Designing with AI
in RQI product suite

Principles, patterns and a process for designing outcome-driven intelligent experiences

Let's never start by asking

“How do we add AI into our product?”

Instead, ask

“What problem are we designing to solve, and what outcome are we trying to create?”

There is nothing fundamentally new about this principle.

This is how we have always designed good products. We would never ask, “Where can we add more tables to the product?” AI should be treated in the same way.

What is different is the range of responsibilities the product can now potentially take on.

Users don't care whether something is powered by AI.

They care whether the product helps them achieve better outcomes with less effort.

AI should

Remove work. Take responsibility for work. Make something possible that wasn't practical before.

It should not simply make the same work happen in a different way.

Principles

4 principles to guide our thinking

Principle 1

Design for outcomes, not existing workflows

Don't assume the tasks users perform today are the tasks they should perform tomorrow.

Start with the outcome, then explore how intelligence could change the path to it, even if that eliminates steps, workflows or responsibilities entirely.

Don't ask“How can AI help the user complete this workflow?”

Ask“If we designed for the outcome instead of the workflow, what would the product do differently?”

Principle 2

Context is the foundation

AI becomes valuable when the product understands enough context to recognise what matters.

Explore what becomes possible when the system can connect signals across data, understand relationships, recognise emerging situations and use that context to determine what might need attention.

The user shouldn't have to continuously monitor the product and connect the dots themselves.

Ask“What could the product understand that the user currently has to figure out themselves?”

Principle 3

Let the system do the work

Don't stop at helping users perform tasks faster.

Explore what work the product could take responsibility for entirely.

Users can define intent, boundaries and exceptions. The system should handle the rest where it can do so reliably and safely.

Ask“What work could the user hand over to the product rather than doing step by step?”

Principle 4

AI needs to earn its space

Intelligence should expand what a product can make possible, not make existing functionality appear smarter.

Challenge whether AI is fundamentally improving the experience or merely adding another way to interact with something we already built.

If AI doesn't remove work, improve an outcome or unlock something that wasn't feasible before, question whether it belongs in the experience.

Sometimes the best solution will not involve AI at all.

A process for mapping AI opportunities

The goal is not to look for places to insert AI.

Start with an outcome and systematically explore how intelligence could change the product's responsibility for achieving it.

  1. Define the outcome
  2. Map what drives the outcome
  3. Explore intelligence patterns
  4. Compose the experience
  5. Decide how much responsibility the product should take
  6. Find the smallest valuable intervention
  7. Release, learn and expand

1 · Define the outcome. For example:

Equipment readiness

Keep equipment ready when and where it is needed.

2 · Map what drives the outcome

Identify the information, signals, dependencies, decisions, people and systems that influence that outcome.

For equipment readiness, that might include:

Equipment usagePerformance signalsMaintenance historyHardware status Upcoming station bookingsTraining schedulesAvailable replacement equipmentService requirements

Instead of jumping immediately to a solution, map the system around the outcome.

This creates the context needed to explore what the product might understand or do differently.

3 · Explore intelligence patterns

Use the following patterns as lenses for exploring what becomes possible.

They are not a checklist, and a solution may combine several or none of them.

4 · Compose the experience

The patterns become more powerful when they work together around the same outcome.

Take equipment readiness.

Instead of building an isolated AI troubleshooting assistant, imagine a product that takes increasing responsibility for keeping equipment ready across an organisation.

Illustrative example

Detect

Station 7 is behaving differently from comparable stations.

Anticipate

Based on recent signals and maintenance history, Station 7 is likely to become unavailable before next month's training.

Recommend

Affected sessions could be moved to Station 4, and Station 7 should be serviced.

Act

The product prepares the required schedule and service changes for approval.

Automate

Where predefined rules and confidence allow it, routine changes are handled automatically.

Orchestrate

Bookings, equipment availability, schedules and service processes are coordinated around the outcome.

The result isn't simply “an AI feature.” The product is taking increasing responsibility for:

Keeping equipment ready when and where it is needed.

6 · Find the smallest valuable intervention

The complete opportunity may be large, but the first release doesn't have to be.

Look across the experience and identify the smallest capability that could create meaningful value independently.

Detect
First release
Anticipate
Recommend
Act
Automate
Orchestrate

The important distinction is

Start small in implementation, not in thinking.

Understand the broader outcome first. Then identify the smallest piece worth releasing.

7 · Release, learn and expand

Release the capability and evaluate whether it actually changes the outcome.

Then reassess the opportunity before building the next capability.

The goal isn't to execute a predetermined AI roadmap.

If one capability eliminates the need for another, that's success.

The mindset

The process can be summarized as:

  1. Outcome
  2. Understand the context
  3. Explore intelligence patterns
  4. Compose the experience
  5. Decide product responsibility
  6. Find the smallest, valuable intervention
  7. Release and learn

And sometimes the answer will be AI.Sometimes it won't.

Designing with AI in the RQI (Resuscitation Quality Improvement) product suite. Examples are illustrative ideas, not today's product.