Launching today
Vester
Claude Code for Investing. Level up your Portfolio
13 followers
Claude Code for Investing. Level up your Portfolio
13 followers
Your personal army of investment analysts. With 12+ live data feeds and 50+ curated finance skills, Vester is a highly autonomous agent built to make "smart-money" investing 10x easier. Covers equities, crypto, and macro with live data from prediction markets, government filings, economic indicators, on-chain analytics, sentiment feeds, and more. Every number sourced and traceable. Beat the major AI labs head to head on the vals.ai Finance Agent benchmark public set. Totally free to start.
Hey Product Hunt,
I am Alex, Co-founder & CEO of Vester here, and Iām pumped to share our first public launch with you.
Vester is designed to sort through noisy markets, complicated datasets and conflicting facts - find ground truth for investors. Our team is at war with information overload.
Somewhere early on, I realized the incredible opportunity that being a great investor presented, and I graduated college obsessively following markets.
While endless hours of analysis and a never-ending pulse on markets served me well as fresh grad, 45 hours per week spent on podcasts and news (I have screenshots) is not a practical default.
Meanwhile, the velocity of information is only speeding up. It feels like decades happen in weeks. Naturally it ends up overwhelming most investors.
For the serious investor, Vester is designed to be a force multiplier. The portfolio command center that I have always wanted. Vester shows you what matters, when it matters.
For the casual investor, Vester is an equalizer. No more flying blind.
Whichever one you are - we hope you will give Vester a try, and share your feedback!
Thank you for your support!
Alex
Hey PH š
Lead AI Engineer at Vester here.
Happy to answer technical questions, but wanted to share a bit of what went into building this.
The core challenge with financial AI isn't the data ā it's trust. A model that confidently returns the wrong price or hallucinates a metric is actively dangerous. So most of my work went into grounding: every response traces back to a live data source, and every claim in the output is cited with a clickable reference to the exact row, document, or API response it came from. Not a footnote ā the actual source.
The second hard problem was context. Investment conversations aren't one-shot queries. They're multi-turn, where question 8 depends on what was established in question 3. Most general-purpose models degrade badly here. Getting coherent reasoning across a long conversation without drift took more iteration than anything else.
On the Vals.ai Finance Benchmark ā public, head-to-head against GPT, Claude, and Gemini ā Vester scored 72%, more than any frontier model. That result is meaningful to me because I know exactly what we measured to get there.
Would love feedback from anyone stress-testing it on complex crypto or macro questions ā that's where it's built to shine.