In this step-by-step video, you’ll discover why evidence not just personality or persuasion moves prospects to commit and transforms your pipeline results.

This video will help you to turn cold email replies into high value meetings using dynamic, personalized lead magnets exactly how we at StackOptimise does it for seven figure B2B clients every month.

AI Agents vs Workflows vs MCP: the simple breakdown for RevOps and GTM teams Everyone's throwing these terms around, but most explanations are either too technical or too surface-level. Here's how I think about each one: Workflows = the assembly line Think Zapier, n8n, or Make. You define the exact steps: 'When this happens, do that, then this.' Good for: predictable, repeatable tasks where you know all the steps upfront. AND you don't need the system to think, just do. RevOps example: When a deal hits Closed Won → update Salesforce → send Slack notification → create onboarding task in Monday GTM example: New lead from form → score in HubSpot → assign to SDR → send welcome email sequence The limitation is you have to anticipate every scenario. No flexibility when things get messy. AI Agents = the smart assistant They can reason through problems, make decisions, and adapt to new situations because you use an LLM (gpt, claude, perplexity etc). Good for: complex tasks that require judgment, research, or creative problem-solving RevOps example: 'Analyze this quarter's pipeline data and identify accounts at risk of churning, then draft personalized retention emails for each' GTM example: 'Research these 50 prospects, find relevant pain points, and write personalized outreach messages that reference their recent company news' The limitation is its less predictable. Sometimes brilliant, sometimes needs guidance. MCP = the universal translator Model Context Protocol. It lets AI agents talk to your tools directly - Salesforce, HubSpot, Slack, whatever. Without MCP: You copy data from Salesforce, paste it to ChatGPT, copy the response, paste it back With MCP: The agent pulls data from Salesforce, processes it, and updates records automatically Good for: eliminating the copy-paste dance between AI and your actual work systems RevOps example: you tell an agent via natural language to query Salesforce for stalled deals, analyze patterns, then automatically create follow-up tasks and update opportunity records GTM example: you tell an agent via natural language to research prospects in Clay, writes personalized emails, then load them directly into your outreach tool Workflows handle the routine stuff that requires no reasoning. Agents tackle the thinking-heavy work. MCP makes it all seamless via natural language direction. The real unlock is orchestrating them together. Start simple though. Pick one process that's eating your time. Map it out. Then ask: What parts need rules? What parts need reasoning? What parts need seamless data flow? That's your roadmap. Credit to Generative AI on the image --- If you want Spring Drive to help you leverage AI for GTM Engineering, DM me. | 51 comments on LinkedIn

If you're wondering which LLM to use, here's my breakdown. Why they're good...and BAD. 1. Open AI remains a fundamental pillar and one of my go-to LLMs. O3 is genuinely powerful for research - the depth it can go when you need comprehensive analysis is impressive. It's references are pretty good but definitely need to be checked over - it rarely spits out something completely irrelevant, but might well spit out something outdated. 4.1 and 4.0 are great fast, efficient tasks - we use the mini runs a lot in Clay when doing simple but effective research at scale. 2. Claude is where language gets interesting. Sonnet 4 is excellent for semantics and nuanced writing that's hard to match. When you need something that doesn't just sound right but feels right, Claude tends to deliver. Opus 4 brings more horsepower when you need both quality and complexity, and has the ability to build some crazy stuff. However, I have found that it can sometimes hiccup on research that o3 wouldn't on... 3. Gemini 2.5 is the one I'm still figuring out. I know this is a fan favourite for quite a few, but so far I've found that it hasn't added a huge amoutn to what others already do for me. Although when I questioned it for 3 hours at 1am about whether or not Google want to kill cold email - it was rather nice about the whole thing. 4. Perplexity has carved out something specific: research legitimacy. I ran a recent test across o3, opus 4, and perplexity's deep research, and perplexity smashed the others on highly relevant and on point references. When I need sources that actually check out and citations that hold up, it's becoming the go-to for me. 5. Grok surprised me with its conversational flow - if I want to work through a problem that doesn't have a clear binary solution, I find its fast and comprehensive. This is actually where I don't quite like Perplexity - I find it sometimes won't go as far as I want it to, even if what it has provided is good. I'm not going to say too much on the recent misgivings on Grok (iykyk), but certainly in the use of GTM, it's great to work with. My advice is to mix it up - but you probably don't need to use like 4 just to be confident you're not missing something. You'll probs end up with your favourite 2 or 3 depending on the use case. And guess what, that's fine. What's your favourite LLM? Comment below let's rack up the votes... | 21 comments on LinkedIn

Get the complete 2025 guide to cold email deliverability coming straight from the agency sending 1,000,000+ emails a month for top B2B brands!

Claygent Navigator is kind of nuts. Think a ChatGPT or Claude agent for as many contacts/companies as you want, all at once. Organized. I don't know if I'm alone in this, but I personally like to see the steps that are being taken when an agent is doing research to identify whether or not my prompting is adequate or if I need to make changes to provide more clarity so that the agent can do its best work. You can now do that in Clay with Claygent Navigator. They released this maybe a couple of weeks ago, but it didn't make a big splash. It should've. It's a massive upgrade from the traditional Claygent that we're all used to. While Claygent has been great for the last couple of years, Claygent Navigator unlocks the next level of agentic research at scale. It's expensive, I'll be honest, I think I paid 8 Clay credits per row. But in small volumes, it can absolutely be worth it. In this use case, we're running LinkedIn ads for one of our clients and we're looking to enrich the data as it comes in from a lead form, and send it to them in real-time via a webhook rather than the native HubSpot integration in order to facilitate speed. Something that's important to them is the size of a social media team. Identifying the size of a social media team could be done previously with a handful of columns stacked on top of each other to find information (whether it's finding jobs that exist with social media in the title, job postings, etc.). Now you can do this all with one agent in a single column with the same level of accuracy. For me, this is a massive efficiency unlock and will be using this for our clients moving forward. If you need a hand optimizing your inbound pipeline, get in touch. There's a million different workflows that we've built already, and there's no limits on what we'll be building moving forward. | 22 comments on LinkedIn

We automated 70 outbound tasks last quarter. And honestly… should've done it sooner 😅 So, here's what we learned Most outbound teams are stuck doing manual work that AI already handles better. Research, personalization, follow-ups, CRM cleanup... the list goes on And we mapped it all out: 70 Outbound Tasks AI Does Faster Than Your Team From prospecting & research to pipeline hygiene & rep coaching organized into 6 categories you can use right away Tasks that free up your team to actually sell 👉 Comment "70" and I'll send it your way. Make sure we are connected (ps. if you're curious which tasks made the biggest difference for us... it's all in there 🙌) | 688 comments on LinkedIn

I spent 1000+ hours on Clay & GTME Implementations Here are 29 lessons I wish I had known earlier: 1. You can use Clay without using credits 2. Clay can scrape local businesses 3. Clay can look up and update the CRM 4. Clay is the ultimate cold email tool 5. GTM Engineer is now one of the highest-demand jobs in tech 6. You can scrape LinkedIn posts and mentions 7. You can do account research in Clay 8. You can scrape Google News from Clay 9. Becoming a Clay expert changed my life 10. Clay is not only an outbound tool 11. Clay can analyze a company’s 10K report 12. Clay is a LinkedIn automation tool 13. You can create templates in Clay 14. Clay can navigate websites and scrape any piece of info 15. Clay can get financial data from any company 16. Clay can scrape e-commerces 17. You can enrich emails with 10+ providers in Clay 18. You can use Clay as an HR tool 19. You can track open jobs in Clay 20. You can replace many parts of the tech stack with Clay 21. You can connect webhooks in Clay 22. You can connect any API to Clay 23. Clay is an amazing cold call operations tool 24. You can manage and enrich inbound leads with Clay 25. You can build AI formulas in Clay 26. Clay is a martech and sales revolution 27. You can de-anonymize web traffic in Clay 28. Clay is not expensive 29. Subscribe to my Clay newsletter and download my book for more Enjoy! | 30 comments on LinkedIn

The modern sales funnel: Prospecting: Apollo.io, Ocean.io, TheirStack, Commsor 🦕, Hifive 🙌, Crossbeam, RB2B, Teamfluence™, Tally, Chili Piper Enrichment: Clay, Persana AI, BetterContact, BounceBan (Verify risky emails without sending messages) Engagement: Smartlead, Instantly.ai, Saleshandy, EmailBison, HeyReach, Expandi.io, GetSales, Dripify.co, Reply, lemlist, Outreach, Salesloft Sales: Senja, Deeto, Testimonial, Boast, trumpet 🎺, Paage, Valuecase, Distribute, PandaDoc, Qwilr, Proposify, Nusii Proposals S.L. | 83 comments on LinkedIn

We send over 1 million cold emails every month - and these are the 47 outreach copywriting rules that make them convert.

We’re able to serve 283 active clients with an extremely lean team, 100% due to our LMS. What I'm about to share about creating a great one has taken me 5+ years to learn. When an employee joins your company, what do they need to know to be successful? This flips your methodology from teaching what you want to cover to teaching what they actually need. After building training systems for hundreds of employees across different roles, here's the framework that works: The 4-Module Structure: 1. Universal Knowledge - Everyone on your team should understand your core service or product. If you're a cold email agency, create "How to Create a Cold Email Campaign." Sales reps, CSMs, developers - everyone needs shared language and understanding. 2. Company Foundation - Cover your history, competitive landscape, different products, and customer journey from start to finish. This gives employees context for why the company exists and how it operates. 3. Industry Fundamentals - Not everyone comes in knowing your space. Create basic training for B2B sales, key acronyms, proposal processes. Especially important for developers from other countries or new hires switching industries. 4. Role-Specific Training - Cover all operational tasks for each specific position. Day-to-day responsibilities, processes, and how to execute their job successfully. Customize this entirely per role. The point is that you structure your LMS around what people need to know, not what you want to teach them. Think from the employee's perspective - what knowledge gaps would prevent them from succeeding in their role? Save this framework and start building your LMS the right way. The infographic below breaks down each module so you can implement this immediately. | 14 comments on LinkedIn

Are you looking at the right data? Because wrong numbers lead to the wrong decisions Most teams track opens, volume, and surface-level activity None of which actually tell you if outbound is working And when you track the wrong numbers, you misdiagnose everything - Weak messaging looks fine -Bad lists slip through -Follow-up gaps stay hidden -And your pipeline stays unpredictable Track the metrics that drive revenue: -Reply Rate →message market fit -Positive Reply Rate → real intent -Meeting Conversion → follow-up strength -Lead-to-Opportunity → outbound quality -List Accuracy → garbage in is garbage out -Time-to-Meeting → engine speed -Follow-up Completion → consistency Track them and outbound becomes predictable Save this so you never track the wrong metrics again If your outbound isn’t working DM me “SYSTEM” Follow Nick Palasz for more | 66 comments on LinkedIn

AI is about to break outbound in 2026 unless you use it the right way Most teams use AI to write emails but the winning teams use it to read the moment Here’s the difference 👇 AI should: – translate signals into buying intent – cluster prospects by real-time movement – route replies based on intent – personalize to what’s happening now – keep sequences from going stale AI should NOT: – choose your ICP – set your strategy – replace human judgment – write unchecked copy – pretend it understands relationships Here’s the 2026 rule: AI is the engine. Humans are the driver If you want my full AI Outbound Playbook 2026 Comment “AI” and I’ll send the resource in pdf version | 90 comments on LinkedIn

If you’ve ever questioned your career after a cold email this carousel is for you Because let’s be honest outbound has a way of humbling even the most confident rep 😅 One minute you’re convinced today’s the day the next you’re staring at a lead list that looks like it survived a hurricane, refreshing your inbox like it owes you money, and sending follow-ups you’ll definitely think about at 2AM. These are the 7 stages of outbound we ALL go through (even if we pretend we don’t) 👇 If it stings a little, that’s how you know it’s real Follow Nick Palasz for more | 78 comments on LinkedIn

The 5 solutions I wish I’d had when I started my sales career 15+ years ago (that help me consistently book 70+ meetings/month) 1 — FullEnrich Your outbound dies the moment your data is wrong. FullEnrich fixes that before your reps even touch a lead. → Cleans & verifies every contact → Confirms emails & phone numbers → Removes dead records instantly 2 — Perplexity Real-time intelligence while you browse a prospect’s world. → Scans sites, posts, PR, product pages → Surfaces buying triggers + priorities → Generates POVs instantly 3 — RetroFix (YC S24) Call prep that feels like cheating. → Auto-scans your calendar → Builds insight briefs for each meeting → Delivers pains, angles, talk tracks 4 — Alta | AI Revenue Workforce Outbound that reacts the moment intent spikes. → Outbound agent triggers personalized messages → Inbound agent qualifies warm leads instantly → Both adjust messaging based on behavior 5 — Instantly.ai High-personalization at scale without burning domains. → Behavior-based sequences → Smart send logic → Maintains deliverability If you want more AI-native outbound systems follow Alan Ruchtein, the playbooks only get sharper from here. | 63 comments on LinkedIn

How to Write the Perfect Breakup Email for Cold Outreach

Growth Engine X is an outbound agency that sends 8,000,000 emails per month on behalf of their clients. We specialize in data segmentation, AI messaging, and creative campaigns.

This is the exact cold email system that generated $1.45 million in revenue and $7 million in pipeline for one of our many clients over 22 months. I'm going to breakdown in depth and reveal the complete strategy, infrastructure, and metrics behind consistent monthly results using only cold email outreach.​​

I'm revealing the secret to high performing cold email campaigns with subject lines that actually get your emails opened! In this video at StackOptimise shares two proven strategies for crafting the perfect cold email subject line drawing from real data, high-volume campaigns, and advanced personalization using Clay and Smartlead.

I've sent ~15 million cold emails in 2025 so far. Yeah, you read that right. FIFTEEN MILLION. And here's everything you need to know about what ACTUALLY works in 2025: | 22 comments on LinkedIn