Nicolas Dominici

July 16, 2026

I said don’t build the app by default.

That doesn’t mean “never build.”

It means: don’t build to feel like you’re making progress.

Build when the market tool blocks the business — not your builder ego.

The symmetric mistake

Two traps:

  1. Custom for sport — vibe-coding, demos, reinventing CRM/notes/calendar
  2. SaaS out of fear — forcing a weird process into a tool that fights you every day

Both are expensive.

The first in maintenance.

The second in friction and shadow work (eternal spreadsheets, parallel WhatsApp, “the truth is in Maria’s head”).

Signals that custom makes sense

Build (or integrate seriously) when several of these are true:

  1. Differentiated process — how you sell or operate is part of the advantage, not a generic flow
  2. Connected proprietary data — value is in the loop across systems, not an isolated module
  3. Rules no SaaS expresses well — ownership, states, exceptions, permissions
  4. Manual work cost > maintenance cost — with a real technical owner
  5. Channel where the team already works — the system meets WA/email/CRM; it doesn’t force another UI

If you only have #1 as “we’re special” without evidence, pause.

What to build then

Not “an app.”

A business system:

  • reliable intake and follow-up
  • source of truth
  • visible next actions
  • boring integrations
  • automation and AI after the process

That can be:

  • glue code + n8n + CRM
  • a small module on top of existing tools
  • agents with files and permissions
  • custom software where it actually hurts

Form follows the loop.

Not the other way around.

What doesn’t justify custom

  • “We want something ours” with no metric
  • A dashboard nobody uses to decide
  • Fashion features (chatbot, agents) with no exception owner
  • Replacing Notion/Calendar because it “looks more pro”

Then go back to Don’t build the app by default.

How I decide with a client

  1. Draw the loop on one page (enter → decide → next action → measure)
  2. Try the 80% with existing tools if the loop is standard
  3. Custom only at the bottleneck the market doesn’t cover
  4. Name who maintains it

If step 4 is empty, there is no project.

There’s an expensive hobby.

Relationship to Zintekia (no price pitch)

When I help companies, the deliverable isn’t “automation.”

It’s reliable behavior: fewer leaks in sales and ops.

Sometimes that’s configuration.

Sometimes it’s a custom system.

Judgment is what I’m buying too when I choose a stack.

I help decide and build the minimum layer the business needs — custom only when it truly hurts. If you’re stuck between SaaS, glue, or custom software, contact me.

You can also email me at nicolasdominici@outlook.com or DM me on LinkedIn.