The best part of cold email is everything before you press send. 😂 Buying domains feels like progress. Loading up 5,000 prospects feels like progress. Watching the campaign go live feels like you're about to print meetings. Then... nothing. So you change the subject line. Then the CTA. Then the tool. Then you convince yourself deliverability is the problem. A few weeks later you finally ask the question you should've asked on day one… Would these people even care about what I'm offering? That's usually where the real work starts. Cold email isn't hard because of the emails. It's hard because good targeting and a great offer can't be automated. What's the lesson that cost you the most in outbound? | 152 comments on LinkedIn

If you still need five tools to prospect in 2026, you're working way too hard. 19 B2B plays you can steal today, all from one search 👇 Finding companies was never the hard part. Knowing which ones, why now, and what to say is. I've been running Prospeo inside Claude. These are the most relevant 19 filters for any B2B angle - you can get all these in one search: TARGET BY HOW THEY SELL 1/ Product-led: Business Model = self-serve + free trial → Claude: outreach that leads with time-to-value. 2/ Sales-led: Revenue Model = sales-led + demo → Claude : write for a demo buyer, not a trial. 3/ Sells to your buyer: sells to [buyer title] → Claude: partnership angle, same audience. 4/ Lookalikes: Company Lookalike = [best account's domain] → Claude: open with why you fit teams like this. TIME IT WITH A SIGNAL 5/ New leader: Exec Events = new CEO/VP, last 60 days → Claude: short note tied to their first 90 days. 6/ Just raised: Funding = last 90 days → Claude: tie their raise goal to your value prop. 7/ News-based: Company News = launch/expansion/M&A → Claude: reference it, don't pitch the product. 8/ Traffic up: Website Traffic = trending up → Claude: scale-up angle. 9/ Traffic down: Website Traffic = trending down → Claude: growth-audit offer. QUALIFY BY WHAT THEY ARE 10/ Compliance-driven: Certifications = SOC2 / ISO / GDPR / HIPAA. 11/ VC-backed growth: venture-backed + revenue + headcount growth. 12/ Has API / uses AI: AI Attributes = has API → Claude: integration pitch. 13/ Team in one country: Headcount by Location = [range] in [country] → Claude: write for that market. 14/ Competitve takedown: Competitive Intel = mentions [competitor] → Clude: switch pitch, no trash-talk. 15/ Lists a brand you know: Key Customers = has [brand] → Claude: reference the shared connection. GO DEEP ON THE ACCOUNT 16/ Ranks for a keyword: Google Discovery = [keyword] → Claude: niche-aware opener. 17/ Says an exact phrase: Company Keywords = [phrase] → Claude: their own words back at them. 18/ Hiring now: careers page hiring [role] → Claude: fill-that-gap angle. 19/ Whole buying committee: Smart Job Title = expand [title] → Claude: one message per role (buyer, user, blocker). How to run it with Claude: 1. Talk to it: Prospeo MCP inside Claude, "build a list of [your ICP]" — no exports 2. Check before you spend: Search Analytics, export only the segment worth the credits. 3. Hand off clean: push a tiered list to your CRM, best-fit accounts get the call first. Don't pitch on the first touch. Open with something useful, then iterate. Do you find these useful? P.S. If I earned your trust, I'll set you up on Prospeo and automate all your TAM sourcing at a better price than Apollo. Book a time with me. | 94 comments on LinkedIn

Today Clay is launching Account Research Agents! This is the first of a series of always-on GTM agents we'll be releasing this summer. Here's how it works: these agents monitor all your accounts, reading new call recordings, emails, signals, and warehouse data. They can synthesize and write key updates to your CRM or data warehouse automatically, such as who the champion is, what stage the deal's in, what the risks are, why you lost. No more harassing your sales team to update Salesforce. In closed beta, teams used the agent to flag high-usage customers for expansion, revive closed-lost deals with the real loss reasons, and quickly ramp new reps with full account history. Account Research Agents are now available on Enterprise, Growth, and Launch plans. Check it out here: clay.link/t3GoXtv | 20 comments on LinkedIn

We trained an AI on 3,019,536 emails and 2,355,916 linkedin positive replies generated by our customers at lemlist (real numbers). Unlike ChatGPT or Claude, it's specialized in outbound. And it doesn't need a separate subscription. We call it lemAgent (for now). Here's the idea 👇 Today, launching a campaign means jumping between 3 tabs: ChatGPT to write, a lead tool to find people, lemlist to build the sequence. An hour later you've got a half-built campaign and a lead list that's almost right. lemAgent kills that: one conversation. Say you're an SDR at Pennylane (lemlist's cutomer) and selling to CPA firms. You tell lemAgent: "Find accounting firms with 10-50 employees and build me a campaign to reach their partners." From that one sentence, it: → Finds the accounts: searches our 650M database and pulls accounting firms in that size range and your region, with the partners' names and contact info → Ranks them by buying intent: like firms with multiple open position for junior accountants (drown in manual work), or firms that visited your website, or even firms with recent m&a activities. → Builds the full sequence: email + LinkedIn, right timing, right steps → Personalizes every message: references the firm's recent hiring or growth, not a generic "Hi {{firstName}}, I hope this email finds you well" One important thing about this agent: it's not autopilot. Because we don't think autopilot works for outbound today. So we built a copilot, eg "human in the loop" It asks you questions about your business, your USPs, your ICP to actually understand before building a campaign that can generate meetings and revenue. lemAgent builds the campaign, but you still run the show. Nothing goes out without your sign-off, so it doesn't harm your brand. A few weeks into the beta, 10% of our daily active users are already on it. Now the fun part: we need help to find a better name than lemAgent. We were thinking something spatial as "lem" stands for Lunar Excursion Module, but ideally human, like "Ariane." Drop your idea in the comments. If we pick yours, I'll send a nice gift 🎁 | 29 comments on LinkedIn

I think the best cold email campaigns have 3 things in common. A campaign that drives ROI doesn't need all 3, but each one of these points drives significantly more replies. Ever since we started doing free tests for companies to work with us, we've launched quite possibly 3-5x more campaigns with different offers than we previously would. Before, I thought I could tell from one look if a campaign was going to work, now, I'm basically never surprised. I've tried to put my finger on what makes a great campaign versus others and this is what I've been thinking. 1. Great campaigns communicate the promise to either help companies make more money, save 90% of time, save 90% of costs in a way that the recipient didn't know about with a next step that is easy to say yes to. The best type of offer to convert cold traffic is an offer that helps people make more money for sure. If you can, always lead with that. If you can't, the time savings / cost savings you can provide has to be orders of magnitude higher. When Instantly.ai first released a sequencer with unlimited inboxes where everyone else was charging per mailbox, the community went crazy. And that was maybe a 99% cost reduction depending on your scale. Save Time/Save Money offers have to be orders of magnitude better than what they are doing today. 2. Great campaigns communicate a mechanism that the recipient didn't know existed. I think the increase in amount of outbound messages people are receiving is maybe responsible for 20% in the decrease in outbound effectiveness. (I'll make another post demonstrating this). The bigger driver of the decrease in outbound effectiveness is that prospects have access to more information than ever before. They know that performance outbound lead gen, bookkeeping, HR management, payroll automation, cyber security, recruiting, are all things they can get access to. And when they're ready they will find you. They don't know that - Trigify.io can find customers talking about their exact problem - That negative google reviews can be removed with a proper appeal process - That Forbes is more pay to play than you might think - That the Donut can target newsletter subscribers by income/interest - That Starbridge can get them a list of governments that have open RFPs. So the second thing great outbound campaigns do is present a mechanism they didn't know was possible. 3. The best outbound campaigns explain the value in taking the next step AND the next step is easy. It costs Young Dental $200-$300 to get a dentist to request a sample through Facebook ads. With us, it costs ~$40. Their campaigns if anything are a time saving / reducing risk of a bad patient experience type offer. Not make more money. And these dentists don't know that these products exist. But the call to action being to get a free sample set sent to their office is such a no brainer that they get tons of responses. Last part in the comments, this was too long | 22 comments on LinkedIn

Last week Clay released an API after all of the smartest Clay pundits in the world told me they’d absolutely never ever do it. What does THAT tell you about the state of the world? My Take: You’re an idiot if you’re building UI for people. The argument, made to me by INCREDIBLY smart people, was that a public API would hurt Clay's business by making it less sticky. Clay was sticky BECAUSE workflows lived in the UI. Let people build on an API, and those workflows go elsewhere. It also devalues Clay's vendor relationships, because you no longer have to log into Clay to access that data. Which was their other "moat." Now, it's possible that 100% of the revenue Clay actually cares about is Enterprise, and that revenue will be VERY slow to move onto an API/CLI. If that's the case, Clay may have done this to get UGC going again. UGC comes from the SMB segment, and you can't kill stickiness that's already dead there. If that's the play, tip of the hat to Kareem and the Clay team. Wise move. But even so: Clay just did something that, if you asked Claude whether they should do it, Claude would say "no, that kills your moats and your stickiness." That's a HUGE statement about the state of the world today. #1: I love that at least ONE other CEO isn't blindly taking direction from Claude. #2: This way of doing things is in such high demand that a $5B darling unicorn is willing to satisfy it, even though it's very not good for the stickiness of their business. What am I getting at? If I were starting a new business today (and I am)... I would service this use case and do nothing else (which I am). UI for humans is dead. Long live APIs. 🏴‍☠️ | 71 comments on LinkedIn

1 in 4 of your posts should convert. The other 3 just warm the room. Here is the 4-tier content score we run at Growth Today: In the same way we tier an outbound list, we also consider the same logic in the content strategy. Each content piece has a job, and the tier a reader lands in decides the message they get later. Here is the 4-tier content score: TIER 3, BoF, convert ↳ a real asset: a lead magnet, a case study result, a teardown with proof ↳ With the highest intent, the people who engage here raised their hands ↳ The DM it earns: reach out within a day, name the thing they grabbed, or if it’s not relevant, just keep your outreach short + sweet TIER 2, MoF, prove ↳ How you think, how you solve a problem, a contrarian take with a method ↳ This is the workhorse; it makes the reader think you know your stuff ↳ The DM it earns: a personal note that references their comment, no pitch yet TIER 1, ToF, reach ↳ a story or a stat that attracts and pulls a broader audience ↳ grows the top, new followers, future pipeline ↳ The DM it earns: none yet, a like is not a buying signal TIER 0, skip ↳ posting for the sake of posting, no point of view, no asset ↳ it fills the feed and trains the algorithm on weak engagement The trap is not knowing your ICP enough. If that’s the case, it is very easy to spend creating content (in any format), chasing vanity metrics, and nothing is going to 1) drive inbound pipeline 2) speed up closing due to increased trust and education 3) attract talent, investors, and partners based on your goals Tools to get started: Scripe + HubSpot/ Attio To improve: Apify + Python + Claude So we run a weekly split: mostly MoF to prove it, one ToF for reach, one BoF to convert. Then the engagement on each post feeds the same signal score our SDR is using for outbound, supported by GTM engineers. Content and outbound are not two jobs. They compound when they work together. What share of your posts last month were just reach? Comment SCORE, and I will send you the 4-tier framework we use for both. P.S. We build signal-based GTM engines for B2B teams, where content, ads, and outbound run on one score. If that is the motion you want, book a call through the site or DM me. | 122 comments on LinkedIn

Claude's Fable model finally made my dream of automatically generating leads for companies I've never even spoken with a reality. I didn't lift a finger to launch. I've always thought the best customer experience we could have is to show someone that we can generate leads for them before they pay us anything. So we started running free tests for people to prove it. I've always wished it could happen automatically so it could become a growth motion bigger than what my organic reach / outbound can convert. Fable finally cracked it. From sub agents that source the right fit companies to run campaigns. To building the ICP list in Prospeo, BlitzAPI and GetLeads (depending on the job) To the hardest part, copywriting that would actually convert. Fable was the model we needed to finally handle all of the unexpected nuance of this endeavor. We ran our first 10 completely automatic campaigns and 3 of them got leads. We have multiple steps to never mention the company name that we are testing this for so and to never mention any social proof. We are purely testing if the value proposition resonates with the prospects. We generated leads for Trove RevolutionParts Luxury Presence The replies are all quoted in the graphic below with the ICP and main hook of the messaging. Again, we don't mention any of the company names or anything. Just the value prop so we can validate the messaging at scale. | 28 comments on LinkedIn

Today Clay is launching our MCP for reps. There are two ways AI gets adopted across GTM orgs: centralized and decentralized. Right now there's tension between the two. Reps want to move faster than Ops can build workflows for them, so reps go rogue and DIY their own AI tools. But there's no governance. Reps are pulling "new" leads that are already in your CRM and running outreach on stale data. Clay's MCP for reps fixes that. Reps can now prospect with Clay data and workflows right inside Claude, ChatGPT, Microsoft 365 Copilot or Glean. Ops maintains control the whole time: which workflows each rep can run and how many credits they can spend, with only high-quality, net-new records written back to the CRM. Ops gets centralized governance without giving up flexibility and speed. Reps spend less time digging and context switching, and more time actually selling. I think Rob, the owner of Veniero's, a 132-year-old NYC pastry shop, can explain it better than I can. Watch below 👇 P.S. Join our CEO Kareem on July 21st for a behind-the-scenes look at Clay's Product Roadmap. We're shipping 12 new features in 12 weeks and he's going to walk through them live. Sign up here: clay.link/YvTGhGt | 32 comments on LinkedIn

You can have the perfect cold email. It still won't book meetings. Because cold email has never been about one email. It's about everything that happens before and after it. ⤷ Who you target. ⤷ Whether your lead list is accurate. ⤷ Whether your offer gives someone a reason to reply today instead of next month. ⤷ Whether your subject line earns the open. ⤷ And whether you follow up when they don't reply the first time. People spend hours rewriting the copy because it's the easiest part to see. The real problems usually sit somewhere else in the process. ⚠ A weak ICP. ⚠ A poor lead list. ⚠ An offer nobody cares about. ⚠ No follow-up strategy. That's why we treat outbound like a system, not a writing exercise. When every link works together, meetings become predictable. When one link breaks, even the best email struggles to perform. Which part of the chain do you think most teams get wrong? Let me know in the comments. Follow Nick Palasz for more cold email and outbound strategies. | 120 comments on LinkedIn

Prospects signal they want to talk to you in 4 different places. Most B2B teams only track one. Here is the 7-step system we use to capture all 4: ↳ Your website (who visits, how often, which pages) ↳ LinkedIn (who reacts, comments, reposts) ↳ Hiring pages (e.g., companies hiring for RevOps need operational expertise or bandwidth) ↳ Tech changes (companies adding new tools have a budget) …Or swap a few while listing out what is relevant for your ICP to book more meetings through cold calls, emails, LinkedIn touches. Here is the 7-step system: 1/ See who is interested Pull signals from all 4 places into one place and keep this info in your CRM. CRM = source of truth ↳ RB2B, Trigify, Common Room 2/ Group them by company One person from a company is a hint. Five people from the same company are a hot account. Treat the signals as one full picture, not just 50 random leads. ↳ Clay 3/ Rank them by fit Not every interested company is your buyer. Sort them by how well they match your best customers. Best matches first. Think of: Tier 1, Tier 2, Tier 3, and unassigned/untiered. ↳ Claude/OpenAI that goes into Salesforce/HubSpot 4/ Find all the deciders One person rarely buys alone. Find the 3 to 5 people in each company who actually decide. Who pays. Who pushes for it. Who tests it. Think of: buyers, influencers, end users, and blockers. Give them a different score ↳ Clay, Apollo 5/ Stack your contact tools One tool finds about half the emails. Run several in a row. The next one catches what the first one missed. You only pay when a real email is found. ↳ BetterContact, Prospeo, FullEnrich, Icypeas - just to mention a few. 6/ Write as you know them Before you write a word, look up what is happening at their company right now. Hiring, funding, and a new product. Use the real thing. Not "Hi {First Name}" with the name swapped in. Sources that your competitors are unlikely to use: your own first-party data (eg, product usage, ad interactions, etc.), custom data unique to your business, community signals (eg. Trustpilot, Slack, Reddit community etcÖ ↳ Claude, Trigify 7/ Match the channel to the fit Best accounts: multichannel with calls Medium fit: mix of manual + automated steps, fewer touchpoints Low fit /unassigned: automated ↳ Nooks/Orum (parallel dialers) or Lemlist/Hubspot (if your team doesn’t make 1000s of calls per day) Rule we never break: no pitch on the first touch. Open with something useful. The trick is the system, not any one tool. How many total signals/insights are you monitoring? P.S. Feel free to reshare this if you know somebody in your network who would benefit from this. | 108 comments on LinkedIn

Clay finally shipped their Claude Code CLI & API, along with Workflows, a completely new way to build in Clay with unlimited runs These two features are probably the biggest product improvement they've released in the last 2 years. What this means: You can now build in Clay entirely from your terminal. That means you can use your own skills and best practices to build workflows, directly with natural language. It's MUCH better than Sculptor was, which kinda flopped in my opinion. Accompanying the new CLI is Workflows, the primary way Claude will build in Clay for now. Essentially a Zapier style workflow builder rather than a Clay table, it's also an amazing improvement in my opinion. After years of scrolling through 100 column tables and clicking columns, I can't tell you how happy I am about this. Workflows sync up to Audiences and have unlimited runs... which means you're no longer limited at all by the 50k row limit. FINALLY! They've also released audiences to anyone on their new plans, which means that you have now unlimited records synced into Clay, which is a huge improvement as well. For anyone who's worked on a dataset with 50k+ records, you know why. Now you can sync in your entire CRM with natural language, have Claude build a workflow for you that can run on unlimited records, and it is easier to see if-then branches and whatnot and use all of the Claygent system functions you already have in your system. I frankly didn't think the day would come when Clay would release a Claude Code CLI. I thought they were going to force folks to log into their UI so they wouldn't become "another tool in your stack", especially with their orchestration narrative. I was wrong. Huge kudos to Clay for this. Excited for the rest of the summer of launches! | 21 comments on LinkedIn

This email booked 5 meetings with $1 billion orgs Here’s how it could be even better For context this is from one of my clients David, he: - Booked 33 SQOs with billion dollar orgs in Q4 - Sourced $2m+ in qualified pipeline - Brought in $500k+ closed won (proof in the comments) He sells a very niche product of spare part manufacturing software to $1b+ revenue manufacturing companies in North America. The buyers are hard to reach, very analytical, and are often set in their ways. Yet we were still able to get up to a 6% reply rate with outbound. But it could be better. So this is exactly what I’d change about his email: 1. COI, the impact is okay but it’s not very clear as to what that gap would actually cost. This is key, especially when prospecting into ENT orgs. 2. Vague solution, the solution again is okay but very broad. If we added more context as to what we do it would make it easier to understand. 3. CTA, again I like this CTA but think we should A/B test a few different ones to see what works best, I may opt for a shorter one due to the body length of the email. Email is simple, you need to nail 3 parts: - Who the messaging is for - Why you’re reaching out - Why they should care If you do that, you’ll get replies, book meetings and build pipeline. Fall short at any stage? And you’ll end up in the sea of generic emails. LMK what you think of this email in the comments! | 29 comments on LinkedIn

You can now build on Clay right from your terminal in Claude Code, Codex, or Cursor.

FOUR basics you need to learn Before diving into ‘agentic AI’ and building random automations Skip them and your automations will always underperform. 1️⃣ How to Prompt AI A simple 5-part structure: ↳ Input → what you're giving it ↳ Action → what you want done ↳ Output → the format you want back ↳ Prefix→ the role/context it should take on ↳ Constraints → the rules and guardrails Nail this and AI actually does what you meant, not what you typed. 2️⃣ Creating Projects Claude Project or ChatGPT Project. Same idea: a dedicated "folder" inside your LLM that specialises in one area of your work or life. E.g. a Contracts project that drafts agreements from your meeting transcripts and email threads. Now you basically have a beta "AI agent" at your fingertips. 3️⃣ MCP Connectors MCPs mean you don't need to worry about duct-taping tools together using APIs. Think of it as glue between tools. As soon as you connect tools using MCPs, you can perform actions in those tools without doing it manually. 4️⃣ Building Simple Automations 1. Repetitive → what process always follows the same steps? E.g. finding candidates always includes: → filtering LinkedIn Recruiter based on 5 filters → adding to list → exporting CSVs and enriching with mobile and email → sending multi-channel outreach 1. Trigger → what is the first action that kicks off that sequence? 2. Action → what happens after the trigger fires? 3. Result → what is the end goal of the automation? (In this case: reaching out to candidates to propose their dream job.) Sue me. But I think anyone teaching you more complex 14-step prompt engineering formulas is: - either too smart to try and replicate - overcomplicating your life too much PS: If you want me to share resources to learn these 4 things… DM me "screw AI agents" and I promise to send something useful across. (Make sure you have registered for the Unlock GTME Waitlist - Link in top comment)

Nobody will remember how many emails you sent. They'll remember the results you created. It's easy to get caught up chasing numbers that look good. But the metrics that matter are usually much simpler. Here are a few things people actually remember: ✅ The meetings you booked. ✅ The pipeline you generated. ✅ The deals you closed. ✅ The ROI you delivered. ✅ The customers you helped win. ✅ The revenue you brought in. ✅ The opportunities you created. ✅ The problems you solved. ✅ The impact you made on the business. Remember: Being busy isn't the same as being effective. ↳ Focus on outcomes. ↳ Ignore vanity metrics. ↳ Let results do the talking. P.S. Which metric do you think people spend too much time obsessing over? | 103 comments on LinkedIn

Every week, your posts collect dozens of warm leads. 7 steps to turn that list into booked meetings. Here is the 7-step system we run to turn it into booked meetings: 1/ Post for the buyer you want Write straight to the person you want to book. Make sure to create content that resonates with your IDEAL DREAM customer. To get started: use your insights or speak to the AE who closes the deals To improve this: pull insights from @Gong, Fellow AI and any notetaker to extract To make it even better: pull insights from public research, including defined Reddit, Slack, LinkedIn, and other communities To make this an autonomous engine: Python script runs on a scheduled run that pulls insights weekly from all the calls and summarizes/updates a database that includes this information ↳ Your own content calendar 2/ Pull everyone who engaged Likes, comments, reposts. Pull this automatically using the tools of every profile that interacted with your best posts (or competitors and relevant thought leaders) ↳ Apify or PhantomBuster or Trigify.io or Clay 3/ Enrich the list Raw profiles tell you nothing. Add company, role, seniority, and firmographics so it becomes a real prospecting list. ↳ Clay 4/ Rank them by fit Not everyone who engages is a buyer. Sort by how well they match your best customers. Best matches first. ↳ Claude or OpenAI, pushed into Salesforce or HubSpot (source of truth) 5/ Stack your contact tools One tool finds about half the emails. Run several in a row so the next one catches what the first one missed. You only pay when a real email is found. ↳ BetterContact or Prospeo or FullEnrich or Icypeas or AI ark or Blitz API RECOMMENDATION: Have a minimum 2 paid waterfall contact data to maximize the coverage 6/ Add context before you write Before you write a word, look up what is happening at their company right now. Hiring, funding, and a new product. Use the real thing. FRAMEWORK: INSIGHT → GIVE VALUE (no pitch on the first touch) ↳ Claude, OpenAI 7/ Reach out by fit If you’re getting started: automated steps. Done is better than perfect! Ideal: best accounts: multichannel with calls Medium fit: mix of manual + automated steps, fewer touchpoints Low fit /unassigned: automated ↳ Nooks/Orum (parallel dialers) or lemlist/Hubspot (if your team doesn’t make 1000s of calls per day) How many warm leads are sitting in your notifications right now? P.S. If you’re a founder / GTM executive, and you or your team is buying a cold list while ignoring their own engagement, DM me. | 138 comments on LinkedIn

If I wanted to Become a GTM Engineer… I'd do it in this order. Earlier this week, a student emailed me with a simple question. "If you were starting from scratch in 2026, would you learn Clay first or Claude Code?" I've been getting variations of this question a lot recently. Usually followed by: "How do I land one of those $5k to $7k/month remote GTM Engineering jobs?" My answer usually surprises people. In this week's Clay Cult Newsletter, I break down the roadmap I'd follow if I had to start all over again. Including: → Why learning B2B sales is more valuable than learning another tool → When Claude Code should become part of your stack → Why Clay makes much more sense once you understand the workflow → The mistake I see almost every beginner make → The simple 5 question checklist I use before building any automation If you're trying to break into GTM Engineering, or you're wondering what to learn next, this issue should save you months of heading in the wrong direction. You can read the full newsletter here: https://lnkd.in/e4avKX9b

2026 AI FIRST GTM Tech Stack A lot has changed since last year (crazy how fast) so here's a breakdown of the most interesting things I discovered, tested and put in my s++ TIER LIST General Productivity Wispr Flow Claude Design GTM Ops Claude + Deepline (93 integrations in 1) Clay (best for enterprise) Email Sending Tools EmailBison Smartlead Instantly.ai LinkedIn Tools Hey Reach (best for linkedin automation) Sales Navigator (best for network and social selling) Cold Calling Dialers Salesfinity Nooks Intent Signal Providers HG Insights (most accurate tech stack data for enterprises) Vitelis (most powerful open data engine right now) RB2B (company visitors) Mobile Enrichment Forager (best mobile coverage) FullEnrich (best waterfall of providers) Cold Email Infrastructure sending.ac (Microsoft Azure domains + addresses) 1-1 Outbound Tools Outreach (best for enterprise and big team) Commsor 🦕 (warm intro and GTN) CRM Salesforce (best for enterprises) HubSpot (best for scale-ups) Attio (best for brokies) What's the single most interesting piece of tech stack you added in 2026? - Do you want my best AI & GTM Engineering Trainings? Go to Michael Saruggia and click on 'View Newsletter' | 24 comments on LinkedIn

3 boring ways to use AI that can get you high intent leads every day automatically. Multiple tools can set this up for you, you just need to pick your favorite. Pick your favorite AI coding harness. Cursor, Codex, Claude code. Anything that can run a scheduled task every day. These tools can do all of these strategies on their own. But consider layering Parallel Web Systems or Exa on top for some more purpose built searching. However you want to do it, just ask your AI buddy to do any of these three things. I want to be clear that all of these start with defining who your very best customers are. 1. Ask AI to find 3 perfect companies based on some trigger event per day. Keep a local csv so it doesn't show you the same 3 companies again for whatever time duration you want to set it for. I had Parallel find 3 companies that are hiring for GTM Engineer roles as you can see in the image. It's boring but getting this data collection done on a daily basis with something as smart as the parallel.ai task agent is a great use case. 2. Ask AI to find an industry event that it is coming up and find the sponsors. If we could run only event campaigns on cold email for the rest of time, we would be the most successful agency ever. Of course, we can't because there aren't a ton of events. But when they do come, you want an Agent finding the new ones and scraping the sponsors / attendees if possible because the list is so high intent. 3. Automate your Dream 100 outreach When people ask me how outbound is going to have to be different, I consistently believe that it will evolve to have to bring true value in the message. Chet Holmes created the dream 100 outreach as a thought process of "what are the lengths you would go to acquire your dream 100 customers." Once you answer that question, most of the things you would do are totally possible for AI to do on your behalf. You can use your token allotment to run a skill on 5-10 companies per day or more and create the dream 100 resource and then you can manually send (or automatically of course) to those prospects. My second post today will go over one we made yesterday. Copy and paste this to your agent friends and comment how easy it is to set up. It's something I think anyone can do no matter your AI skill level.