The ‘Loop of Loops’: rethinking Product Discovery in the AI era

This is the fourth blog in Mike’s 8-part series about how AI can become a force for clarity and acceleration across the broader software development lifecycle.

The Loop of Loops Product Workflow

Building on this foundation, the team defined a three-phase iterative workflow:

1. Problem discovery
2. Solution discovery
3. Delivery

Each phase operates as its own refinement loop, with explicit checkpoints for human review; overall these are contained within the larger interactive lifecycle.

This ‘loop of loops’ reflects established agile practice and other frameworks such as the Double Diamond Design Process.

AI fundamentally changes the economics of each loop. Activities that were previously time-intensive become fast enough to repeat multiple times without too much effort. Teams can explore more options and make larger bets within the same time, without increasing risk.

The workflow begins with problem discovery, where AI is used to:

• Transcribe and summarise interviews
• Synthesise interview outputs alongside curated business and technical inputs

Once a clear understanding of problems and needs emerges, the process moves into solution discovery, where teams:

• Translate synthesis into product stories with acceptance criteria
• Define user journeys and refine stories accordingly
• Prioritise stories and load them into a task tracking system

To support multiple initiatives running concurrently, iteration-specific directories are introduced to compartmentalise discovery work.

An example repository directory structure, with shared context in yellow, and product-specific content in blue

The final phase is the delivery loop, where teams:

• Confirm what was actually built
• Update backlog and story status
• Produce an as-built product specification

“When AI is generating artefacts, unchecked drift compounds quickly. Explicit human review at each checkpoint prevents this, creates space for judgment and correction, and ensures that accountability for outcomes remains with the product manager.”

Defining the workflow established the ‘what’ of each phase; the next challenge was ensuring ‘how’ each step could be executed consistently and repeatedly.

Orchestration

Each activity in the workflow requires explicit guidance on what context the LLM should read, what work it should perform, and what outputs it should produce. This guidance was encoded using Claude Code’s slash commands.

For example:

• /iter creates a new iteration and captures its intent
• /synth produces a structured synthesis from discovery artefacts

Each command applies consistent context and instructions, enabling repeatable execution (with an audit trail).

Workflow command reference across delivery phases

As the workflow matured, the underlying orchestration mechanisms evolved as well.

Skills and future orchestration

Since this work began, Anthropic introduced skills within Claude Code – a more structured evolution of slash commands that bundle templates, rules, and scripts into reusable units. Unlike explicit commands, skills can be inferred and applied automatically based on user intent, making them well suited to iterative and agentic workflows where tasks do not follow a fixed sequence.

For clarity, this article references slash commands, but in practice future iterations of the approach will increasingly rely on skills as the primary orchestration mechanism, especially as work becomes automated with agents.

With orchestration in place, the workflow can be examined phase by phase, starting with problem discovery.

Read the full Evolving product discovery for AI delivery eBook

About the author

Mike Mitchell is a Product and Strategy Principal at Equal Experts, helping organisations adopt modern product practices and integrate AI into their product lifecycle in practical, sustainable ways.

His current focus is on helping product teams do more meaningful strategic work with AI, both by accelerating the routine work that crowds it out, and by making AI a more proactive, multi-modal collaborator.

Mike brings 35 years of experience in enterprise software, with a foundation of engineering followed by two decades of product leadership. He is a founder of multiple startups, has advised dozens more, and has led multiple enterprise transformations. Just prior to EE, Mike served as a fractional Chief Product Officer, guiding early- and growth-stage firms to deliver value faster through stronger product strategy and operating models. He holds a degree in Computer Science & Engineering from MIT.

Mike’s next blog in this series covers how AI is transforming Problem Discovery.