Build-in-Public — Brian Benwah

// context

What is this?

I'm building brianbenwah.com using the exact AI employee system I teach — n8n workflows, Claude doing brand research and content generation, AWS infrastructure underneath.

I'm not going to show you a finished product and tell you it was easy. I'm going to show you the build as it happens: what I planned, what I built, what broke, what it costs, and what I'd do differently.

This is the proof. Not testimonials. Not income screenshots I can't verify. The actual build, in public, with honest numbers.

// status · updated as things change

Current numbers.

The burn is real. The $0 revenue is real. The goal is $2,000/month minimum within 12 months — and I'm tracking it here, in public, whether it's going well or not.

That's the point.

// brianbenwah.com · AI employee build running

Status running

Started 2026

Revenue $0

Monthly burn ~$800–$1,000

12-mo goal $2,000/mo minimum

Stack Claude · n8n · AWS · Postgres · Docker

// what's running

The tools. What they do. What they cost.

// current stack · costs tracked in build log

Claude (Anthropic)
  Role: AI brain — content generation, brand research,
        voice matching, copy drafting
  Cost: tracked per-run in build log

n8n (self-hosted)
  Role: workflow automation — the connective tissue
        between every tool in the system
  Cost: server cost tracked in build log

AWS
  Role: infrastructure — hosting, storage, compute
  Cost: tracked monthly in build log

Postgres
  Role: database — storing brand context, content history,
        workflow state
  Cost: included in AWS cost

Docker
  Role: containerization — keeps everything portable
        and reproducible
  Cost: $0 (open source)

I'll update the cost column as real numbers come in. No estimates, no "approximately" — the actual invoices.

// log · reverse chronological

The build, entry by entry.

Each entry covers what I planned, what I built, what broke, and what it cost. The log starts from the beginning — including the parts that didn't work.

// entry 001 · 2026-06 running

Starting from zero.

// what's built

Brand profile, voice profile, positioning strategy — all generated and refined using Claude
n8n workflows for content research and drafting — running, producing output
AWS infrastructure — live, costs being tracked
This website — built using the design system and copy framework the AI employee produced

// what's not built yet

A content distribution system (in progress)
An email capture and nurture sequence (planned)
A community or offer (model undecided — tracking options)

// what broke

Nothing catastrophic yet. The first n8n workflow needed three iterations before the Claude output matched the voice profile. That's normal — the tuning phase is real work.

// what I'd do differently

I'd write the voice profile before the brand profile. The voice shapes everything downstream, and I built it second.

// cost this period

// entry 001 costs

AWS tracking (invoice pending)

n8n tracking (server cost)

Claude tracking (API usage)

Total (all-in) ~$800–$1,000/mo burn

// next

Content distribution system. Getting the output from the AI employee into the actual channels — YouTube, Instagram, the blog. That's the next build.

// roadmap

What I'm building next.

[ ]

Content distribution system

n8n → YouTube / Instagram / blog pipeline

[ ]

Email capture + nurture sequence

Plain, no-manipulation, here's-what-I-know format

[ ]

Community or offer

Model undecided; tracking options. Will be announced here first, with honest numbers.

[ ]

First real cost breakdown

Full invoice-level breakdown of what it costs to run this AI employee for one month

I'll update this list as things get built. Crossed off means done. Amber means over-budget or delayed. Green means running.

// the reason

I'm not doing this for the content.

I'm doing this because it's the only honest proof I have right now.

I don't have client results. I don't have a track record of people who've used this system and made money. I'm pre-launch.

What I have is 20+ years of building systems, a global internal AI app I helped build at AWS, and this brand — which is running on the exact system I teach.

The build log is the proof. If the system works, you'll see it in the numbers. If it doesn't, you'll see that too.

That's the deal.

// stay in the loop

I post the builds as they happen.

The build log lives here. The shorter updates — what I'm working on, what just broke, what I figured out — go on YouTube and Instagram. Follow wherever you spend time. No platform gets the "real" version.

YouTube

// youtube.com/@brianbenwah

Full build walkthroughs — the complete plan → build → run sequence.

Instagram

// @brianbenwah

Behind-the-scenes, shorter updates, what I'm working on right now.

Twitter / X