AIAI-First Marketer

Less busywork. More marketing.

We rebuilt how our marketing team works with AI. Same team, no developer, no big new budget. Everything from our talk is here, free to take.

10 weeks

30 days


Same team. No developer. No big new budget.

The future of AI in marketing isn't generative. It's outcomes.

1

Give AI one place to read.

A knowledge base of everything you ever made. No source, no answer.

2

Let it do the busywork.

AI finds, drafts, checks, versions and measures. People own the ideas.

3

Count it with three white lights.

Data, tools, results: three checks before a number reaches sales.

Effort isn't a result. Here's what counted.

Leads matched to their account

4 in 10

9 in 10

Event follow-up prep

2 hrs

2 min

Sales cadences

1,300

16

Brief to first draft

10 weeks

30 days

Tungsten Automation marketing operations, before and after each rebuild.

Match on the website domain, enrich, and use intent for the gaps. Bad addresses come out before anything sends.

A person checks how the AI reads the numbers. The run stops when a number looks wrong.

Every figure traces back to the CRM. Any gap goes up front.

The rule: automate where a mistake is cheap and easy to spot. Keep a person where it's expensive, or quiet.


One kit. Five templates and every skill from our build.

One download, one zip. Open it and start with template 01.

Download the kit (.zip)
01

One slow process

Measure it, then place every step.

02

Pilot pitch

Pitch the pilot on one slide.

03

Knowledge base and first skill

One place the AI reads. One skill that asks first.

04

Before-it-goes-live check

Five checks before anything goes out.

05

Three white lights and red light log

Three checks before a number reaches sales.

+

The Marketing Studio

Fifteen skills that take a brief to approved copy. Type /start: it asks which team you're on and what you want to do, then routes. Fill in one file and it writes in your voice. It never invents a fact.

  • start
  • house-rules
  • build-knowledge-base
  • brand-system
  • find-assets
  • campaign-concept
  • sme-interview
  • meeting-source
  • personas
  • write-copy
  • brand-check
  • timeline
  • tailor
  • review-and-handoff
  • project-setup
View the skills on GitHub ↗

Public builds of what you saw today. No internal data. Yours to adapt.

Find yourself. Then take one step on Monday.

One view of what works, with a source behind every number.

  1. 1

    Pick one slow process

    A campaign everyone knows takes too long. Map its steps.

  2. 2

    Show the gap, not the tool

    Weeks from brief to draft. Budget spent making, not thinking.

  3. 3

    Bring IT and data in early

    Who owns the licences, the connectors and the data you'll trust?

  4. 4

    Pilot one campaign, then re-test

    One quarter, one before-and-after. Then ask for the next step.

The starter kit has a template for each of these steps.

Two AI-first marketers. One team.

IP

Inês Piçarro

I lead marketing's move to AI: the knowledge base, the design system, the skills and agents the team works with every day. Before that, ABM at Cisco across EMEA and digital for Vodafone. I also teach the basics of AI, one-to-one.

At the talk: Part 1, Foundation · Part 2, One idea. Every account.

DQ

Daniele Quartana

Daniele builds the systems that turn marketing into pipeline: lead routing, data quality and reporting, now with agents built in Claude Code. He started out selling windows door to door, and still thinks like a seller.

At the talk: Part 3, What counts.