Viktor.comAn AI coworker that actually does the work
AlphAI
Financial news for AI agents and trading bots
7 followers
Financial news for AI agents and trading bots
7 followers
Market-moving financial news, structured for AI agents and trading automations. Live REST API and MCP server for Claude Desktop, Cursor, and your own agents.
Hey PH 👋
AlphaAI started out as an "AI stock analysis" site. You picked a ticker, we pulled prices and indicators plus recent news, ran a stack of prompts over it and handed back an analyst-style report. It demoed well. We paid for the data, nobody paid us. Turns out people don't buy prompt output. Anyone who cares about the analysis already has ChatGPT or Claude and their own prompts.
What they actually used, quietly, was the news feed the analysis was sitting on. A few were piping it straight into their own agents. So I killed the analysis product and rebuilt around the part that got used.
Now it does one thing: financial news that shows up already machine-readable. Collectors pull from GDELT, SEC EDGAR filings and publisher RSS, and every story gets per-ticker impact analysis, a category and a 1-10 relevance score. We look at ~12k articles a day and fewer than 1 in 5 make the feed. A SEC Form 4 insider filing turns into a scored news item about 6 minutes (median) after it lands on EDGAR https://alphai.io/insider-trading-tracker.
You read it as a REST API or an MCP server (https://alphai.io/mcp), so Claude Desktop, Cursor or any MCP client talks to it with no glue code. The free tier is real: 20 req/min and 100 req/day on both surfaces, no card. There's a live playground with no key and no signup: https://alphai.io/developers
One honest caveat: no price data. We used to carry quotes and even 15-minute delayed data was one of our biggest bills, so we dropped it. Your broker does that better anyway. This is purely the news side, and the relevance score exists because most of what we scan on any given day is junk.
If you wire news into agents, I'd genuinely like to know what metadata you're missing and whether a tool surface like this matches what your agent actually asks for. Happy to get into the pipeline, the scoring, or what it's like to kill the product you named the company after.
How does the data get structured so fast, and do you cover earnings whispers or only formal announcements?
@zilansalm5w7b Hi Zilan!
Two paths: SEC filings (Form 4, 8-K) get parsed from EDGAR into structured rows the instant they're filed, no model needed in fact. General news comes from GDELT, and a background worker enriches each article as it arrives: tickers, category, a per-ticker take, and a 1–10 relevance score. It runs at ingest, not on query, so the feed's already structured when you hit it.
On whispers: tricky question! We cover formal announcements as first-class (SEC), plus the news/analyst chatter around earnings tagged under earnings, BUT we're not a proprietary "whisper EPS" consensus service. It's the news and analysis around earnings, not a crowd-sourced whisper number.
Thanks for the right questions!