I Ran 7 Powers on My Own Startup

I listen to Acquired whenever I walk the dog. And we have a very energetic dog, so I’ve become something of a superfan. Every teardown ends the same way: Ben and David grade the company against Hamilton Helmer’s 7 Powers, the seven sources of durable advantage from his book 7 Powers: The Foundations of Business Strategy. It’s a genuinely useful lens, and after enough episodes of listening to them apply it to Nvidia or Costco, you start sizing up the companies you know best.

So I decided to apply it to this+that.

My first observation is that AI has changed the source of power for tech companies. Technology itself is no longer a moat. For most of my career, it was. You had some domain knowledge, or a deeply technical insight, and built something unique. The difficulty itself held competitors at bay for a year or more. That world is gone. In the AI era, a capable team can build almost anything in about a week. Our team built the Brain, the knowledge layer that powers this+that, in even less time. If they can do that, so can an equally talented team across the street.

This raises the only question that matters. When building is free, where does durability actually come from?

The four we fail

Helmer’s seven powers are scale economies, network effects, counter-positioning, switching costs, branding, cornered resource, and process power. Four of them aren’t available to us, so I’ll start there.

Scale economies. This power shows up when your cost per unit drops as you get bigger, the way it does for a chip foundry. Software used to qualify because the marginal cost of another user rounded to zero. AI broke that. Every action our assistant takes runs an inference call, and inference is a variable cost that scales with usage, not against it. A bigger user base doesn’t make the next task cheaper to serve. We negotiate better model pricing as we grow, sure, but that’s a discount, not a moat. Grade: no power.

Process power. This is the Toyota Production System: an advantage built into how the organization works that rivals can’t copy even when you show them the factory. It takes years and an entire culture. We’re a startup that rewrites half its assumptions every quarter. It’s hard to see how an early-stage software company can have process power before it reaches strong product-market fit. Grade: too early to be real.

Cornered resource. An airtight patent, an exclusive supply deal, a deposit nobody else can mine. We don’t have one, and what startup does in 2026? There’s no input we control that a competitor can’t also reach. Grade: no power.

Network effects. The news here is only mostly disappointing. There’s a real effect when a whole team runs on this+that: shared workflows, a common knowledge layer, the value climbing as more of your colleagues join. But it’s bounded inside a team. It’s not the global, winner-take-all flywheel that a marketplace or a social network gets, where every new user anywhere makes the product better for everyone. A competitor can win the team next door without ever touching ours. That might change if landing one team starts pulling in the rest of the company, each team bringing in the next, until a team-sized effect becomes a company-sized one. We’re not there yet. Grade: weak, and local.

Four of the seven, gone or thin. That leaves three that actually hold up.

The three that are left

Three powers survive the audit: counter-positioning, switching costs, and branding. What they share is the reason they matter now. None of them can be handed to a competitor by an afternoon of coding. They come from position and from accumulated state, not from cleverness, and in an era where cleverness is a commodity that’s the only kind of advantage worth building on.

Counter-positioning, like Switzerland

Counter-positioning is Helmer’s subtlest power. You hold a position that an incumbent can’t copy, not because they lack the engineers, but because copying it would damage the profitable business they already run. The incumbent looks at your move, reasons about it correctly, and decides to pass. Conscious refusal is the moat.

Ours is neutrality. this+that feeds its Brain from communications across every ecosystem at once: Gmail, Outlook, Slack, Teams, Google Chat, Telegram. We treat every channel as a first-class citizen of your inbox, not a second-class import.

Now look at that through Microsoft’s eyes, or Google’s, or Notion’s. I mean this generously, because their position is rational. Those companies make their money by pulling you deeper into their ecosystem. Microsoft’s grounding story is M365. Google’s is Workspace. Notion wants to be the place your knowledge lives. Ingesting a competitor’s channels as equal citizens cuts directly against the lock-in that funds the whole operation. They could build it in a week, same as us. They won’t, because the copy would weaken the business they’re actually defending. That’s counter-positioning in its textbook form: the incumbent reasons correctly and declines.

And the payoff of neutrality compounds in a way I didn’t appreciate until I watched it happen. Every company has a knowledge base, and every knowledge base rots, because a human has to stop working and go stock it. A brain fed by your communications doesn’t have that problem. It fills itself from the work. The neutrality is what makes the feed complete, and a complete feed is what makes the brain worth trusting.

The same trap runs one layer down, in where your knowledge lives. Microsoft doesn’t just want your mail and chat in Outlook and Teams, it wants your documents in SharePoint. Google wants them in Drive. But plenty of teams don’t run a single vendor’s whole stack. They pair Gmail with Notion, or Outlook with Confluence, taking the comms they like from one place and the knowledge tool they like from another. We meet that reality instead of fighting it. Through MCP, the open protocol, this+that writes straight into the knowledge tools you already use, updating a Confluence page, a Google Doc, or a GitHub repo as the work happens. Your knowledge stays where you keep it, and we keep it current there. That’s the same neutrality applied to the knowledge layer, and the suites can’t copy it, because their whole pitch is that you should consolidate onto them.

Switching costs: the brain and workflows

The second live power follows from the first. A brain that fills itself from your communications gets better the longer your team uses it. It learns your accounts, your projects, your shorthand, the standing decisions nobody writes down. After a year, it holds a model of how your organization actually runs, and that model is proprietary. You can’t easily export it to a competitor, just like you can’t easily export your contacts from Salesforce. You can do it, but it’s time-consuming, and you have to update the processes and tools around the data at the same time. So, why would you?

That’s a switching cost in the cleanest sense. Leaving doesn’t mean re-importing your data. It means giving up accumulated understanding that took a year of real work to build and starts from zero somewhere else. The longer the brain runs, the more expensive the exit, and none of that depends on us being smarter than the next team. It depends on time and use, which a competitor can’t fast-forward.

The brain is the deep version of this, but not the only one. A team that has been with us a while has also built up a stack of workflows wired into its tools and its habits: the routing rules, the approval chains, the automations that quietly run the week. That’s a more ordinary switching cost, the kind any automation platform earns, but it’s real, and it sits on top of the brain rather than competing with it. The knowledge is one layer, the machinery that runs on it is another, and leaving means rebuilding both.

This is also where we’re aiming next. If neutrality is what makes the feed complete and the brain is what makes the feed compound, then the work in front of us is the part in between: a comms-neutral process for keeping the brain current as the work flows in, across ecosystems rather than locked to any one of them.

Branding

The third power is the quietest and, in this category, maybe the most underrated. Branding is durable when a name carries trust that a competitor can’t manufacture on demand. You’re handing an assistant your inbox, your calendar, and your company’s working memory. That’s an enormous amount of trust to ask for, and trust doesn’t ship in a sprint. It accrues from years of conservative claims, honest limitations, and not betraying the access you were given. A competitor can copy our features by Friday. They can’t copy a track record of having been careful with people’s data, because a track record is something you accumulate, not something you ship.

The real race

Counter-positioning is temporary. It holds only while the market is small enough that incumbents would rather protect their existing business than chase ours. When the prize gets large enough, they bite the bullet, eat the damage to their walled garden, and build the neutral thing anyway. Every durable counter-position eventually faces that moment. I don’t get to assume ours is special.

So the real race isn’t a feature race, because code isn’t a moat anymore. The race is to get each customer’s knowledge into a brain, whether that brain is ours or one we reach over MCP, and to build the automations that keep it current and run on top of it, before that window closes. Switching costs and brand are the only powers that survive the incumbents recognizing the opportunity. There’s no shortcut for building trust and accumulating knowledge.

That’s the actual strategy. Not a clever architecture, because the architecture is a week of work for anyone. The position, and the head start on filling it.

If you want to watch a brain start filling itself from your team’s work, you can try this+that. Free during beta.