How to build a marketing agent that turns LinkedIn engagement into cold email leads

Cold email reply rates have been sliding for years, and the culprit is obvious once you say it out loud: everyone's inbox is now flooded with AI-generated spray-and-pray outreach. The old trick of buying a list and blasting it doesn't work anymore. What still works is targeting people who are already showing you they care about your category — right now, in public, on LinkedIn.

That's the entire premise behind a new style of marketing agent that's quietly replacing traditional SDR teams for lean startups. Instead of guessing who might want your product based on job title or company size, you watch who's liking and commenting on posts from creators in your niche. Then you scrape those people, find their contact info, and reach out while the topic is still fresh in their head. It's intent-based outbound, built entirely on public engagement signals.

This guide walks through how to build that system yourself — the tools, the logic, and the order of operations — plus a bonus workflow for turning your team's internal conversations into a daily LinkedIn content engine.

Why does LinkedIn engagement matter more than firmographic data?

Traditional cold outbound targets people based on static attributes — company size, industry, job title. The problem is that a VP of Marketing at a 200-person SaaS company might have zero interest in your product this quarter. Engagement-based targeting flips that logic: if someone just left a comment on a post about, say, AI SEO tools, they've told you exactly what's on their mind today.

The workflow starts by identifying 10 to 20 creators and company pages whose audience overlaps with your ideal customer. Reply rates are down across every channel as AI-generated volume floods inboxes, and the answer is to target intent signals instead of static firmographics, using LinkedIn likes and comments as evidence that a person cares about a specific topic right now. You don't need to cover every niche influencer either — a handful of outliers in any niche capture most of the relevant audience, so chasing full coverage adds cost with thin marginal return.

How do you actually scrape the engagers?

This is where Apify comes in. Apify is a scraping infrastructure platform with pre-built "actors" for pulling data from almost any public source, including LinkedIn. Apify is the scraping API layer: one key, endpoints for LinkedIn, Twitter, and more.

For this specific job, you want actors built by API Maestro. Cody uses API Maestro's LinkedIn actors for profile posts, post reactions, and post comments, then runs a script that takes a post URL and returns deduped engager profiles. A few specific actors worth bookmarking:

• LinkedIn Post Reactions Scraper — pulls every reaction on a given post
• LinkedIn Post Comments & Reactions Scraper — grabs comments, replies, and nested engagement in one pass, and scrapes in batch all LinkedIn post related data including comments, stats, reactions, replies and media attachments
• LinkedIn Profile Posts Scraper — monitors a creator's feed so you catch new posts as they go up

The practical setup: pick your 10-20 creators, run the profile posts actor on a schedule (daily or every few hours), grab the URLs of new posts, then feed those URLs into the reactions and comments actors. What comes out the other end is a raw list of LinkedIn profile URLs — the people who just told the world they're interested in your category.

How do you turn a LinkedIn URL into an email address?

A LinkedIn profile URL alone isn't useful for outbound. You need a real email address or phone number, and this is where waterfall enrichment comes in. The concept is simple: waterfall enrichment lets you search sequentially across multiple tools until you find a valid match.

The tool most associated with this technique is Clay, and its own documentation explains the mechanics clearly: once the enrichment process starts, Clay dynamically moves through the waterfall — if a provider finds a valid email, the process stops for that contact, and if no email is found, data credits are refunded and Clay moves to the next provider until an email is found or all providers are exhausted. That "pay only for hits" model matters because you're often enriching hundreds of engager profiles a week.

For a LinkedIn-first workflow specifically, the sequencing looks like this: LinkedIn URLs go to GitLeads first, then Apollo, then Origami or Prospeo for whatever remains unmatched. Each provider is strong in different geographies and seniority bands, which is why stacking them beats relying on any single source — a single provider often lands in the 40 to 65 percent range on a typical B2B list, while a well-ordered three-provider waterfall commonly reaches the 80 to 90 percent range.

Worth noting on the legality side: this isn't scraping private data. You're pulling from data brokers who've already aggregated and licensed this contact information — the same providers that power tools like Apollo.io and Hunter.io.

How do you set up the cold outreach and reply handling?

Once you have verified emails, the next step is the actual sending infrastructure. Instantly.ai is the dominant tool here for volume cold email, and for good reason — the platform now includes a B2B lead database of 450M+ contacts, email verification, a built-in CRM with calling and SMS, an AI Copilot for autonomous campaign management, and website visitor identification.

The setup process inside Instantly generally follows this order:

1. Connect your sending domains and inboxes (most operators run several dedicated domains, separate from their primary company domain, to protect deliverability)
2. Import your enriched lead list via CSV or direct integration
3. Build your sequence — Instantly's AI Sequence Writer can draft a first pass, though the AI copy generation is useful for first drafts, and quality varies depending on the prompt, so expect to edit rather than publish directly
4. Set sending limits per inbox (the general guideline is to stay conservative — around 30 emails per account per day, per Instantly's own guidance) and let the platform rotate across accounts automatically

For the reply management piece — the part where an agent actually handles incoming responses instead of a human — Instantly's Reply Agent is built specifically for that. The platform includes an AI Sequence Generator that drafts entire email sequences from a prompt, a Copilot that assists with campaign analytics, and an AI Reply Agent that handles initial responses to interested prospects automatically. This is effectively the "SDR in a box" piece of the system — it qualifies replies, books calls, and hands off only the leads worth a human's time.

What about LinkedIn DMs as a second channel?

Email alone leaves a lot of engaged prospects unreached, especially ones who rarely check email but live inside LinkedIn. Running a parallel LinkedIn DM sequence to the same engager list — timed a day or two after the cold email — gives you a second touchpoint using a completely different channel, which noticeably lifts response rates compared to email-only outbound. The messaging should reference the specific post they engaged with, since that context is exactly what makes this approach different from generic outreach in the first place.

How do you turn internal conversations into a content engine?

The second half of this system flips the script: instead of finding customers, it turns your own team's raw material — sales calls, podcast recordings, Slack threads — into a steady stream of LinkedIn posts.

The mechanics are straightforward once you see them laid out. You take a transcript (from a sales call, a podcast episode, or an internal meeting), run it through an LLM with a prompt tuned to extract genuinely interesting insights, and format it into a LinkedIn-native post. Automating the transcript-to-post pipeline typically involves:

• A transcription tool (many teams already have this from calls in Zoom or podcast recording software)
• An LLM API call (Claude or GPT-4 class models both work) that extracts 3-5 distinct "hooks" or insights per transcript
• A drafting step that turns each hook into a short, platform-native post
• A scheduling layer — tools like Buffer or native LinkedIn scheduling — that queues posts across team members' individual profiles

The advantage of sourcing content this way is that it's never generic. It comes straight from real conversations your team already has, which means the insights sound like an actual practitioner talking, not a marketing team paraphrasing a blog post. Running this across an entire team multiplies the effect, because each person's LinkedIn network gets fresh, credible content daily instead of the company page posting into the void once a week.

What's the real cost of running both systems?

Neither of these agents is free to run, but they're dramatically cheaper than headcount. Apify scraping runs on a pay-per-result model — low pricing: $2 per 1k posts for post-level data, with comment and reaction scraping in similar ranges. Waterfall enrichment through Clay charges only on successful matches, and Instantly's pricing scales with sending volume, not per-inbox fees the way older tools did. For a solo founder or small team, the whole stack can run for a few hundred dollars a month — a fraction of what a single SDR hire would cost.

Where should you start?

Don't try to build both systems simultaneously. Start with the LinkedIn engagement scraper — it's the one that directly produces revenue signal. Pick five creators in your space, run the Apify LinkedIn actors on their last few posts, push the results through a Clay waterfall, and load whatever emails come out into Instantly. Watch what happens to your reply rate compared to a cold list you bought off a data provider.

The gap will tell you everything you need to know about why intent beats firmographics — and once that pipeline is running on autopilot, the content engine becomes the natural next build, feeding the exact audience you're now reaching with proof that you understand their problem better than anyone else in their feed.