Independent dev: how to work solo without becoming an agency

The myth of agency as natural evolution

There’s a narrative in the dev market that goes something like this: you start as a freelancer, gain experience, hire people, become an agency, scale. As if working solo were a temporary stage until you “really grow up.”

I disagree. After 8 years working in web development, I went in the opposite direction. I left a full-time service relationship with an agency to work as an independent dev, with my name as the brand. No team, no office, no ambition to become an agency.

This wasn’t for lack of options. It was strategy. Agencies bring overhead that doesn’t make sense for everyone: people management, payroll, rent, sales team, finance. For those who want quality and control over their work, keeping operations lean is a choice, not a limitation.

The model that works for me

I work with sites, landing pages, and webapps. My current focus is digital presence for aesthetic dentistry clinics. The model is simple: single-payment packages between R$1,500 and R$3,000, hosting handled by the client (which in most cases is free on Vercel).

No monthly fees. No mandatory maintenance contracts. The client pays once, receives the site, and that’s it. If they need adjustments later, I charge on demand.

This structure works because:

• The ticket is high enough that I don’t need absurd volume
• The scope is well-defined (site/LP with 2-4 week delivery)
• Operational costs are nearly zero (computer, internet, free Vercel)
• I have no employees, so I have no payroll

With 2-3 projects per month, the math works. With 4-5, it works well. Without needing to manage anyone but myself.

A niche solves the acquisition problem

The biggest fear of working solo is “where do clients come from?” When you serve everyone, you compete with everyone. When you pick a niche, acquisition changes.

I chose aesthetic dentistry for practical reasons. The service ticket is high (dental veneers cost R$10-15k per case), the professional understands investment, and the market is large enough to sustain demand.

My prospecting works like this: Instagram DM and WhatsApp. I send direct messages to dentists who post aesthetic cases. It’s not traditional cold calling. It’s a personalized approach, referencing the professional’s work, showing how a site can generate real returns.

I have a prospecting landing page that works as supporting material in these conversations. It’s not an automatic closer. It’s a credibility piece the dentist can review before deciding.

Process compensates for lack of team

Agencies have teams to split tasks. Independent devs have process. If the process is good, you deliver on the same timeline and with higher quality, because one person with full project context makes better decisions than three people who only saw a part of it.

My workflow operates in layers:

• I define the scope and site structure in an .md file with detailed specifications
• I build with Astro (SSG, zero client-side JavaScript when possible)
• Automatic deployment on Vercel via Git
• Lighthouse 95+ as a minimum target, not a bonus

Every project follows the same base. Same stack. Same deploy process. What changes is the content, palette, typography, and copy strategy. This means I’m not reinventing the wheel on each project. I’m applying a system that already works.

With GPM2, Soline, and Tok Final, they all follow the same structure. Astro, Vercel, Lighthouse above 95. The client receives a fast, well-ranked site that doesn’t depend on plugins to function.

Tools that replace a team

Not having a team doesn’t mean doing everything manually. It means using tools that handle coordination, review, and execution without needing another person.

What I use daily:

• Claude and Claude Code to generate specs, review copy, audit SEO, and accelerate code
• Astro as the framework (SSG, maximum performance, no unnecessary frontend framework)
• Vercel for deployment (automatic, free, global CDN)
• Git for version control (no surprises, easy rollback)
• NotebookLM to organize sales knowledge (I have 15 books synthesized there)

Each of these tools eliminates a role that in an agency would be a person. Copy review? AI. Deployment? Automatic. Knowledge management? NotebookLM. This doesn’t replace competence, but it multiplies capacity.

The limits of working solo

It would be irresponsible not to mention the limits. There are things an independent dev can’t do:

• Projects with 1-week deadlines and large scope (no team to split work)
• Clients who need 24/7 support (I’m not an agency with a support desk)
• Projects requiring complex original design (I work with functional design, not branding from scratch)
• Volume above 5-6 simultaneous projects (quality drops)

Knowing when to say “no” is part of the model. If a project doesn’t fit my process, I decline. This feels counterintuitive when you’re starting out, but it’s what protects quality and reputation long-term. Every time I turned down a project that was too big or too rushed, I had capacity to take something better the following week. The projects I regret are not the ones I declined.

There’s also the isolation factor. Working solo means not having someone to bounce technical decisions off day-to-day. No code reviews from a colleague, no quick hallway conversation about whether to use a context provider or prop drilling. I handle this with AI tools for a second technical opinion and networking in online dev communities. Claude Code, for example, catches issues I would miss after staring at the same component for hours. Dev communities on Discord and Twitter fill the gap for broader architectural discussions. It doesn’t replace a teammate, but it works in practice.

Building a tech business without startup cosplay

The tech market is fixated on scale. “How many employees do you have?” “What’s your MRR?” “When are you raising funding?” These questions make sense for SaaS. They don’t make sense for services.

Service scales differently. It scales through pricing, niche, and efficiency. If I can deliver in 2 weeks what an agency delivers in 6, my margin is better even charging less. If I focus on a niche where the client ticket is high, I don’t need 30 projects per month.

The goal isn’t to build an agency. It’s to build work that pays well, allows autonomy, and doesn’t depend on heavy structure to function.

• Do you have a standardized stack you apply to every project?
• Is your operational cost low enough to live on 2-3 projects/month?
• Can you say “no” to projects that don’t fit your process?
• Do you have a defined niche or do you serve whoever shows up?
• Do you use AI tools to multiply your capacity?
• Does your pricing reflect the value you deliver, not the time you spend?

An independent dev isn’t a freelancer who didn’t grow. It’s a professional who chose a different operating model.