AWS Bedrock builder tools

Launching Industry-First GenAI Features - Agents & Knowledge Bases: 0 to 5.2M Tools in 90 Days

Bedrock Builder tools solve the problem of building AI applications when there's literally no roadmap to follow. Starting with nothing but an idea and a 90-day deadline, Knowledge Bases and AI Agents went from concept sketches to powering over 5.2 million real-world applications—proving that speed and quality can be allies.

Role:

  • Independently led strategy and design
  • Mentored & onboarded other designers

Responsibility:

  • Launch industry-first features in 90 days.
  • United two racing teams to maximize customer value
  • Establish scalable foundation

Timeline:
May 2024 - July 2024

Problem statement

Building AWS Bedrock builder tools from 0 to Market Leader
‍

With Microsoft, Google, and OpenAI racing to market and 1,000+ business opportunities at stake, AWS had 90 days to launch industry-first GenAI builder tools i.e. Knowledge Bases and AI Agents, starting from absolute zero: no MVP, no playbook, no standards, and a team that didn't even know what the first version should include.

Building foundation

Starting from ground zero with no playbook

Technical deep dives:
I dove into technical docs to understand the complex architecture and went beyond basic journey maps to map out advancement logic and conditional flows.

‍Research & Discovery: How do you research something that doesn’t exist? Knowledge Bases were so new you couldn’t Google them, so I reverse-engineered the problem by researching with 45+ participants, studying their daily frustrations and current workarounds to uncover what we actually needed to build.

From answers to action

The Agent opportunity

The research revealed something bigger. Users didn't just want accurate answers, they wanted action. Instead of just searching "Delta flight prices," they wanted systems that could find AND book flights automatically.

This was exactly what AI Agents could deliver: entity that execute autonomous tasks on user's behalf.
‍
Another team was building Agent technology without UX expertise. I aligned both efforts to maximize customer value. The challenge: Agents needed to launch in 40 days, Knowledge Bases 60 days later.

Strategic Planning & Delivery

Balancing user needs with business constraints

Instead of compromising user needs for timelines, I created a strategic roadmap to deliver Agents in 40 days, Knowledge Bases in 60, integrated for maximum value.

‍4-week sprint:
Week 1: Blueprints across 5 teams
Week 2: 45 flows, 250+ screens designed
Week 3: Leadership alignment (2 GMs, 1 Sr PE, 1 VP)
Week 4: Removed friction, prevented 200+ bugs
‍
‍

Primary persona and there user journey

Click to see detailed view

Design execution

Rapid Design at Scale: 45 Flows in 40 Days

I implemented a focused four-week design process.

Week one established the blueprint through five-team collaboration. Week two delivered 45 main flows and 250+ screens capturing the end-to-end experience.

Week three secured leadership alignment on our integrated RAG-Agent approach through presentations to 2 GMs, 1 Sr PE, and 1 VP. The final week focused on quality, removing CX friction through leadership escalations and eliminating 200+ potential bugs before implementation.

Result: This systematic approach transformed complex requirements into a cohesive console experience, ready for launch in just 40 days.

Impact

Design led impact at scale

Re:Invent keynote

Featured in AWS CEO keynote 2023 - rare recognition for new features

5.2M+ tools

From 6.5K at launch to 5.2M+ applications created by 10K+ customers

$545M revenue

Direct revenue impact from platform adoption

150+ AWS consoles

Design patterns now power 150+ AWS consoles across the platform

Customer praise

Adidas credited improved findability, Leanspace called it "super powerful"

Platform shift

Became blueprint for future GenAI feature development at AWS

Feature walkthrough

SHUBHA GARGAVA

Need more information?