The Lab

Practical AI systems for real work.

I build and experiment with AI workflows and small systems to solve real problems across business development, content, programmes and everyday work.

n8n Claude Supabase Google Workspace + more

Live workflow

TRIGGER

New Signal Opportunity detected

Enrich Context Research company

LOGIC

Score & Prioritise Assess fit

DATA

Update CRM Store opportunity

Draft Outreach Grounded messaging

HUMAN

Review Approve next step

> scanning for opportunities...
> 12 new companies found
> enriching context with AI...
> 8 high-fit opportunities
> drafting outreach...
> ready for review ✓

Selected builds

Small systems. Useful outcomes.

Most of these started as something I needed for myself. The useful part is not the automation itself — it is figuring out where automation genuinely improves the process.

All Business Development Content Personal Ops Research

New comment

1 Detect keyword

2 Send DM

3 Reply publicly

4 Log activity

Social · Automation

Instagram Comment → DM Agent

↗

Detects selected comment keywords, sends relevant DMs, manages public replies and tracks activity while avoiding duplicate executions.

n8n Instagram API Supabase

Job Hunting Agent

01 Find relevant roles

02 Score role fit

03 Generate tailored CV

04 Draft cover letter

05 Track applications

Career · Agentic workflow

Job Hunting Agent

↗

Finds relevant roles, scores fit, tracks previously seen jobs and produces tailored CV and cover letter drafts for shortlisted opportunities.

n8n Claude API Google Docs Google Sheets

This week's meal plan

M Pumpkin oats

T Chicken rice

W Broccoli egg

T Salmon porridge

Personal Ops · Telegram

Baby Meal Planning Bot

↗

Generates weekly meal plans from an approved feeding guide, previews them for confirmation and logs approved weeks without duplicates.

n8n Telegram Google Sheets Claude API

Content Radar

01 Trending topics

02 Relevant examples

03 Filter weak material

04 Weekly digest

Content · Research

Content Radar

↗

A recurring content-research workflow that surfaces relevant examples across selected niches, filters out weak material and produces a concise digest of what is worth watching.

n8n AI research Email digest

My approach

Useful beats impressive.

Start with the problem.

The workflow should exist because something is repetitive, messy, slow or easy to forget.

Keep humans where it matters.

Research and drafting can be automated. Judgment and key decisions often should not be.

Measure real impact.

Time saved, opportunities created and errors reduced matter more than how complicated the workflow looks.

Build for maintainability.

A system I can understand and maintain six months later is usually the better system.

Tools I work with

Current toolkit.

n8n

n8n Workflow orchestration

Claude Reasoning & generation

Supabase Structured data

Google Workspace Review & outputs

More Constantly testing

Building in public

More useful systems to come.

The Lab is where I document what I'm building, what works and what turns out to be much less useful than expected.

Let's connect →