Architect

23 June 2026

How to drive value with AI

Productivity wins. Workflow transformation. Innovation.

Lots of companies see AI deliver great productivity wins. Someone drafts a sales proposal. Another person designs a pack with AI. A third person writes a couple of social posts.

These are real wins. The time saved is real. But they're treating the warm-up like it's the workout.

There are three types of value you can generate with AI. Most teams never get past the first one. Not because they can't. Because they don't know what comes next.

When I surveyed 529 senior marketing leaders, the single most common challenge was building agents and automation. 140 separate verbatims about the same gap.

  • "Take the step from AI Operator to Strategic User."

  • "Scaling from ad-hoc AI to agentic workflows."

  • "Moving from experimentation and task productivity to embedded workflows."

They can feel there's something beyond productivity gains. They just can't see the path.

That's what the AI Value Framework is for. Three types of value. A way to name where you are and see what comes next.


Productivity wins

This is where almost everyone starts. And it's important.

Same workflow. AI handles some steps. A human drives the work. AI assists on individual tasks. Your writing assistant checks grammar. Your compliance tool flags risks. Your reporting tool summarises the numbers. One task at a time. High control.

The value is linear. You save 20 minutes on task A. You save 15 minutes on task B. Add them up and you might claw back an hour a day.

Good. But you haven't redesigned anything. The workflow is identical. You've just made some of the steps faster.

💬 If AI is only saving you 20 minutes a day, you haven't redesigned anything yet.

I hear this constantly from marketing leaders. "I use the LLM chatbots, but haven't built any agents." "I use ChatGPT mostly but know there is much more to explore." That's productivity wins. Valuable. But not transformative. And most teams stop here because it feels like progress.

Examples of use cases:

  • A content writer using AI to draft a social post

  • A lawyer using AI to summarise case files

  • A marketer using AI to check tone before sending

  • You draft the content, AI checks compliance. Very popular in healthcare.


Workflow transformation

This is where the shift happens. You stop optimising individual tasks and look at the entire process. The question changes.

Not "How can AI help me do this task better?" but

  • "Why does this workflow have seven steps when it could have three?"

  • "Which steps can AI do better or on par with humans?"

  • "Which steps need human judgement?"

Workflow transformation means you completely re-engineer how the work gets done. Steps get removed or added, not just accelerated.

Take content creation. Most teams write posts separately for LinkedIn, Instagram, TikTok and email. Four different processes. Four different people. But the core message is usually the same.

What if the human owns the unique messaging and AI writes different assets for all channels using your voice, your examples, your guardrails? You've just cut creation time in half and improved consistency. That's not doing the same work faster. That's doing different work.

Or look at proposal-to-delivery. Calls happen. Notes exist somewhere. Someone drafts a proposal. Someone else creates a presentation. Someone else builds a project plan. Four tasks. Two days. Three people.

What if a single agent read the call notes, drafted the proposal, created the presentation and generated the project plan in sequence? Same output. Unrecognisable process.

💬 You're rethinking steps, not just speeding them up.

Examples of use cases:

  • You come up with messaging as input. AI writes different assets for all channels using your templates, examples and guardrails.

  • A single agent reads call notes, drafts the proposal, creates the presentation and generates the project plan.

  • News articles automatically repurposed into blog posts, email newsletters and podcasts.

Case study: ElevenLabs

Built an automated case study engine. It interviews customers via voice agent, writes the case study, edits and publishes it.


Innovation

This is rarer. And it's where the most interesting work lives.

Innovation means new products. New formats. New experiences. You stop asking "How do we do this faster?" and start asking "Why are we doing this at all?"

Think about FAQ pages. They exist because customers have questions. Every company builds one. But what if instead of an FAQ page, customers talked to an AI avatar trained on your content? Not a chatbot with canned responses. A conversational experience that gives personalised answers from your actual knowledge base. Better experience. Lower cost. Better answers. I'd prefer an AI avatar on my bank app to answer all my questions rather than going through a long FAQ.

Examples of use cases:

  • An AI avatar that answers client questions using your content.

  • Websites designed entirely to feed LLMs. Not for humans to read. For AI to learn from. New format. New purpose.

Case study: Harvey

  • Built personalised demos for lawyers. Not one generic deck. Every prospect sees a demo with their own data, their own scenarios, their own edge cases. The buyer sees their world, not a hypothetical. Conversion improves. Sales cycle shortens.

Case study: Wealthsimple

  • Built Willow. An AI voice agent that makes outbound calls to new customers. Not an assistant helping a human make calls. A completely redesigned onboarding journey. 10% lift in conversion.


How to progress

You don't jump from productivity wins to innovation in one step.

AI is like fitness. You don't go from never exercising to running marathons.

You build. You meet yourself where you are.

Master productivity wins. Feel the value. Build trust in the technology and in your own judgement. Then look at your workflows and ask: what if we redesigned this completely?

Once workflow transformation is embedded, innovation becomes obvious instead of impossible. You start noticing the bigger questions because you've already proven that how we do this can change completely.

💬 The question isn't "Are we using AI?" The question is "What value are we generating?"

Actually useful AI. In your inbox.

Monthly newsletter with use cases, playbooks, case studies from top companies and invites to live webinars and events.