Dharmesh drops an AI company research agent

Researching a new prospect or startup usually means opening a dozen tabs: LinkedIn for execs, traffic estimators for growth, review sites for sentiment, and a spreadsheet to dump it all.

HubSpot co-founder Dharmesh Shah decided to automate that grind. On November 11, he launched Company Research Agent at companyresearch.ai, a standalone agent designed to condense deep company reconnaissance into a clean, single-page dossier.

One prompt instead of ten open tabs

Dharmesh initially built the agent for his own deal screening, but quickly realized it directly serves sales reps, business development teams, and customer success managers.

The tool pulls from multiple web channels alongside several proprietary and paid data sources. Reports include firmographics, top executives, website traffic, search marketing data, and G2 product reviews. For publicly traded companies, it even pulls financial metrics like revenue multiples and revenue per employee.

Users can customize the dossier by reordering or stripping out sections, and they can prompt the agent with specific follow-up questions. Generated reports can be shared via public links that require no login, or pushed straight into HubSpot. Dharmesh noted he is already building a Smart Note feature to automatically attach custom summaries directly onto HubSpot company records.

Why it matters

Sales intelligence is one of the clearest, immediate ROIs for agentic software, but most tools are either bloated enterprise databases or basic scrapers that miss the mark.

What makes this project clever is its focus on workflow integration. Instead of trapping insights inside another siloed chat window, Dharmesh is bridging third-party paid signals, G2 reviews, and existing CRM records into one fluid output.

The friction here is data depth: for early-stage or stealth startups, public and commercial footprints are sparse, meaning hallucination or shallow outputs remain a risk. Still, for anyone running outbound sales or market research, it is worth testing right now simply for the time it saves on pre-meeting homework.