Research is a key part for anyone working in Sales. No matter if you work in Outbound or Inbound teams, or as an Account Executive, having a deep and informed view of your prospective account or lead is essential.

Research is time consuming. LLMs have helped a lot to speed up this research and get a much more comprehensive view on a company or person in a much faster time frame. You can achieve a lot by using the Deep Research capabilities of e.g. ChatGPT, Claude or Gemini.

While these new capabilities are a big step forward, they also fall short of what is often needed to thoroughly prepare for an upcoming Sales meeting:

  • What are a person’s posts and public statements about a particular topic?
  • Where was a company or a person mentioned in articles, videos or posts of industry associations, events or communities?
  • What does a company publish in its regulatory reports or statements?

Deep Research functionalities of the standard LLM providers only capture this in parts and are always a black box. You can use LLM-native search engines to significantly improve your results and control what you search for.

This can make the difference between a meeting where you have highly specific upfront information about your prospect and the typical “let’s see what they say” meeting.

How to get comprehensive and cheap research reports

The key to better research reports is to equip your LLM with enhanced search capabilities, using LLM-native search engines. These search engines are able to understand your intent and context, leading to far better results than just using typical Google-like searches.

Two search engines are state-of-the-art for this type of work:

  • Exa
  • Parallel

Connecting these tools to your LLM is easy. There are two ways:

  • Connect to their MCP
  • Connect to them via API

I can recommend this article to understand more about what an MCP is.

In simple terms, think of an MCP as a tool for your LLM to speak to another tool. It allows your LLM to use Exa or Parallel.

An API is also a way to communicate with these tools. Try out asking your LLM what an API is and to explain it to you in simple terms, it works pretty well!

If you need help setting this up, let me know!

Things which become possible using these search tools

Here are a couple of examples of what you can do once you have these capabilities connected:

  • Crawling personal blogs, podcast appearances, and interview quotes
  • Community & Association Mentions: Finding what industry peers, conference agendas, or niche communities say about the target account
  • Pulling annual reports, regulatory filings, or strategic press releases to identify budget allocation and pain points.

Best practices on how to use these search engines

For starters, the best way to use these search engines is by using your LLM of choice. You don’t go “into the tool” in Exa or Parallel, you connect your LLM to it and have it use them. To not have your LLM start from scratch on how to best use these tools, here are some learnings to steer your prompts:

Say what kind of category you want to search for:

Are you looking for news, blog posts, PDFs or personal sites? Define this to get better results and remove junk.

Set a date window:

When specifically looking for news, define a date window of news you think is relevant. This avoids surfacing news which are “old”.

They are CHEAP, so iterate!

Even search queries which are comprehensive only cost a few cents. So don’t get frustrated when the first times you use these tools, you don’t get what you hoped for. Brainstorm with your LLM how to change the queries and try again.

FAQ

What makes Exa and Parallel different from standard Google search?

Google looks for exact keyword matches, which usually brings up generic marketing pages optimized for SEO. Exa and Parallel look for meaning and context. Instead of matching words, they understand what you are actually trying to find and bring back clear, useful details for your AI.

Do I need to know how to code to use this?

No. While developers use these tools, sales reps can easily set them up in everyday tools. You can connect them to ChatGPT Custom GPTs, Claude Projects, or no-code platforms.

Is my prospect data safe when using these tools?

Yes, but practice good data security. You are using these tools to pull public information in, not push private company data out. Just stick to searching public company names, executive names, topics, and website URLs.

How expensive is this compared to standard sales tools?

It is extremely cheap. A single search query usually costs a few cents. Even if you run a deep, complex research, costs remain very low.

What should I do if the search results aren’t helpful?

Iterate! If your results are off-target, try narrowing the date range or picking a specific category (like “news” or “company filings”). You can even copy the bad results, show them to your LLM, and ask: “Why did we get this junk, and how should we change our search terms to fix it?”