Research Target Accounts With an AI Agent, Lusha and lemlist
By Joris van Huët
Enterprise Interim CMO & Marketing Leader · 15+ years · 50+ orgs
Published
2026-10-26
The short answer: if the first email to a named account needs a reason that is true this quarter, have an n8n agent read Lusha's dated company signals, propose one angle with its source, and park each lead in lemlist's To launch view, where a person approves the angle before anything sends. On monthly billing that costs $162.90 a month: Lusha Pro at $69.90, lemlist's Email plan at $69 and n8n Starter at $24, on the Lusha, lemlist and n8n pricing pages, checked on 26 October 2026. For example, at 12 minutes of hand research per account, 100 accounts take 100 x 12 = 1,200 minutes a month; the reviewer's job shrinks to reading one sentence and opening one link.
Disclosure: I may earn a commission through some links on this site. It does not change what this page says.
Account research before outreach fails in two ways. There is none, or there is a fact that was true last year. Lusha's company signals carry a date, and its news events carry the article's own link, per the company signals reference. That is what makes an agent safe here: it does not need to know anything about the account, only to read, cite and propose. A person decides.
Before you build:
- The credit question, answered on ten accounts. Lusha's API counts each signal type in a request toward credit usage and leaves the rates to your plan, per its signals overview and API introduction. For example, if each type cost 1 credit per company, 100 accounts x 3 types would be 300 credits, half of Pro's 600 a month on the pricing page. Run ten accounts first and read
creditChargedin each response before you trust that sum. - What you already have. The account list with one contact per account, a lemlist campaign whose first email opens with
{{angle}}, a Lusha API key, and n8n 1.88.0 or later, which Lusha's MCP setup page requires for its n8n route. - An angle policy in one paragraph. Which events count for your offer (a funding round, a hiring surge in your buyer's department, an executive hire, a product launch), how old is too old, and the exact words the agent writes when nothing qualifies: no angle.
- A baseline. Positive replies per 100 contacts on your current first email, from the last campaign.
The workflow, account by account
- Trigger. A Google Sheets Trigger on Row added passes the account, its domain and the contact's first name and email.
- The agent reads the signals. An AI Agent node takes the angle policy as its system message. Its one tool is an MCP Client Tool connected to Lusha: HTTP Streamable, URL
https://mcp.lusha.com, Header Auth withx-api-key(Lusha MCP setup). Under Tools to Include, choose Selected and tick only the signals tools (MCP Client Tool), so the agent can neither enrich contacts nor export anything. Signals cover the last 6 months unless the request sets another start date, and a news event arrives with a summary, a publication date and the article's URL (company signals). - It proposes, with evidence. Through a Structured Output Parser, the agent returns four fields: the angle in one sentence, the fact behind it, its date and its source URL. Or it returns no angle. The policy forbids any fact without a URL.
- The lead goes to lemlist and waits. An HTTP Request node using n8n's Lemlist credential (n8n Lemlist node) calls Create Lead in Campaign with
deduplicate=trueand putsangle,angle_factandangle_sourcein the body; lemlist stores any extra key as a custom variable, usable as{{angle}}. SetupdateStrategytofillEmptyOnly: under the default,overwrite, an empty value clears what lemlist already holds for that contact. Rows marked no angle go to a sheet tab instead. - A person checks the angle. With auto-launch off, each lead waits in To launch for manual review; lemlist's auto-launch article treats the two modes as mutually exclusive. The reviewer previews the email with the angle filled in, opens the source, and then launches the lead, rewrites that step for this one person, or removes the lead, as the campaign guide describes. Nothing sends before that.
- The record. The sheet logs one row per account: approved as written, rewritten, or removed.
Company events are about the account. Contact signals such as a promotion or a job change are about a person, and GDPR Article 21(2) extends the right to object to profiling related to direct marketing, while Article 21(4) requires that right to be brought to the person's attention at the latest at the first communication (GDPR). Keep angles at company level unless your first email covers the rest.
The free version, with a sheet and a browser
Take 20 accounts. For each one, open the company's newsroom and careers page, and write one dated fact, its link and a one-sentence angle in the sheet. That is the agent's job done by hand, and it tells you whether your angle policy produces anything worth sending.
Lusha's Free plan, $0 with 40 credits a month, gives you spot lookups, and its comparison table marks the MCP server on every plan; signals, full API access and webhooks appear among what Pro adds (pricing page, checked on 26 October 2026). lemlist's 14-day trial runs the Multichannel features except in-app calls, with 200 credits (lemlist pricing). Its help centre lets custom variables come in during a CSV import (import article), so put the angle in a column, import, and review in To launch exactly as above.
There is also an in-tool middle route. lemlist's AI Variables write a column per lead from a prompt, with optional web search and AI company context, paid with lemlist credits or a supported AI provider (AI Variables). The same article warns that Generate all rows overwrites existing data in that column.
What the paid tools add
Lusha Pro adds the signals themselves: funding and other news with links, hiring surges by department and location, headcount trends over 1 to 12 months, all listed in the signals overview. n8n adds one run per new account, the same policy applied every time, and a log. lemlist's Email plan carries the API and an MCP server for unlimited users (lemlist pricing).
| Piece | Free route | Paid route, monthly billing | Source |
|---|---|---|---|
| Lusha | Free: $0, 40 credits a month | Pro: $69.90, 600 credits a month, API, signals, webhooks | Lusha pricing |
| lemlist | 14-day trial, 200 credits | Email: $69, 50,000 emails a month, API and MCP | lemlist pricing |
| n8n | Community Edition on your own server | Starter: $24 a month for 2,500 executions | n8n pricing |
| Model behind the agent | A key from your AI provider | n8n Cloud Gateway credits, or that same key | Gateway credits |
All checked on 26 October 2026. On yearly billing the same pages show Lusha Pro at $52.45, lemlist Email at $55 and n8n Starter at $20 a month. If the sequence also needs LinkedIn steps, lemlist's Multichannel plan is $109 per user a month on monthly billing (lemlist pricing).
One request option deserves a sentence in the agent's instructions. Lusha's signals endpoint accepts allSignals as a single value, and each signal type in a request counts toward credit usage (signals overview), so the shortest request is also the one that bills the most. A credit-metered API earns most from the broad call; name two or three types in the policy and keep the rest out.
How to know it worked
The holdout is built into lemlist. Make the angle step an A/B test: Version A opens with {{angle}}, Version B is the same email without that line. lemlist sends each version to a random half of the leads, and its A/B article recommends 50 to 100 leads per variation before drawing conclusions. Compare positive replies per 100 leads by version, against the baseline you took.
Watch the reviewer's verdicts as well. My rule of thumb: if more than one angle in three is rewritten, fix the policy, not the reviewer. And log creditCharged per account for the first month, so the Lusha line on the bill is a measured number rather than a guess.
If Version A does not beat Version B after 100 leads each, the research is not earning its keep for this offer. Stop it, and keep the hours.
When the hard part is not the workflow
Sometimes the wiring is fine and the trouble sits above it: nobody has decided which jobs an agent should run in the team, or who approves what. That is the agentic marketing work I sell, so weigh this paragraph accordingly.
FAQ
Can the agent launch the email itself?
Not in this design. lemlist's API has a Launch Lead endpoint, and the agent is not given it: its only tool reads Lusha signals, and the HTTP step after it only creates waiting leads. Auto-launch stays off.
What if Lusha finds nothing for an account?
The agent writes no angle and the row goes to its own tab, never into the angle campaign. A person writes one, or the lead goes into a separate campaign with the plain email. The agent never fills the gap from memory, because the policy requires a URL for every fact and the reviewer opens it.
Where do the per-tool details live?
Lusha's own screens and credits are covered in how to use Lusha, lemlist's mailbox setup, warm-up and launch steps in how to use lemlist, and n8n's approval patterns in how to use n8n.
ABOUT THE AUTHOR
I am Joris van Huët, an enterprise interim CMO. 15+ years, 50+ organizations, including ING, P&G, Nestlé, BNP Paribas, WeTransfer and Vinted. I take on innovation projects (venture building, design sprints), agentic marketing (setting up and running AI agents) and hands-on multi-channel work. See the track record.
I wrote and published this with AI assistance, and I answer for it. Claims about my own experience are limited to the track record above, and a statistic links to its source or is labelled as an example. I sell interim and fractional CMO work, which is why this site exists. How this site is written.
The agents in this article, built into your workflows and handed over to your team: the first one within two weeks.