of this work is conceptually hard. It is, however, cognitively expensive and operationally fragile. When one person is out, things stall. When volume spikes, errors increase. When the business changes shape—as it always does—processes lag behind reality. This is the environment in which workflow tools and AI-assisted automation are being the most exciting — not because leaders want novelty, but because the old mechanisms stop working.

Taking a step back, we should also note that this situation has appeared before, many times, under different technological labels. and structures. Long before “automation” was fashionable, businesses struggled with the same structural issue: too many handoffs, too much implicit knowledge, and too much reliance on memory. Accounting ledgers, standardized forms, spreadsheets, and later enterprise systems were all successful attempts to reduce the cost of coordinating and analyzing company data. Each wave followed a similar trajectory. Early adopters endured awkward tooling and unclear payoffs. Late adopters dismissed the tools as distractions until the surrounding ecosystem made the old way untenable. Over time, what once felt optional became assumed infrastructure. But for every Excel adopter, there was a Lotus Notes adopter that got stuck with the non-standard. AI-enabled workflows can sit squarely in this lineage. For us, they increasingly appear to be the next spreadsheet not the next social app everyone forgot about. They can powerfully structure company data, reducing strain on managers. They encode the “when this happens, then that should follow” logic that otherwise lives informally inside organizations. The pattern that matters is not technological sophistication, but substitution. Whenever a business replaces human routing—deciding who should know what, and when—with a repeatable mechanism, it trades flexibility for reliability. Historically, that trade has favored reliability once a business reaches a certain scale.

Under this hypothesis, the return on effort does not come from eliminating roles or accelerating everything indiscriminately. It comes from fewer dropped signals, fewer delayed reactions, and fewer hours spent reconstructing what already happened. This framing helps explain why the setup often feels disproportionate. The work is front-loaded because the organization must articulate its assumptions. What counts as an exception? What information actually matters? What should happen automatically, and what should escalate to judgment? Answering those questions is slow because it exposes ambiguity. But that ambiguity existed already. The workflow merely makes it visible. The hypothesis also explains why results can feel incremental rather than spectacular. A well-designed workflow rarely announces itself. It quietly prevents a missed follow-up, a stale report, or an unnoticed change. Over time, those prevented failures accumulate into something that feels like operating leverage.