n8n & AI workflow automation

Give your team back ~20 hours a week.

I'm Dyaksa. I find the copy-paste, chasing and checking that eats your team's week, and replace it with automations that run on your own server, never guess when they're unsure, and cost nothing extra as your volume grows.

Book a free process review See the results ↓

~1,000 h

staff hours a year handed back, across five workflows

estimate

≈ A$43k

in staff time a year, at A$42/hour

estimate

60%

less wasted ad spend with rule-based automation

real result

< 5 weeks

for an invoice workflow to pay for itself

benchmark

What changes for your business

Every workflow I build is judged on four things a business owner actually cares about.

Time back

Hours of copying, checking and chasing handed to software, so your people spend their day on work that needs a person.

Lower cost

Each workflow pays for itself in weeks. It's self-hosted, so there are no per-task fees that grow with your volume.

Fewer errors

Inputs are validated, duplicates blocked and every result checked. Anything uncertain goes to a person instead of being guessed.

Your data, your control

Customer, financial and ad data stays on your own infrastructure, with a log of every decision the automation makes.

Results, before and after

Five workflows from my portfolio, each compared with the manual process it replaces. Every figure is labelled as an estimate, a benchmark or a real result.

Lead Capture & CRM Sync

Sales

Every lead answered in under a minute, not hours later.

Before

Leads from several forms copied into the CRM by hand. Replies go out hours later, and the same person gets entered twice.

After

Every lead is captured, de-duplicated and answered automatically, and sales are pinged the moment it arrives.

Time saved

~225 hours a year

estimate

Money saved

≈ A$9,500 a year

estimate

Errors

No duplicate contacts; bad emails stopped at the door

Lead data goes straight from your form to your CRM, never through a third-party automation service.

Tested: new, returning, invalid and outage scenariosHow it works →

Invoice Processing & Approval

Finance

5.2× faster per invoice, and it pays for itself in under five weeks.

Before

About 90 minutes a day opening PDFs, typing six fields, checking totals and routing approvals from memory.

After

Invoices are read, checked and routed automatically. Your team sees only the ~5% that genuinely need judgement.

Time saved

7.4 hours a week

benchmark

Money saved

≈ A$14,300 a year

benchmark

Errors

Error rate 1.56% → 0.63%

benchmark

Invoices are processed on your own server, with no upload to an outside OCR service. Large amounts always go to a manager, and every decision is logged.

Tested with real PDFs: over limit, under limit, unreadableHow it works →

Ad Spend Optimisation

Marketing

Budget moves only when a result is real, not noise.

Before

Daily dashboard checks and gut-feel calls that pour budget into ads that only look like winners.

After

Spend and revenue are checked every 6 hours. Proven winners are scaled in safe steps, losers paused, and replacements queued.

Money saved

60% less wasted ad spend

real result

Time saved

~160 hours a year of manual review

estimate

Errors

No budget changes on lucky streaks

Revenue and spend data stay in your systems. Budget increases are capped, and every change is logged with its reason.

The 60% is from my own paid-media work at EcomobiHow it works →

Ad Creative Production (AI + video)

Content

A creative concept goes from brief to every ad size overnight.

Before

An afternoon per concept exporting image and video variants one by one, and broken files that sometimes go live and waste spend.

After

One spreadsheet row produces every image and video variant, and each file is checked before it can reach your ad account.

Time saved

~200 hours a year

estimate

Money saved

≈ A$8,700 a year

estimate

Errors

0 broken files reach live ads

Your prompts and creative strategy stay on your own machines, and hidden prompt data is stripped from every image.

Tested: generation, video renders, stuck-queue alertHow it works →

Server Monitoring & Nightly Reporting Data

Operations

Problems fixed before customers notice, and backups you can trust.

Before

Servers and backups checked by hand, outages reported by customers, and a backup discovered empty on the day you need it.

After

Nightly checks, verified backups and automatic restarts. A person is paged only when an automatic fix doesn't work.

Downtime

~45 min → ~6 min per incident

estimate

Time saved

~100 hours a year of checking

estimate

Errors

Bad data stopped before it reaches your reports

Runs entirely on your own infrastructure, and database passwords never appear in logs.

Tested: disk full, empty backup, bad data loadHow it works →

Secure app connections →Logins and data sync between your tools that don't break, double-charge or leak. Automation health dashboard →Know the moment any workflow stops doing its job, even if it never errors. How I measure ROI →The before/after method behind the invoice numbers, with every figure reproducible.

How I work with you

No big-bang rollout. We start with the process that costs you the most and prove the saving before moving on.

Time the process

We map how the work is done today and time each step, so we automate the step that actually costs you, not the one that's merely annoying.

Build with a safety net

Every automation checks its own work and hands anything uncertain to a person. Nothing fails silently.

Test before it touches real data

Each workflow runs through its real-world scenarios first. My portfolio's seven workflows pass 53 automated checks.

Hand over and measure

You get alerts, a plain-English runbook, and the before/after numbers, measured on your process.

Why you can trust the numbers

Automation claims are easy to inflate, so every figure on this page says where it comes from.

real resultObserved

From my own work: the 60% cut in wasted ad spend at Ecomobi.

benchmarkModelled with a published method

The invoice figures come from a before/after model you can rerun. Its dataset is labelled as sample data.

estimateStated assumptions

Volumes and minutes for a typical small team at A$42/hour. See every assumption. In a real project, week one replaces them with your numbers.

About me

I'm Ida Bagus Wicitra Dyaksa (Dyaksa), based in Bali and working remotely. I came to automation from operations, where the job was keeping a high-volume production line on schedule, and from paid media, where I first used rule-based automation to stop wasting ad spend.

The method is the same everywhere: find the step that's the real bottleneck, remove it, and make sure you'll know if it ever comes back. I build with n8n on infrastructure you own, so you keep control of your data and your costs.

Tools: n8n (self-hosted), Google Workspace, Slack, CRMs and business APIs, PostgreSQL, generative AI (ComfyUI / Stable Diffusion), JavaScript and Python.

Operations Staff · BAIADATamworth, Australia · 2023 – 2025

Kept a 24.5+ tons/day operation on schedule, and found and fixed recurring bottlenecks.

Digital Marketing Contractor · EcomobiDenpasar · 2018 – 2020

$4K/month ad spend at 180% average ROI, and 60% less wasted spend through automation.

Manufacturing Operator · Primo FoodsSydney, Australia · 2022 – 2023

High-volume, food-safety-compliant production with batch quality checks.

B.Eng, Maritime Transportation Engineering · ITSSurabaya · 2016

Logistics and process simulation.

Which process is costing you the most?

Tell me about it. I'll send back a short plan: what to automate first, what it should save, and how we'd prove it.

Email wicitradyaksa@gmail.com LinkedIn GitHub