The argument in one line.
Clay's new CLI plugin lets an AI coding agent operate its lead-generation platform directly from a chat prompt, collapsing what used to take days of manual list-building and outreach-writing into a single spoken request.
Read if. Skip if.
- You run cold outreach (email, DMs, calls) for an agency or B2B service and want a cheaper way to build prospect lists.
- You already use an AI coding agent for other tasks and want to see it operate a business tool through a CLI plugin.
- You're curious how AI 'agent' tools scrape and qualify leads automatically instead of returning a raw contact dump.
- You sell to consumers rather than businesses — this is a B2B decision-maker targeting workflow.
- You're looking for a permanently free tool — the walkthrough runs on a limited free-credit tier of a paid platform.
The full version, fast.
Clay, a lead-generation platform now valued at five billion dollars, released a CLI that plugs directly into an AI coding agent, so instead of manually operating Clay's interface you describe the list you want in plain language and the agent runs it for you. The walkthrough sets up the agent in a code editor, installs the Clay plugin, and prompts it to find ten kitchen-fitting companies in one city along with each owner's verified work email and the company's unique selling point. It then has the agent write a personalized, non-AI-sounding cold email for each lead using that USP, and finally merges list-building and email-writing into one reusable workflow that can be re-triggered for any future batch of leads.
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.

01 · Cold open
The free-leads promise stated up front, with the caveat that ChatGPT works too.

02 · Introduction to Clay.com
What the lead-gen platform is, why it mattered when it launched, and how it aggregated hundreds of competing tools into one API.

03 · The feature that changes everything
The platform's new CLI plugin, which lets a coding agent run the whole system instead of a human operating the interface.

04 · The simple setup process
Installing the coding agent in a real code editor, installing the CLI plugin, and enabling bypass-permissions mode.

05 · How to claim 2,000 free leads
Authorizing the platform account to unlock a promotional free-credit tier tied to the plugin launch.

06 · Building the lead list
A single voice prompt with hard filters returns ten qualified companies with verified decision-maker emails and each company's USP.

07 · The targeting mistake everyone makes
Why most outreach defaults to a broad TAM list instead of a smaller, better-converting SAM list, and how cheap AI list-building removes that excuse.

08 · Writing automated custom emails
Turning the lead list into personalized, non-AI-sounding cold emails that use each company's USP as the hook.

09 · One automated workflow, start to finish
Merging list-building and email-writing into a single reusable Clay workflow triggerable from the agent.
Lines worth screenshotting.
- A lead-generation platform released a CLI in 2026 that lets an AI coding agent operate its whole system directly from a chat prompt, with no manual interface required.
- The underlying platform grew from a zero valuation to five billion dollars in four years by aggregating hundreds of separate lead-generation data sources behind one API.
- Before aggregator platforms existed, building a qualified B2B prospect list from scratch took days of manual research spread across dozens of separate sources.
- The platform's built-in scraping agent visits a target company's website and runs an AI query against it, filtering out any lead that doesn't match stated criteria like 'no bathroom companies, kitchen specialists only.'
- Most outreach campaigns target TAM (total addressable market — everyone who could theoretically buy) instead of the smaller, higher-converting SAM (serviceable addressable market — who a company actually wants to work with) simply because TAM lists are easier to build.
- A single spoken prompt with hard constraints — location, specialization, decision-maker seniority, verified email, and company USP — was enough for the agent to return a qualified lead list with no manual filtering afterward.
- Feeding a company's own unique selling point back into its cold email as the personalization hook is what keeps AI-written outreach from reading like a generic template.
- A one-off lead-generation prompt and a one-off email-writing prompt can be merged into a single reusable workflow that regenerates the full list-plus-outreach pipeline on demand.
- GEO (generative engine optimization) outreach pitches a free audit of whether AI chat tools recommend a business by name — a pitch angle that only exists because buyers now ask AI instead of Google.
- Enabling 'bypass permissions' mode lets a coding agent run terminal commands and install plugins without individual confirmation prompts, trading oversight for setup speed.
An AI coding agent can now operate a business tool, not just write code.
Wiring an AI coding agent into a lead-generation platform's CLI turns list-building and outreach-writing into a single spoken request, and the same TAM/SAM targeting logic applies whether a human or an agent is doing the work.
- The platform's core pitch when it launched years earlier was aggregating hundreds of separate lead-generation tools and data sources behind one API, so no single source's blind spot limits a search.
- Before aggregator platforms existed, building a qualified prospect list meant manually working across separate single-purpose sources, because each early tool only covered one channel.
- Reaching a multi-billion-dollar valuation in a few years is one signal of how much unmet demand there was for a single aggregated lead-generation layer.
- A CLI plugin that lets a coding agent operate the entire platform from chat removes the need for someone to manually run searches and filters inside the platform's own interface.
- Handing lead-generation execution to an AI agent removes the ongoing 'manning the system' step that previously ate hundreds of hours even after the underlying data aggregation already existed.
- Getting a coding agent to control a third-party CLI plugin requires running it inside a real code editor and terminal, not a browser-only or sandboxed version of the agent.
- Enabling a permissions-bypass mode removes the need to individually approve each terminal command the agent runs, trading manual oversight for setup speed.
- Once a plugin is installed, the agent can confirm its own setup status on request, removing a manual verification step.
- A free-credit incentive tied to a new plugin launch is how platforms recruit their next wave of users before those users convert to a paid tier.
- The value of a free-credit tier is time-limited by design — the offer can be revoked once its promotional period ends.
- A single prompt specifying location, industry, exclusions, decision-maker seniority, a verified email requirement, and a request for each company's USP was enough for the agent to return a usable list with no manual filtering afterward.
- The platform's built-in scraping agent does the qualifying work of visiting each company's site and checking it against stated criteria before including it in the list.
- Requiring a verified work email, not just a guessed format, is what makes a scraped list actually usable for outreach instead of one that bounces.
- Most outreach defaults to targeting TAM (everyone who could theoretically buy) because it's the easiest list to build, not because it's the most effective one.
- A smaller SAM list — the people a business actually wants to work with — costs less to build once an AI agent is doing the work, and produces a higher reply rate.
- Cheap, fast list generation removes the old excuse for defaulting to a broad, low-quality TAM list over a narrower, higher-intent SAM list.
- Feeding a company's own unique selling point back into its cold email as the personalization hook is what separates a targeted email from a templated blast.
- Explicitly instructing an AI agent to write 'casual' and to avoid sounding like AI is necessary — without that constraint, agent-written outreach defaults to generic marketing language.
- An agent can generate and batch multiple personalized emails from the same lead list in the same session that built the list.
- A lead-generation prompt and an email-writing prompt, once each is proven separately, can be merged into a single reusable workflow that regenerates both in one trigger.
- Once list-building and outreach-writing are combined into one workflow, generating the next batch of leads no longer requires manually operating any software interface.
Terms worth knowing.
- Clay
- A lead-generation and sales-workflow platform that aggregates hundreds of data sources behind one API, letting users build and enrich prospect lists without manually querying each source.
- Claygent
- Clay's built-in AI agent that visits a company's website, searches for specified keywords, and returns only the leads that pass an AI-evaluated qualification check.
- Clay CLI
- A command-line plugin Clay released that lets AI coding agents operate Clay's platform directly from a chat prompt instead of through its web interface.
- TAM
- Total Addressable Market — every potential customer a business could theoretically sell to, often used by default as the (least selective) target list for outreach.
- SAM
- Serviceable Addressable Market — the smaller subset of the total market a business actually wants to work with, typically converting better but taking more effort to define.
- API waterfall
- A system that automatically tries multiple data providers in sequence until one returns a match, so no single source's coverage gap blocks a search.
- GEO (generative engine optimization)
- The practice of optimizing a business's online presence so AI chat tools and search engines mention or recommend it by name in generated answers.
Things they pointed at.
Lines you could clip.
“I started a marketing agency that's now made tens of millions in revenue.”
“Clay went from zero to a five billion dollar valuation in just four years.”
“Very few companies will build a hyper personalized, hyper targeted list of SAM where they only reach out to the people they genuinely wanna work with.”
“Subject line, does ChatGPT know Solo Kitchens?”
Word for word.
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
The bait, then the rug-pull.
An agency owner opens with a blunt promise: build a list of qualified B2B decision-makers, completely free, using nothing but a chat prompt to an AI coding agent. What follows is a screen-recorded setup of a lead-generation platform's brand-new CLI plugin, which turns the agent into an operator for a platform now valued at five billion dollars.
Named ideas worth stealing.
TAM vs SAM
- TAM — Total Addressable Market
- SAM — Serviceable Addressable Market
TAM is everyone who could theoretically buy; SAM is the smaller set of prospects a business actually wants to work with. Most outreach defaults to TAM because it's easier to build, even though SAM converts better and costs less to reach.
API waterfall
Chaining together many different data-source APIs so that if one source lacks a match, the system automatically tries the next, removing any single source's blind spot.
How they asked for the click.
“Put a link in the description for the setup page... use that link and you'll get 2,000 free credits.”
The free-credit pitch repeats three times (mid-video setup pointer, an explicit dedicated segment, and the final sign-off), each time framed as an early-investor bonus that could be revoked once the promotional window closes.







































































