Topic 06 · Build with technology
Understand what a technology makes possible, what it requires, and whether it is useful for the product.
Articles about AI, automation, data platforms, technical systems, and the product choices that come with them.
10 articles
Common questions
01
Which capability creates real user value rather than novelty?
02
What data, infrastructure, and safeguards does it require?
03
How should an intelligent system communicate uncertainty?
Recommended reading
It covers the basics. The articles below explore related questions in more detail.
A good place to start
01
Filed under
Artificial intelligence · Data platforms
Choose AI bets by the learning advantage they can create: define the user decision, evidence loop, economics, operating limits, and proof required to scale.
11 min read
02
Artificial intelligence
Design AI features around user decisions: set the system’s role, expose useful evidence, support correction, and evaluate reliance as well as output.
Continue03
Engineering
Turn a product promise into clear data requirements across collection, contracts, quality, freshness, ownership, privacy, serving, and failure recovery.
Continue04
Artificial intelligence
Treat every automation as an operational promise: define its trigger, authority, decision boundary, recovery path, owner, and evidence before choosing a tool.
Continue05
Delivery
Translate product promises into system requirements: authoritative state, visible delay, failure recovery, workload shape, observability, and safe change.
ContinueMore on AI & Technical Product
6 more articles
01
Artificial intelligence
Build a task-specific evaluation set, choose credible graders, define release thresholds, and turn production failures into regression tests.
02
Artificial intelligence
Define the workflow, data, serving, monitoring, and recovery conditions an ML system must satisfy before users depend on its predictions.
03
Artificial intelligence
Treat personalisation as a decision policy: define eligibility, choices, outcomes, and how the product behaves when evidence is weak.
04
Artificial intelligence
Govern AI across its lifecycle with clear decision rights, scoped use, evidence, release gates, supplier controls, incident response, and retirement.
05
Artificial intelligence
Design AI-assisted work around a clear task, context contract, review burden, evidence trail, correction path, and accountable human decision.
06
Artificial intelligence
Choose, evaluate, and operate NLP or LLM capabilities inside a bounded workflow—with source provenance, human authority, monitoring, and rollback.