Now accepting first-cohort clients · 2026

I build AI automation
that removes the drag
from operations teams.

Independent automation specialist building production-grade AI workflows for founders and operations leaders. Five fully-built reference systems below — each one a real, working pipeline you can watch in action.

Book a consultation → See recent work
Reference builds
5 fully operational
Build cadence
5–7 days each
System response
< 30 sec
Coverage
24/7 autonomous
Studio · 2025–2026

Reference builds

Five complete automation systems I've built to demonstrate what's possible — each one is an end-to-end pipeline you can watch run. They're the templates I'll start from when we build the one that solves your bottleneck.

● Case 01 / Lead Operations

Lead qualification & routing for digital marketing agencies

Sales teams burn 3–5 hours a week researching inbound leads. By the time anyone replies, the warmest prospects have already moved on to a competitor. This system replaces the entire research step — enrichment, scoring, and routing happen before the lead lands in front of a human.

Format
Reference build · 6 days
Toolkit
n8n GPT-4o Apollo Airtable Slack
−100%
Manual lead research time
< 30s
Hot-lead Slack alert latency
1–10
Every lead scored against ICP
Watch on Loom →
● Case 02 / Voice AI

24/7 voice agent for restaurant reservations

Restaurants miss bookings whenever staff are busy at the host stand. Every unanswered call is a lost reservation — and with group bookings on the rise, lost revenue compounds nightly. The agent answers every call, in natural language, around the clock.

Format
Reference build · 7 days
Toolkit
VAPI n8n GPT-4.1-mini Airtable
24/7
Reservation coverage, including after hours
0
Missed bookings from unanswered calls
NL
Handles natural date phrasing
Watch on Loom →
● Case 03 / Document Processing

AI invoice & receipt extraction pipeline

Finance teams manually re-type invoice data from scanned PDFs and email attachments. Records fall behind, transcription errors slip through, and reconciliation eats whole afternoons. The pipeline turns any uploaded document into a structured Airtable record in under ten seconds.

Format
Reference build · 5 days
Toolkit
n8n GPT-4o Vision Airtable Slack
~10s
From upload to structured record
100%
Line-item extraction including totals & tax
0
Manual data entry required
Watch on Loom →
● Case 04 / Customer Support

AI inbox triage & auto-responder

Support inboxes become walls of noise — urgent tickets buried under newsletter replies and pricing questions, response times slipping, the team constantly fire-fighting. This system triages, prioritises, and drafts a reply for every inbound message before it ever reaches a human.

Format
Reference build · 5 days
Toolkit
n8n GPT-4o Airtable Slack
3-axis
Classification: category · urgency · sentiment
Auto
Draft reply generated for every ticket
100%
Tickets logged with unique ID & metadata
Watch on Loom →
● Case 05 / Knowledge Systems

RAG-powered support bot grounded in internal docs

Customers wait hours for answers to questions the documentation has already covered. Knowledge lives in PDFs nobody opens, and the same questions reach the team queue every single day. This bot grounds answers in your actual docs — and escalates honestly when it doesn't know.

Format
Reference build · 7 days
Toolkit
n8n GPT-4o Pinecone Airtable Slack
Grounded
Answers cite your actual documentation
Auto
Escalates to human when confidence drops
Logged
Every interaction stored for evaluation
Watch on Loom →
"

Good automation should be invisible — a process that used to take an afternoon now just happens, and nobody on the team thinks about it again.

— Working principle
About

A bit about me.

The shortest version: I build automation systems for small operations teams, work alone by choice, and treat every engagement like the system will run unsupervised for years — because it will.

I'm Normidah — an independent automation engineer based in the Philippines. I started building with n8n because I kept watching small teams burn entire afternoons on work that should run itself. The gap between what modern AI can do and what most businesses actually use is enormous — and most of it isn't a technology problem, it's a workflow problem.

I work solo, by design. That means I'm the person on your discovery call, the one designing the system, and the one debugging it at 1am if it ever needs that. There's no junior pattern-matching your bottleneck to last quarter's template — every workflow gets built from scratch around the specific problem it's solving.

If we work together, you get fixed pricing, daily progress updates, and a system you fully own at the end. No retainer lock-in, no vendor hostage situations, no surprise invoices. The deliverable is a workflow your team can run, modify, and trust — long after the engagement closes.

How I work

A short, opinionated
engagement model.

The most reliable automation projects start narrow. Pick the single highest-leverage bottleneck, ship it in a week, then expand from a working foundation. The engagement model below is built around that idea — designed for low risk, fast feedback, and real production results.

01 / Discovery
A focused 30-minute call.

We map your operations, find the bottleneck that's costing the most time, and define what "working" looks like before any code is written.

↳ Free · No commitment30 min
02 / Proposal
A clear scope within 24 hours.

Fixed price. Defined deliverables. Exact integrations and credentials list. You'll know precisely what you're getting before you say yes.

↳ Fixed price< 24 hrs
03 / Build
Production system in 5–7 days.

Built and tested against your real data — not a sample dataset. Error handling on every external call, monitoring on every step, alerting on every failure path.

↳ Self-hosted n8n5–7 days
04 / Handoff
A Loom, docs, and the keys.

Video walkthrough of the architecture, written runbook for your team, and every credential properly stored. You own the system end to end.

↳ Fully documented1 day
05 / Support
30 days of included care.

Bug fixes, tuning, and adjustments based on real-world usage are covered for the first month. Production-grade reliability is the deliverable, not the launch.

↳ Included30 days
06 / Retainer
Or build out the next system.

Once one bottleneck is solved, the next one becomes obvious. Retainer engagements are how most clients move from one workflow to a connected stack.

↳ OptionalMonthly
Toolkit

A deliberate,
opinionated stack.

I use the same core tools across every engagement. Familiarity compounds — fewer surprises in production, faster delivery, easier handoff to your team. No experimental dependencies that disappear in twelve months.

Orchestration
  • n8n (self-hosted)
  • Claude Code
AI Models
  • GPT-4o
  • Claude Sonnet
  • Whisper
Voice & Telephony
  • VAPI
  • Twilio
Vector & Retrieval
  • Pinecone
  • text-embedding-3-small
CRM & Data
  • Airtable
  • Google Sheets
  • HubSpot
Enrichment
  • Apollo.io
  • Clearbit
Notifications
  • Slack
  • Gmail (OAuth2)
Deployment
  • Self-hosted n8n
  • Docker · VPS
Common questions

What clients
actually ask.

Honest answers to the questions that come up on most discovery calls — pricing, ownership, what happens when things break, and what I won't take on.

01
I don't see testimonials — what's your background?
I built out a portfolio of complete, working reference systems before taking on paid clients — the five above. That was deliberate: I'd rather you watch the actual work than read a wall of logos. First-cohort engagements open in 2026, and early clients get fixed pricing, extra attention, and direct access throughout the build.
02
How much does a typical project cost?
Fixed-price builds start at $1,200 USD and scale with complexity. Most engagements fall between $1,500–$4,000 for a single workflow including discovery, build, documentation, and 30 days of support. First-cohort clients get founder pricing — lower than the rates I'll list publicly once there are testimonials in this section.
03
Who owns the workflow once it's built?
You do — completely. Workflows run on your n8n instance, against your credentials, in your accounts. I provide a full runbook and a Loom walkthrough at handoff so your team can run and modify the system without me.
04
What happens when something breaks at 2am?
Every workflow ships with three layers of resilience: continue-on-fail on every external API call, structured error logging to Airtable, and Slack alerts to a dedicated channel. Most failures are visible within 30 seconds and recoverable without my involvement.
05
What kind of projects do you turn down?
Anything that needs a frontend, anything that requires real-time machine learning training, and anything where "AI" is the goal rather than the solution. If a deterministic workflow would do the job better, I'll tell you on the call.
06
How does your AI not hallucinate?
Every AI call returns strict JSON validated by a Code node downstream. Parse failures route to a manual-review queue with sane defaults. For RAG workflows, answers must cite retrieved context — when context is missing, the system escalates rather than guesses.
// Now accepting first-cohort clients · 2026

Let's build the
system you've been
meaning to build.

Thirty-minute discovery call, no commitment. We'll find the bottleneck worth automating and you'll leave with a clear sense of scope, timeline, and cost — whether or not we end up working together.

Book a discovery call → WhatsApp Email
Usually responds within a few hours · Manila time