AI Share
& Learn.

How we're building the agency of tomorrow, today.

14 APRIL 2026 · ALL HUMAN · NICK BERNON, DELIVERY

Today's session

01AI at allhuman today
02How are our roles changing?
03How I'm using it as PM

01

Part 01

What we've
built.

AI at allhuman today

Tokamak / allhuman.com redesign

AI-driven outputs.

Damir Mehmedbasic

AhCMS

Jasmin Luka

Intent-Driven Landing Page

Orla Madden

Design System

Ashwini

QA Automation

Damir Mehmedbasic · AhCMS

One developer. One day.

Typically takes

1.5–2 sprints

Damir completed this in a single day

Free · Fully customisable

AhCMS →

Jasmin Luka · Intent-Driven Landing Page

A landing page that adapts
to each visitor.

01 Raw signals Time of day · referrer source · scroll depth · engagement time · tab returns · device type
02 Derived intent Persona (LinkedIn → executive · GitHub → builder · Google → marketer) · Urgency score (0–100) · Journey stage
03 What changes Headline · CTA text · Service order · Accent colour, over 20 variants in total

Try it live · open these side by side

Executive view → Builder view → Marketer view →

Orla Madden · Design System

A full design system,
documented by AI.

Tokens

Colour, typography, and spacing defined and documented. The foundation everything builds on.

Components

Core UI components built and named to match code conventions. Brand-aligned with Katya over two weeks.

Documentation

Generated using Claude. Phase 1 and case study templates complete in Figma. Pending sign-off.

Work page → Case study →

FiTrade · First live client engagement

Our first AI-managed
client engagement.

Nick

AI-managed PM: backlog documentation, ticket drafting, meeting summaries. Full project knowledge base maintained automatically. Stage 1 live.

Marko + Ashwini

Test cases generated from Azure DevOps tickets and Figma screenshots.

Same team. More coverage. The quality bar does not move.

02

Part 02

The model.

How are our roles changing?

The Generalised Workflow

AI generates. Humans direct and approve.

Role What AI does now What changes next

PM / BA

Auto-processes meetings; logs decisions; maintains task ledger; generates backlog items

Project context self-maintains. Human role = review and decide.

Designer

Figma designs, component documentation, and working code. Transitions, animations, and interaction design tested via Claude Code MCP.

First layer of development. Specs feed code directly. No handoff gap.

Developer

Picks up from the designer's code. Integrates, extends, and builds on what AI has already generated.

AI operator: sets standards, governs output, makes architectural decisions.

QA

Generates test cases from tickets and designs. Human validates.

AI generates, maintains, and proposes fixes. Human remains the quality gate.

Every role, sharper. Every output, held to a higher bar.

03

Part 03

Using AI
as a PM

My experience managing a project.

AI Operations Layer Active

Using AI in my day-to-day.

The benefits include:

Meetings to Actions

AI triages transcripts, normalises speaker names, extracts decisions, logs activity, and generates presentations.

Daily Brief and Tasks

AI syncs the calendar, prioritises my task list, and delivers a daily briefing on request.

Board Diff and Health

AI compares successive board exports to surface changes, spot drift, and track project health over time.

People Signals

AI runs sentiment analysis on conversations, identifies team member unhappiness, and proactively flags possible causes.

Human clarity.
AI velocity.

ALLHUMAN · TOKAMAK · 2026

AI Share& Learn.