Inspiration
Investment research often focuses on an asset’s performance while overlooking the person’s goals. The same investment can mean different things to someone saving for retirement and someone who needs the money in six months.
We built AlphaForge AI to connect investment ideas with the goals, constraints, and evidence that matter.
How we built it
Users describe their investment idea in plain language. AlphaForge identifies relevant questions, retrieves public financial data, and uses analytical tools to investigate historical performance, risks, and trade-offs.
We built the interface with Streamlit and used Amazon Bedrock for understanding requests, planning research, and explaining results. Financial calculations run in Python, with sources and limitations shown alongside the findings.
Challenges we faced
Our biggest challenge was making answers useful and well supported. Correct numbers do not always lead to sound conclusions.
We worked on linking claims to evidence, handling follow-up corrections, and making incomplete analysis visible. We also balanced deeper investigation against response time.
What we learned
More agents and longer answers do not automatically produce better research. Clear responsibilities, reliable tools, and meaningful validation matter more.
AlphaForge remains a research prototype. Our next steps are stronger portfolio analysis, broader data coverage, and more testing on unfamiliar questions.
Built With
- agents
- ai
- altair
- amazon
- amazon-web-services
- bedrock
- generative
- multi-agent
- python
- strands
- streamlit