Stop Manually Researching Companies: Let AI Turn Email Lists Into Full Profiles

You've got a spreadsheet full of email addresses. Maybe it's a lead list, a batch of signups, or a conference attendee export. What you don't have is context—who these people work for, what those companies do, how big they are, or whether they've raised money. Manually researching each one is the kind of task that eats an entire afternoon. Fire Enrich is an open-source tool that takes that list and does the digging for you, turning bare emails into enriched company profiles powered by Firecrawl and a multi-agent AI system.

What It Does

Fire Enrich accepts a list of emails and returns a structured dataset with company details attached. Given something like [email protected], it produces a record containing the company name, industry, employee count, founding year, headquarters, funding stage, total raised, website, and—importantly—the sources it pulled from.

The tool is built on three technologies: Firecrawl handles web scraping and content aggregation, OpenAI handles the extraction and synthesis of information, and Next.js 15 provides the frontend and App Router framework. It runs as a web app you deploy yourself, either locally or on Vercel.

The interesting part is the architecture. Rather than sending one big query to a model and hoping for the best, Fire Enrich uses a multi-agent orchestrator that runs agents in a specific sequence, with each phase building on what the previous one found. It starts by extracting the domain from the email—recognizing that [email protected] is a corporate address. A Discovery Agent then runs parallel searches to identify the company and its basic details. Once it has a company name, a Company Profile Agent uses that name to search for industry classification and market positioning. A Financial agent follows, and so on through the pipeline.

Why It's Cool

The agents actually depend on each other. This isn't a cosmetic "multi-agent" label. The README walks through the exact flow: the domain extraction feeds the Discovery Agent, whose findings about the company name feed the Company Profile Agent, whose industry classification feeds the Financial agent. Each phase narrows the search space for the next one. That's a sensible design—it means later agents aren't guessing blindly, they're working from established facts.

It shows its sources. The enriched output includes a sources array with the actual URLs the data came from. For the Wiz example, that's the company's about page and a TechCrunch funding article. In a world where AI-generated data is often a black box, being able to verify where a "Series D" or "$900M raised" claim originated is genuinely useful.

Parallel searches within each phase. The Discovery Agent runs three concurrent Firecrawl API calls (searching for the company name, the domain, and a "what is" query). The Company Profile Agent does the same with industry-specific queries. This is a practical way to get broader coverage without serializing every request.

The output is immediately usable. The before-and-after example is clean: one email in, a structured JSON object out. Fields like employeeCount come back in ranges (1001-5000), which is honest about the precision you can realistically get from public web data.

Deployment is straightforward. There's a Vercel deploy button, and the whole thing needs just two API keys—Firecrawl and OpenAI. No database, no complex infrastructure to stand up.

How to Try It

Setup is minimal. You'll need API keys from both services first.

1. Get a Firecrawl key at firecrawl.dev/app/api-keys
2. Get an OpenAI key at platform.openai.com/api-keys
3. Clone the repository and create a .env.local file:

FIRECRAWL_API_KEY=your_firecrawl_key
OPENAI_API_KEY=your_openai_key

1. Install dependencies and start the dev server:

npm install
npm run dev

Or with Yarn:

yarn install
yarn dev

1. Open http://localhost:3000

If you'd rather skip the local setup, the README includes a one-click Vercel deploy button that prompts you for the same two environment variables.

The full source and instructions live at github.com/firecrawl/fire-enrich.

Final Thoughts

Fire Enrich is best suited for developers and technically comfortable operators who regularly work with lead lists or contact datasets and want to automate the research step. It's not a hosted product—you're running it yourself, paying for your own Firecrawl and OpenAI usage, and accepting the accuracy limits of web-sourced data. But the multi-agent pipeline and the source citations make it more trustworthy than a single-shot LLM prompt, and the setup cost is low enough that it's worth trying on a real list to see how it performs. If your workflow involves turning emails into context, this is a solid starting point you can fork and adapt.