Working Surfacelink · 17 Nov 2026
03
A tall water mill with a series of wooden sluice gates, where loose grain poured in at the top passes each gate only when a miller seen from behind lifts its lever, and comes out at the bottom as neatly stacked sacks, the whole scene drawn with wide empty margins on every side.
AWS Machine Learning Blog

Accelerating product development with AI-driven lifecycle management (AI-PDLC) in AWS Quick

Vishnu Elangovan, Anupa Bhattacharyya, Debaprasun Chakraborty · 11 November 2026

Company story · decisions · artifacts · Decision records

Read the original
Takeaway

Three Amazon Web Services authors describe six AI agents that turn scattered customer feedback into a signed-off specification, with a person approving every stage before coding begins.

Summary

Vishnu Elangovan, Anupa Bhattacharyya and Debaprasun Chakraborty of Amazon Web Services described AI-PDLC, a set of AI agents in AWS Quick, Amazon's workplace AI product. The agents cover the product work before coding: finding customer problems, choosing which to solve, writing a specification and building a clickable prototype. A person approves the output of each stage before the next begins.

Key points
  • A Discovery agent gathers complaints from tools such as Slack, Jira and Salesforce and scores them by severity and frequency.
  • A Prioritize agent scores the options, and a Working Backwards agent drafts a press release and FAQ, Amazon's format for describing a product before it is built.
  • A Spec Generator agent writes a specification with acceptance criteria, each traced back to the original customer problems, plus a brief for designers.
  • The team confirms the problems, approves the priorities, signs off the specification and approves the build, after which coding agents such as Kiro or Claude Code take over.
Implication

A product team can trace each acceptance criterion back to the customer problem it came from. A reviewer can then check a story against its source. That advice is an inference and is not the authors' claim. The post is a vendor's account of its own product and reports no measured results.

Plays
Product

Require a person to approve each stage of agent-drafted product work, from the list of customer problems to the final specification, before the next stage starts.

Product

Trace every acceptance criterion in a specification back to the customer problem and priority score it came from.

Derived by Working Surface from the article; more in the Playbook. Source line: As implementation accelerates, the upstream product development process becomes even more important.

Source issue

17 November 2026: as AI agents and staff outside engineering produce the stories, specifications and tools that others build on, a team must name who checks them and who looks after them.