Open Source AI Agents: From Research to Outreach

🎥 Prefer watching instead? Check out the full YouTube video →

Last month I released Mira, an open-source agentic AI system for company data enrichment. Since then, Mira has grown with powerful new features like configurable agents, bulk enrichment, and personalized outreach.

All of this is happening in the open on GitHub under the MIT license.
👉 Explore Mira on GitHub

Agents: The Core of Mira

Everything in Mira starts with agents. An agent is a workflow designed for a specific purpose — like research, evaluation, or outreach. Each one has its own configuration where you define:

• What data points to collect
• Which sources to use (Website, LinkedIn, Google Search)
• How to analyze results
• Whether to generate personalized outreach

This flexibility means you can tailor agents to your exact workflow.

Smarter Website Exploration

When Website Explore is enabled, Mira doesn't just scrape the homepage. It looks at the pages most relevant to your data points.

• Looking for press mentions? It checks /press or /news.
• Interested in hiring activity? It checks /careers or /jobs.

Mira goes where the signal is strongest.

Multi-Source Data Collection

• LinkedIn: Gathers structured info from company profiles and posts.
• Google Search: If a specific data point like recent funding is requested, the system automatically generates a tailored search query (e.g., “[Company name] latest funding round”) and extracts relevant details from the top sources.

Together, these sources give a broader, more reliable view of a company.

Analysis & Criteria

Beyond collecting data, Mira can evaluate companies against custom criteria.

For example:

• Companies that raised funding recently
• Companies in specific industries
• Companies meeting your ICP

Mira assigns a fit score with reasoning so you can filter and focus on the right targets.

Outreach Generation

Once the research is done, Mira can draft personalized LinkedIn messages and emails based on your instructions. This turns raw research into actionable outreach.

Running Agents in Practice

You can run agents on a single company or on a bulk list. Here's how they work in practice:

Research Agent

• Collects structured company profiles with sources and confidence scores.
• Useful for enriching CRMs or doing background research before meetings.

Evaluation Agent

• Checks companies against your criteria.
• Produces a fit score + explanation for each company.
• Ideal for turning a long list into a targeted shortlist.

Outreach Agent

• Runs on a list of companies.
• Generates LinkedIn and email drafts tailored to your goals.
• Outputs a CSV with both research and outreach, ready to use in your tools.

Why This Matters

These updates make Mira not just a research tool, but a complete workflow engine. From enrichment, to filtering, to outreach, you can now move seamlessly from data to action.

And since it's open source and customizable, you can create your own agents to match your business workflows.