Feature · Research Agent

Every Email Starts With
Real Research. Not Guesses.

Most cold email tools write from a contact record: name, job title, company. That's not enough to write a relevant email. SendState's Research Agent reads the prospect's website and LinkedIn before generating a word, so every first email reflects something true about who they are right now.

3

Sources scraped per prospect

3

Personalized openers generated

1x

Research runs once, reused automatically

The problem with merge fields

Merge fields feel like personalization. They aren't. "Hi {{first_name}}, I noticed {{company_name}} is in the {{industry}} space" is still a template, just one with variables. Prospects can tell.

Genuine personalization means knowing something real about the person's current situation. What their company actually does. What problems they're visibly dealing with. Whether there are signals right now (a new hire, a funding round, a job listing) that make the outreach timely rather than generic.

Without research

"Hi Sarah, I noticed Acme Corp is in the SaaS space..."

  • · Generic angle, no context
  • · Sounds like everyone else
  • · 1.1% average reply rate

With Research Agent

"I saw you're hiring two AEs while the sales motion is still founder-led, that's a specific timing problem..."

  • · Angle built from live signals
  • · Reads like the sender did their homework
  • · 6.2% average reply rate

How it works

Three phases. One brief. Before send.

1

Live data collection

The Research Agent fetches and reads the prospect's company website, LinkedIn profile, and X/Twitter if provided. It handles different page structures automatically and extracts meaningful text, not raw HTML noise.

Company website · LinkedIn · X / Twitter

2

Signal extraction

AI analyzes the scraped content to build a structured brief: what the company does, who they sell to, their current stage, visible pain points, opportunities, and buying signals (hiring activity, expansion indicators, "pipeline now" language).

Pain points · Buying intent · Fit signals · Tone & style · Tech stack

3

Brief + openers ready

The orchestrator assembles everything into a single research result: company summary, ICP, signals, recommended tone, confidence score, and three personalized opening lines grounded in what was actually found. The email writer uses this brief as its starting point.

Company brief · 3 openers · Confidence score · Sources logged

Research runs once per prospect and is stored. If research already exists and meets quality threshold, it's reused no extra cost, no extra latency.

What the writer actually receives

Every email generation call includes a fully structured brief, not a name and job title.

RESEARCH BRIEF prospect@company.com
COMPANY Acme Corp, B2B SaaS, Series A, 45 employees
OFFER Project management for remote engineering teams
ICP Engineering leads, CTOs at 20-200 person companies
STAGE Scaling post-funding, active hiring
PAIN POINTS Manual sprint tracking, async coordination gaps
OPPORTUNITIES Hiring 3 engineers → team coordination pressure increasing
BUYING SIGNALS Job listing: "Head of Engineering Ops" posted 8 days ago
FIT SCORE 87 / 100
SUGGESTED OPENERS
→ "I saw you're hiring for Head of Engineering Ops: that role usually appears when coordination costs are starting to hurt..."
→ "Acme just closed Series A and is scaling the eng team fast. At that stage, sprint tracking usually breaks before tooling catches up..."
→ "As CTO at a 45-person company post-funding, you're probably feeling the point where your current setup stops scaling..."
TONE Direct, peer-level, no fluff
SOURCES website · linkedin (confidence: 91%)

This brief is what the email writer uses, not {{company_name}}.

What research-backed emails actually do to reply rates

6.2%
Average reply rate with Research Agent
vs 1.1% on static sequences
-30%
Fewer objections per campaign
More relevant emails generate fewer "not right now" replies
0
Manual research hours
Research runs automatically before every campaign launch

How other tools handle research

Most tools skip it entirely. A few have it as an add-on. SendState runs it by default on every prospect before any email is written.

SendState
Clay
Apollo
Smartlead / Instantly
Research before email generation
Live signal detection
Personalized openers generated
Research reused automatically
Part of sending platform

Clay is a powerful research tool for teams who want to build custom enrichment pipelines. SendState is for teams who want research to happen automatically as part of sending: no pipeline to maintain.

Common questions

Yes. When you import prospects and launch a campaign, the Research Agent runs on each contact before generating emails. You don't trigger it manually, it's part of the send pipeline. If research for a prospect already exists and meets the quality threshold, it's reused at no extra cost.

The Research Agent fetches the prospect's company website, LinkedIn profile (if a LinkedIn URL is provided), and X/Twitter profile (if provided). It handles different page structures automatically and extracts meaningful company and signal data from each source.

The Research Agent returns a confidence score alongside the brief. If the score is low, for example, because a website has minimal content or a LinkedIn profile is sparse, the email writer falls back to safer, less signal-dependent angles rather than fabricating context. You can see the confidence score and sources in the prospect's research record.

Not exactly. Clay is a powerful standalone enrichment and research platform built for teams who want to build custom data pipelines with many sources. SendState's Research Agent is built for teams who want prospect research to happen automatically as part of the send workflow: no separate tool, no pipeline to maintain, no extra cost per enrichment row. If you're already using Clay, you can import Clay-enriched lists into SendState.

Apollo enriches contacts from its database: firmographic data, contact info, and some intent signals. SendState's Research Agent scrapes live data from each prospect's actual website and LinkedIn at the time of the campaign, which means it picks up things Apollo's database doesn't have: recent job listings, current messaging, expansion signals, and tone cues from how the company actually talks about itself right now.

See what your prospects are actually signaling.

Every contact you import gets a full research brief before a single email is written. No setup. No extra tools. No manual research.

Start free 3-day trial, no credit card

No credit card · Cancel anytime · Full feature access

AI Research Agent: Prospect Intelligence Before Every Send | SendState