Stephany Oliveros at Zenler AI Summit 2026 — Human-Centred AI Agents and Workflows

Stephany Oliveros showed how to build AI agent workflows using Claude at the AI Summit 2026. Watch the full session with your 60-day Zenler free trial

Stephany Oliveros is co-founder of SheAI — an educational platform helping people learn to use AI without the overwhelm — and she's spent years thinking about why AI keeps disappointing the people who try to use it seriously.

Her answer: it's not the tools. It's the environment you put them in.

Her session at the Zenler AI Summit 2026 took the concept David Newton introduced — building AI systems rather than relying on one-off prompts — and extended it significantly further. Into the world of AI agents, skills, multi-agent workflows and a complete command centre structure built using plain text files.

Rakesh Vallil called it out specifically in the summit's closing remarks.

Why AI Keeps Failing — and the Fix

Stephany opened with the frustration most people experience but rarely articulate clearly.

You start a conversation with an AI tool and it goes well. You come back the next day and it's lost everything. You add more context. It drifts again. You start over.

She explained exactly why this happens — and it's not the AI's fault. There's a structural reason, and understanding it is the first step to building something that works consistently.

AI as Infrastructure — A Landing Page Built in Five Minutes

In the teaser clip below, Stephany makes the concept of AI as infrastructure visible in the most direct way possible: she shows it working.

She talks through her thesis research on AI and human cognition — specifically the finding that AI can liberate cognitive load and help you be more creative, but only when used within the right framework. She also covers the concern that 40% of people who overuse AI start skipping verification, and why responsible use means staying in the loop.

Then she shares her framing: AI is like electricity is for the internet. It's infrastructure. The question is whether AI is directing your work, or whether you are directing it.

And she shows a landing page she built in five minutes from a basic prompt — pulling real information from two brands, real photos from Pinterest, working links, clear calls to action. The whole thing functional and live from a single casual conversation with an AI agent.

Watch it — then read on for exactly how she built the system that made it possible.

Stephany Oliveros — Zenler AI Summit 2026 Teaser

Two Files That Change Everything

Before building anything, Stephany recommends starting with two simple text files.

One captures who you are — your values, your voice, your way of thinking, your standards.

The other captures what your business is — your goals, your model, your clients, your processes.

Together, these files load at the start of every AI session. The AI knows who you are from the first message. You never start from zero again.

She covered exactly what goes in each file, how to build them efficiently, and the one mistake that makes them less effective. That's in the full session.

Skills, Agents and Workflows

Stephany introduced a clear vocabulary for building AI systems: a skill is a detailed instruction for completing one specific task; an agent is a role with a goal; a workflow coordinates multiple agents and skills into a system that runs start to finish without you managing every step.

All text files. No coding. But the difference between having them and not is the difference between prompting and orchestrating.

She demonstrated a live build — a complete landing page from research through to copy and design direction — built by a coordinated workflow in under five minutes from a single prompt.

The full demonstration is in the replay.

The Command Centre

Stephany shared the folder structure she uses to organise everything — four folders and a set of named text files. When this structure is in place, the AI understands the hierarchy automatically.

The complete structure with examples of each file type is in the full session.

Get the Full Session

Stephany's complete session includes the full context engineering explanation, the live build demonstration, the command centre structure walkthrough, and a Q&A getting into Claude Code, Cowork and building your first agent.

Claim your 60-day free Zenler trial and watch Stephany's full session →

What's Inside the 60-Day Free Trial

Full replay of all 14 summit sessions plus the 5-hour bonus workshop
All speaker bonuses including David Newton's AI Creator Handbook, 100 Prompt Playbook, Website Auditor and Social Fix Auditor
The Zenler Accelerator Program
Weekly Live Support Desk with Allie Miriam every Thursday
The complete Zenler platform — all included, no hidden costs

Limited-time offer. After the countdown it reverts to a standard 30-day trial.

Get your 60-day free Zenler trial now →

FAQs

What is context engineering?
Building structured information files that load at the start of every AI session — so the AI knows who you are without needing to be told each time.

Do I need to code to build this?
No. All files are plain text in plain English. Stephany built the entire live demonstration without writing a single line of code.

What's the difference between a skill, an agent and a workflow?
A skill is step-by-step instructions for one task. An agent is a role with a goal. A workflow coordinates multiple agents and skills into a complete system. Stephany explains all three in the session.

How is this different from David Newton's approach?
David builds instruction files and voiceprints in ChatGPT Projects. Stephany builds context files and agent systems in Claude Code. Same underlying principle — different platforms and depth of complexity. Both sessions are in the replays.

What is CAI?
Co-founded by Stephany Oliveros, CAI is an educational platform focused on helping people learn AI with an emphasis on human behaviour and practical adoption.