Hi HN! We’re Bolu and Noah from AnswerGrid (https://answergrid.ai). We’re building an AI-powered web research tool in the form of a spreadsheet. The problem we’re starting with is helping founders who sell to businesses discover the most relevant leads worth the investment of manual outbound (that is, worth putting in the effort to write a personalized outreach message to see if they might be interested).
There’s a demo video here: https://www.loom.com/share/fe4e40fa000b4406910a9ce247079138?..., and you can try the actual product at https://app.answergrid.ai/try-it (no signup needed!)
One of the biggest problems B2B founders have, once they’ve built an initial product, is finding initial companies to sell to. There’s a chicken-and-egg problem here: most companies don’t want to buy from a startup until it’s much further along, but the startup won’t get that far unless they find some customers. Locating those first companies who need what you’re building badly enough that they’ll go with you at an early stage is time-consuming and painful - you’re searching a haystack for a few needles. This is the problem we’re building software to help solve.
We found that for effective lead qualification—that is, narrowing down the haystack of potential companies to a few highly relevant ones—you can’t just search on keywords or industry categories like "AI," "Hardware," or "Healthcare" on other lead generation indexes. Those methods are too blunt to find adopters for an early product that almost necessarily has a narrow initial appeal.
Instead, you’d need to evaluate companies using many loose heuristics. For example, one of our early customers qualifies leads with questions like, "Do they offer subscription or usage-based pricing?" Most existing lead generation tools have forced founders to choose between thoughtful qualification research and increasing the reach of their high-quality outbound.
We found a few existing tools to wrangle CSV exports with additional AI qualifications, but nothing seemed built to help first-time founders like ourselves find the first couple of hyper-relevant companies to sell to. Their feature sets and advertised use cases (like "AI-written personalized messages") seemed optimized for scaling the go-to-market of a mature product—a completely different situation. This led us to focus on making lead gen intuitive for other builders selling for the first time, looking to find their early users.
We’ve built a tool to help founders codify the research heuristics they use to qualify leads that justify a time-consuming investment in human-written outreach. (Side note: while we’re using AI to help qualify leads, we emphatically do not want to help scale AI spam. Our tool aims to help you discover potential users worth writing to manually.)
Here’s how it works: Starting from lead-gen sources like Crunchbase, Apollo, and LinkedIn as a data foundation, we find leads that match a simple high-level company profile: “Series A Biotech startups in the US.” You can then further qualify leads using AI-powered tools such as (1) “web scrape,” which, when given a target URL column, spawns concurrent jobs to scrape (text and screenshots) the target sites of all companies in the table and returns LLM evaluations based on a prompt; and (2) “web search,” which returns answers and citations to natural language questions using information from the web.
These tools afford a greater level of expressiveness than simple keyword searches. We aspire for the tool to match the thoroughness a founder would apply manually on any given lead. But this time, help them do it for thousands of leads at a go.
Once you’ve narrowed prospects down with these granular qualifications, you can then use job descriptions to find the contact information of your potential customer champions.
While we’re excited about other web research use cases, we started with sales qualification because the research outcome is immediately actionable—" Who should I sell to today?"—and the feedback cycles are short—"Was that lead relevant?”.
We are exploring exposing our web research infrastructure through an API to other developers who'd find it helpful to enable web research workflows in their apps. Please reach out (https://tinyurl.com/AnswerGrid-15) if you’re working on a product that might benefit from this.
You can try the tool here: https://app.answergrid.ai/try-it (no signup needed)
We’re excited to hear about any lead qualification tricks you’ve discovered for identifying your early users or what other use cases you think are relevant here. Even if you’re not looking to do lead generation, any product feedback from playing with the demo would still be most appreciated.
* A Note on how the Search box works: It tries to infer the starting Crunchbase filter configs based on your query. If it doesn’t get the configs precisely, you can edit the inferred filters at the top left corner once the grid is loaded, or try starting with a more straightforward query before adding extra LLM qualification columns.
by Andrey
Azimov