AWS says a six-week program helped four customer-facing professionals with no engineering backgrounds build WealthWise, a multi-agent financial advisory prototype that won first place. The program aimed to close the gap between discussing AI and building with it. In a survey of business professionals who use AI tools but do not write production code, 90 percent wanted hands-on experience building AI agents. Although 80 percent had explored Amazon Bedrock and Amazon Bedrock AgentCore, fewer than 1 in 5 had used Strands Agents SDK or built agents with AWS Lambda.

The four participants built five agents for portfolio analysis, risk assessment, financial planning, market insights, and personalized investment recommendations. WealthWise used a dual-server architecture with Node.js and Python Flask, Amazon Nova models, Strands Agents SDK for coordination and conversation memory, four Amazon DynamoDB tables, and live market data. It returned responses to complex financial reasoning requests in under 5 seconds. Each agent could choose tools and combine them to reason across portfolio data, market databases, and financial planning models.

Participants spent approximately four hours per week for six weeks. AWS says its program data showed that participants who completed a phased program retained three times more practical skills than participants in intensive two-day formats. Teams formed groups of 3–4, selected a real problem, attended hands-on training, then built prototypes with mentors’ help. The building phase allowed 2–3 iteration cycles; participants could build a prototype in week 3, then revise or rebuild it in week 5. Four core team members ran the program alongside their regular jobs.

The program used Kiro, an integrated development environment that scaffolds software architecture from natural-language descriptions; Amazon Bedrock for access to foundation models; Strands Agents SDK for composing agent patterns; and AWS Model Context Protocol servers and AWS Lambda agents to support deployable architectures. Participants demonstrated their prototypes in a judging round covering Business Impact, Technical Excellence, Reusability and Scalability, Innovation, and Presentation Quality.

After the program, 23 percentage points more participants rated themselves confident applying AI to practical use cases than expected at the start. Self-rated “Strong or Expert” understanding of agentic AI rose from 27 percent to 82 percent (+55 pp), while the share who felt “Well or Extremely prepared” to identify AI opportunities rose from 41 percent to 85 percent (+44 pp). Participants with only theoretical or limited experience fell from 34 percent to 0 percent. Strands Agents SDK adoption rose from 20 percent to 80 percent (+60 pp), and AgentCore adoption rose from 39 percent to 85 percent.

During the program, 52 percent identified customers who could benefit from their prototypes, and 87 percent expected to apply what they learned with customers within 30 days. Afterward, 90 percent felt well prepared or better to identify AI opportunities in customer conversations; before the program, 44 percent had cited uncertainty about architecture patterns as a barrier. AWS also reported a 100 percent recommendation rate and said 95 percent of participants found the program met or exceeded expectations. The pilot expanded from one team to regional and global deployment. Prototype ideas included healthcare care coordination, content moderation and triage, fraud prevention, customer churn prediction, autonomous decision-making, and Kiro plus AgentCore demonstrations for customer workshops.

AWS recommends seven program practices: choose tools suited to participants’ experience, use a phased six-week structure, pair participants with mentors, require working prototypes and code repositories, measure skills before and after, document materials for reuse, and establish psychological safety. Participants said they applied lessons to their work within the first two weeks. The playbook, evaluation rubrics, and environment setup guides are available to teams seeking to replicate the program.

AWS says a six-week program helped four customer-facing professionals with no engineering backgrounds build WealthWise, a multi-agent financial advisory prototype that won first place. The program aimed to close the gap between discussing AI and building with it. In a survey of business professionals who use AI tools but do not write production code, 90 percent wanted hands-on experience building AI agents. Although 80 percent had explored Amazon Bedrock and Amazon Bedrock AgentCore, fewer than 1 in 5 had used Strands Agents SDK or built agents with AWS Lambda.

The four participants built five agents for portfolio analysis, risk assessment, financial planning, market insights, and personalized investment recommendations. WealthWise used a dual-server architecture with Node.js and Python Flask, Amazon Nova models, Strands Agents SDK for coordination and conversation memory, four Amazon DynamoDB tables, and live market data. It returned responses to complex financial reasoning requests in under 5 seconds. Each agent could choose tools and combine them to reason across portfolio data, market databases, and financial planning models.

Participants spent approximately four hours per week for six weeks. AWS says its program data showed that participants who completed a phased program retained three times more practical skills than participants in intensive two-day formats. Teams formed groups of 3–4, selected a real problem, attended hands-on training, then built prototypes with mentors’ help. The building phase allowed 2–3 iteration cycles; participants could build a prototype in week 3, then revise or rebuild it in week 5. Four core team members ran the program alongside their regular jobs.

The program used Kiro, an integrated development environment that scaffolds software architecture from natural-language descriptions; Amazon Bedrock for access to foundation models; Strands Agents SDK for composing agent patterns; and AWS Model Context Protocol servers and AWS Lambda agents to support deployable architectures. Participants demonstrated their prototypes in a judging round covering Business Impact, Technical Excellence, Reusability and Scalability, Innovation, and Presentation Quality.

After the program, 23 percentage points more participants rated themselves confident applying AI to practical use cases than expected at the start. Self-rated “Strong or Expert” understanding of agentic AI rose from 27 percent to 82 percent (+55 pp), while the share who felt “Well or Extremely prepared” to identify AI opportunities rose from 41 percent to 85 percent (+44 pp). Participants with only theoretical or limited experience fell from 34 percent to 0 percent. Strands Agents SDK adoption rose from 20 percent to 80 percent (+60 pp), and AgentCore adoption rose from 39 percent to 85 percent.

During the program, 52 percent identified customers who could benefit from their prototypes, and 87 percent expected to apply what they learned with customers within 30 days. Afterward, 90 percent felt well prepared or better to identify AI opportunities in customer conversations; before the program, 44 percent had cited uncertainty about architecture patterns as a barrier. AWS also reported a 100 percent recommendation rate and said 95 percent of participants found the program met or exceeded expectations. The pilot expanded from one team to regional and global deployment. Prototype ideas included healthcare care coordination, content moderation and triage, fraud prevention, customer churn prediction, autonomous decision-making, and Kiro plus AgentCore demonstrations for customer workshops.

AWS recommends seven program practices: choose tools suited to participants’ experience, use a phased six-week structure, pair participants with mentors, require working prototypes and code repositories, measure skills before and after, document materials for reuse, and establish psychological safety. Participants said they applied lessons to their work within the first two weeks. The playbook, evaluation rubrics, and environment setup guides are available to teams seeking to replicate the program.