aws-samples/aws-ml-enablement-workshop

What it solves

Many organizations struggle to turn AI/ML ideas into real product growth. The ML Enablement Workshop addresses this by helping cross-functional teams (business, development, and data science) align around customer needs and validate ideas quickly, using generative AI throughout the process. It aims to reduce the high failure rate of ML projects by teaching a practical, customer-obsessed workflow that leads to production releases in about 6 months.

How it works

The workshop has three stages: Day0 (1 hour, can be remote) confirms roles and readiness; Day1 (practical, 3.5 hours) walks participants through Amazon's "Working Backwards" process, where they generate AI-assisted press releases and build mockups for hypothesis testing; Day2 (improvement, 3 hours) integrates learnings from the mocks, updates the press releases, and plans the first steps toward a release. A tool called Kiro (spec-driven generative AI) is used to practice writing specifications, and example outputs (press release, mock app) are provided to illustrate the process.

Who it's for

This is designed for product teams — including product managers, engineers, data scientists, and business stakeholders — who want to work together to identify and validate generative AI use cases for their products. It is also suitable for organizations that want to build cross-functional collaboration and adopt Amazon's product development culture. The workshop is available as open-source materials, so teams can run it internally (with some naming restrictions), or AWS can deliver it if conditions are met.

Highlights

• Learn and practice Amazon's "Working Backwards" process in a short, hands-on format
• Apply generative AI from ideation to mock construction, refining product development habits
• Use mock-based feedback to avoid armchair plans and ground decisions in real user input
• Includes concrete examples (press release, mock app) and customer case studies from companies like MUFG, JINS, and BASE
• Free to use under an open-source license, with optional AWS support for prototyping afterwards