For the past decade plus, starting a company has felt like choosing between two paths:
Agency or software.
High-touch or large-scale.
White-glove or self-serve.
Custom work or standardized product.
Agency businesses are nice because they feel premium. They’re built around judgment, context, and trust. Clients generally talk to real people. Work is tailored. But the downside is that scale is hard; headcount becomes a ‘ceiling’.
Software businesses are nice because they scale. Dashboards, APIs, subscriptions, etc. One product can serve thousands of customers. But software often assumes users already know what they want (and how to use it). For many customers, that’s a bad assumption.
For a long time, this tradeoff was unavoidable. You couldn’t be high-touch and high-scale at the same time:
I don’t think it is anymore.
AI and agents are enabling a third model, something between agency and software. I will call this the modern agent-cy. I suspect the most successful businesses of the next 10 years will be both Agency AND Software.
How will this look?
From the user’s perspective, an agent-cy feels like a premium service. You explain what you’re trying to do. You’re asked smart follow-up questions. Then you get a clear outcome. The experience feels guided, not self-serve.
Behind the scenes, it’s software - workflows, automation, data pipelines, models. The user never sees that complexity (they dont need to…) What matters is the outcome.
I’ve been thinking about this a lot while building ytsponsordb.com. On paper at this early stage it could be either. Could be agency: custom audits, competitive analysis, decks, consult calls. It could also be pure software: log in, explore a database, pay monthly.
Of course the ‘ten-star’ best version would sacrifice at neither end of the spectrum.
The goal is something that feels high-touch but scales to hundreds of customers - ie. premium on the surface and automated underneath.
This is the differentiator that agents will become. AI of the next decade will live not as chatbots bolted onto dashboards, but as orchestration layers. They will sit between the user and the software, translating intent into action. Instead of forcing users to learn tools the ideal system will learn and adapt to them.
Early examples already exist. Products like Boardy.ai feel more like a concierge than a platform (you share context via an agentic phone call, answer a few questions, and receive curated results via chat/text). The interface is secondary. The outcome is what matters.
Importantly, this shift enables emergent business models:
> You can price like a service but operate like software.
> You can deliver confidence without linear headcount growth.
> You can scale expertise without turning it into a generic dashboard.
It also changes how products compete. Traditional software competes on features. Agent-cies compete more on trust. Switching costs go up when a system understands intent and delivers outcomes because the product is more helpful (not necessarily harder to cancel/leave).
This matters today because people are fatigued of the SaaS model. Dashboards are crowded. Self-serve experiences leave users behind. Agencies are valuable yes, but they hit scaling ceilings fast.
The next generation of companies won’t look like traditional agencies or traditional software. They’ll live in the middle.
High-touch experiences powered by software.
Guided, not self-serve.
Outcome-driven, not interface-driven.
Peace,
Ramsey