Marketing tools
built with AI

I started vibe coding in summer 2026 and it changed what one marketer can ship. Instead of waiting for a developer or settling for a template, I describe what I need and work with AI tools to build it. Five things came out of it so far: three on this page, and two more as separate cases below.

One person publishing like a team

Problem

A founder community needs regular fashion-tech content, and I am the only person producing it.

Insight

Finding and drafting is the slow part. Judgement is the part that must stay human.

Move

An agent reads trade news from UK and EU sources, removes repeats and ranks stories by relevance. It writes a research note and a LinkedIn draft, then sends it to me in Telegram with Approve, Rewrite and Skip buttons. A trend mode looks for patterns across stories and says so when there is not enough evidence. If the AI service is down, a fallback writer still produces a draft.

Status

Running in test: 12 stories collected and 15 drafts generated, all waiting for review. Publishing to LinkedIn is the next step and is not connected yet.

Turning a visit into a conversation

Problem

At fairs and trade shows people collect cards and rarely follow up.

Insight

People talk when they have a reason to.

Move

Built two quizzes on the same pattern: a few questions, a match, and an email with the result. One for PRISAKARU at Maison&Objet in English and French, one for Avantex that matches visitors with exhibitors. Contacts land in Airtable or Google Sheets, with GA4 tracking.

Status

PRISAKARU: built for the stand. Avantex: built and tested, the team has not used it yet.

PRISAKARU quiz: language choice, a question in English, a result in French.

A site with logins, built alone

Problem

Fashion 4.0 had a vision and a network, and no website or member area.

Move

Built the platform in HTML, CSS and JS with Supabase for login and data. Applications sync live from Airtable, and accepted members get access to a directory and an opportunities board.

Result

21 members admitted.

Two more tools, as full cases

My own analytics service for a client

An analytics service I built for a client, as a Telegram app, for £0 a month.

View case →

Salt cave funnel, built by an AI team

A game quiz, landing page, payments and referrals for a £3 trial session.

View case →

Connect the agent to LinkedIn, add source checks before approval, and log which drafts get approved so the agent learns my taste.