Why “chatting better” won’t save your business — but building better environments will.

In 2023, “Prompt Engineer” was predicted to be the hottest job of the decade.

We were told the future belonged to those who could whisper the right incantations into a text box.

Two years later, that prediction is collapsing.

Not because AI failed.

Because prompting is fragile.

If your business process depends on someone typing the perfect paragraph into ChatGPT every time something needs to get done, you haven’t automated anything.

You’ve replaced a junior analyst with a faster typist.

The future doesn’t belong to the person who writes the cleverest prompt.

It belongs to the person who builds the best context.

The Empty Room Problem

Imagine hiring a brilliant graduate from Harvard.

You put them in an empty room.

No laptop. No email access. No CRM. No file server.

Then you slide a note under the door:

“Write a marketing strategy.”

It doesn’t matter how smart they are. They will fail.

They don’t know your pricing. They don’t know your customer history. They don’t know your brand voice. They don’t know what you promised in last quarter’s board deck.

This is how most companies treat AI today.

They open a fresh chat window — an empty room — paste in a prompt, and expect magic.

When the output is generic, they blame the model.

“I need a better prompt.”

No.

You need a better environment.

From Prompting to Onboarding

At Imbila, what we’re seeing — especially with agents like Penny Claw — is a shift from Prompt Engineering to Context Engineering.

When Penny was deployed, we didn’t obsess over a 500-line system prompt.

We onboarded her.

  • Read access to the calendar
  • Structured memory folders
  • Invoice templates
  • Brand assets
  • Defined write permissions
  • Clear boundaries of action

We didn’t “tell” her how to be useful.

We gave her the tools to work.

That’s the difference.

Artefacts > Adjectives

Prompt engineers focus on adjectives:

“Act as a world-class, professional, witty copywriter…”

Context architects focus on artefacts:

“Here is our 50-page brand bible in markdown. Here are the last 10 client emails. Here is the CRM export. Draft a reply referencing the live project status.”

One approach is theatre.

The other is infrastructure.

One disappears when the chat window closes.

The other compounds.

The Real Strategic Question

If you’re leading a business, stop asking:

“How are we using AI?”

Start asking:

“Where does our AI live?”

Is it trapped in a browser tab?

Or does it live inside your systems — your documents, your CRM, your shared drives, your operating rhythm?

Prompt Engineering is about talking to the machine.

Context Engineering is about integrating the machine.

The companies that win in 2026 won’t have better prompts.

They will have living context — structured, permissioned, accessible.

They won’t have chatbots.

They’ll have digital employees.

🧭 Imbila Insight

Moving from chatbot experiments to digital employees requires more than subscriptions. It requires structured data, permissions, and workflow thinking.

If you’re serious about this shift, take the AI Assessment and see where your organization stands: https://www.imbila.ai/ai-assessment/

Take your own personal AI assessment if you’re interested in our new AI Mastery Course

AI Workshop - Mastery SessionAI Workshop - Mastery Session A hands-on, personalised session that transforms how you work with AI — tailored to your exact level, tools, and goals Take the Pre-Course Diagnostic → Most AI workshops give everyone the same demo. This one starts with a diagnostic. Before you walk in, we already know yourImbila.AICraig Leppan (Imbila)

“Context Engineering is not just a technical discipline — it’s a governance discipline. Who owns the data? Who defines access? Who audits actions?”