DialReady

Every sales call opens with the same prep: understanding the company, the person, and what they care about before you dial. Done well, it's 20–30 minutes of scattered research - a 10-K here, a news item there, a LinkedIn profile in between - repeated before every meeting and never reused. It's exactly the kind of repeatable work worth automating once.

Instead of a research workflow, DialReady is a single form. The system resolves whatever you give it: a ticker becomes a company, a LinkedIn slug becomes a person, a person's employer becomes a company. It fetches the SEC filing, runs live web searches, and scrapes the public profile - then asks the LLM to synthesize everything into one structured brief.

The brief is honest about what it doesn't know. Every section cites its sources, and a gaps list says what couldn't be verified, so nothing gets invented to fill space.

Next.js on Vercel serves the UI and the API route. The pipeline runs in parallel - SEC filings from EDGAR, search results from Tavily, profile data fetched server-side - then truncates each source to fit one LLM call. The LLM runs on NVIDIA's free NIM endpoint, which keeps the running cost at zero, and a free-tier fallback model catches rate limits.

Zod validates the output against the schema, and every section degrades gracefully if a source fails. The whole thing runs within a 60-second function budget, so it ships on a free Vercel tier with no database and no auth.

What took 20–30 minutes of tab-hopping now takes one paste and under a minute of waiting. The brief fits on one screen: company position, financials, recent moves, the person's background, icebreakers, and a suggested approach - with sources attached so you can verify every claim before the call.