PTA

AI // Studio position

AI is part of the system. Judgment still leads it.

AI is useful when it helps us understand more, test more and remove unnecessary work. It is not a substitute for judgment.

We don't begin by asking where AI can be added.

We begin by understanding the work.

Framework // Where technology earns its place

Not every problem needs the same kind of help.

Requires judgment

  • Strategy
  • Context
  • Taste
  • Trade-offs
  • Relationships

Human-led

Benefits from structure

  • Research synthesis
  • Pattern detection
  • Repetition
  • Data processing
  • Testing

AI-assisted

Can be reliably systematised

  • Defined workflows
  • Validation
  • Classification
  • Routing
  • Routine operations

Automated where appropriate

Principle // The rule

The rule is simple.

If the problem needs judgment, we apply judgment.

If structured processing can make the work faster or more reliable, AI can help.

If a repeatable process can be safely automated, we build the system to handle it.

The technology follows the problem — not the other way around.

System // Where AI enters the work

AI doesn't sit everywhere.

Its role changes depending on what the work actually needs.

Human judgmentAI-assistedAutomated where safe

01

Notice

Mostly human — noticing starts with attention, not tooling.

02

Understand

AI may help synthesise research and surface patterns.

03

Structure

AI may help stress-test assumptions and expose edge cases.

04

Build

AI may support implementation, testing and documentation.

05

Evolve

AI may help surface recurring friction from real use.

See the system behind the work

Practice // What this means in the work

What this looks like in the work.

We use configured AI systems to:

  • challenge assumptions
  • stress-test architecture
  • surface edge cases
  • structure complex information
  • accelerate repetitive analysis
  • support implementation and testing
  • automate defined, repeatable work

We don't hand judgment to the model. The judgment remains ours — AI-assisted work stays subject to human review appropriate to the risk and context.

Position // Closing

We don't AI-enable broken processes.

We understand the work first.
We question the assumptions underneath it.
We design the system around reality.
And then we decide where AI earns its place.

Noticed, not invented.