Skip to content
Julia Raciniewska

Practical AI for founders
and small teams.

I'm Julia Raciniewska. I write about practical AI adoption, building SEAR Plugs, and using automation to run a startup alongside a full-time career. Real workflows, honest experiments, and lessons from building with limited time.

Connect on LinkedIn

Portrait of Julia Raciniewska

I work at the intersection of AI adoption, customer outcomes and entrepreneurship. Alongside a full-time role in high-growth AI, I co-founded a specialist water-sports brand with high B2C demand, stocked by 30+ UK retailers and EU expansion in progress. Running both has forced me to be deliberate about time, systems and automation — this site documents the workflows, mistakes and lessons behind that.

02 / Selected work

02

Flexibility Is Not the Same as Freedom

Remote work and flexible hours are conditions someone else sets. Freedom is what you own: skills, assets, systems and options. Notes on the difference, from someone building both a career and a company.

Read now
03

What I Automated First — and What Was Not Worth Automating

A walk through the first year of automating SEAR tasks: which ones paid back the setup time, and which quietly became more work.

Coming soon

All articles

03 / AI workflows

AI workflows for small teams.

I co-founded SEAR Plugs with Andrea Ferrario — a specialist water-sports earplug brand with high B2C demand, 30+ UK retailers and EU expansion in progress. It's also the live testing ground for this site: below are the AI workflows I've built to run it well, from SEO to outreach to operations.

  • High B2C demand
  • 30+ UK retailers
  • EU expansion in progress
A SEAR Plugs case held up in front of a tropical lagoon
Close-up of a SEAR earplug worn in the ear

Website Optimisation

The Weekly SEO Audit: A Rules Engine Instead of 'The Agent Decided'

Every Monday at 07:00, a scheduled audit pulls Search Console signals, checks them against declarative rules, and emails a prioritised action list — with every finding logged and deduplicated in BigQuery.

Read more

Market Expansion & Outreach

B2B Outreach at Founder Scale: 236 Personalised Pitches, Every One Human-Sent

How we ran SEAR's UK wholesale outreach — segment-specific research and drafting by Claude, with a hard rule that no email leaves the account without my review.

Read more

Operations & Systems

Monthly Paid-Social Review on a Fixed Analysis Budget

A sequenced monthly Meta Ads review — fatigue, audience overlap, budget allocation — designed to fit a strict limit on analysis calls and hours.

Read more

04 / Builder mindset

Start with value, not with an agent.

Choose the least complex architecture that can deliver the outcome — and add human control where risk demands it.

01

Frame the value

  • Strategy fit
  • Baseline
  • Impact hypothesis
  • Success metric
02

Choose the operating pattern

Deterministic automation

Code follows fixed rules for stable, repeatable processes.

AI-assisted task

A person directs AI to complete a bounded task.

AI workflow

A predefined process combines AI and deterministic steps.

Agent

AI dynamically selects steps and tools within defined boundaries.

Use the simplest pattern that can reliably deliver the outcome.

How it runs·On demand·Event-triggered·Scheduled·Continuous
03

Design the system

Context · Instructions · Skills · Tools · Orchestration · Guardrails

04

Set the control level

Human-led · Approval gates · Monitored autonomy · Automated assurance

Control can vary by step.

05

Measure & improve

Quality · Adoption · Reliability · Cost · Business impact

What you learn in measurement, feeds the next “frame the value” decision.

Julia surfing

Off the laptop

Outside work I surf, train and travel whenever I can. Those parts of my life aren't separate from the professional story — they're why I care about building better systems and more ownership over my time.

More on that thinking in Flexibility Is Not the Same as Freedom.

05 / Field notes

Field Notes

Occasional notes on AI workflows, building SEAR and what I'm learning. Written when there's something worth sending.