Financial Datasets
@findatasets
Stock market data for your AI agents.
Dexter now lets you easily select LLMs. Like Claude Code, type /model to choose. Another small step toward a financial agent that feels smooth and personal.
Building a stock market API has been tough. One of the harder parts: currencies. American companies report in USD. Others report in their local currency. Your AI agent does not think about this. If you feed it mixed currencies, the analysis breaks. We just standardized all…
Dexter now works with more LLMs. I’m running evals across all top models. Current support: • openai • gemini • anthropic Want another provider? Send a pull request. Sky is the limit.
I updated our system design for Dexter. This week we gave Dexter memory through the file system. How it works: 1 • Dexter runs tasks 2 • saves tool outputs 3 • passes around output summaries 4 • loads full outputs at the end 5 • generates final answer Before saving each…
The future of finance isn’t closed. It’s open source. Meet Dexter, a deep research agent built in public. It’s crushing evals, improving fast, and every line of code is yours. Finance belongs to everyone.
I added memory to Dexter. Tool outputs now get saved to the file system, and Dexter loads them whenever needed. Key benefits: • better context mgmt • fewer API calls • lower costs All open source. Enjoy.
NVDA filed earnings today. Financials were on @findatasets in seconds. They're still not live on: • Yahoo Finance • Google Finance • MarketWatch • and many others Most platforms wait on legacy data pipelines. We parse filings the moment they hit EDGAR.
Dexter now runs 73% faster. Key changes: • optimized tool use • reduced loops • added output streaming Small tweaks with a huge payoff. Feels so much better to use now.
3 months in and I finally cracked it. Parsing SEC filings for banks. Banks follow the most complex reporting rules and their numbers look nothing like other industries. You run into items like: • time deposits • sub advisory fees • trading liabilities • commercial paper…
Introducing evals in Dexter. We can now see how it performs across a full set of LLM metrics. This includes: • correctness • costs • latency • and more LangSmith from @LangChainAI gives a live view into every run. Very cool to watch in real-time.
I love using Dexter for earnings research. “Walk me through Meta’s latest earnings.” How Dexter works: • pulls latest filings • reads through them • looks for insights and surprises • tells me what I need to know Runs in 20 seconds. Saves me hours.
We hit 23,421 stocks in our Fundamentals API Quick stats: • 1B+ rows • 200 ms latency • history back to 1994 Every SEC filing ingested in under a second. Legacy vendors, start running. We’re coming.
GPT-5.1 is now live in our AI hedge fund! First impressions: • super fast • great value $ • excellent reasoning A huge leap forward.
I ran a ton of Dexter evals last night. Now I know what needs work. Biggest opportunities: • smarter tool calls • better context chaining • cleaner outputs Starting fixes tonight. All code is public and shared - let's build.
Markets reward those who act in real time. This earnings season, @findatasets handled thousands of SEC filings instantly. Highlights: • 4,500+ filings processed • under 1 second each • fully standardized While legacy platforms took days for updates, our users had the data…
I’m adding evals to Dexter this week. We just hooked up LangSmith and can now see every LLM call in detail. This includes: • costs • traces • latency • evaluations Since we’re using @LangChainAI, we get all of this for free. No code changes needed.
The next big win in agents won’t come from bigger LLMs. It’ll come from smarter context management. Agents break down when tool outputs flood the context window in multi-step flows. Fix: move tool use out of context window and into code execution. Agent writes and runs code…
Dexter now breaks down company revenues. By product lines, business segments, geographies, customers, and more. How it works: • fetches raw XBRL data • extracts segments • analyzes patterns • keeps going until done Lets you see exactly how a company makes money. Less…
I just built a new Stock Screener API. Lets you filter 20,000 stocks in < 1 second. Powered by data from: - income statements - balance sheets - cash flow statements - key ratios It’s designed for LLMs and agentic tool use. Live now on @findatasets - soon in Dexter.
Dexter just got a lot smarter. After every task, Dexter now checks if the overall goal is already achieved. While working: • reflects on progress • stops when goal is hit • generates final answer Next: adding more frontier LLMs. We’re using @LangChainAI to make this easy.…
United States الاتجاهات
- 1. Black Friday 311K posts
- 2. #releafcannabis N/A
- 3. Good Friday 50.5K posts
- 4. #FanCashDropPromotion N/A
- 5. #DaesangForJin 55K posts
- 6. #ENHYPEN 208K posts
- 7. #AVenezuelaNoLaTocaNadie 1,040 posts
- 8. ARMY Protect The 8thDaesang 2,980 posts
- 9. Victory Friday N/A
- 10. 2025 MAMA Awards 457K posts
- 11. Third World Countries 42.4K posts
- 12. yeonjun 67.8K posts
- 13. Mnet 211K posts
- 14. Cyber Monday 5,245 posts
- 15. Sarah Beckstrom 264K posts
- 16. Lamar 48.9K posts
- 17. Signing Day 1,169 posts
- 18. TODAY ONLY 63.4K posts
- 19. Mr. President 20.7K posts
- 20. Stormzy 2,079 posts
Something went wrong.
Something went wrong.