Mateusz Sekta

Mateusz SektaSep 10, 2026

I tested dozens of outbound sequencers in the last 6 months. Over the last 2 years, I did hundreds of outreach campaigns. The 4 most important things for me were always: - technical - infra, building sequences, reply agents, API, integrations, - ⁠support - responsiveness, communication, fixing problems - pricing - subscription, credits system, or pay-per-usage - trends - processes agentification, MCP, unified systems for outbound So, based on these ~2 months ago we switched the sequencer and migrated fully to Salesforge.ai 🔥 Last week, Niels DM'd us with a proposal about a partnership. So now, I'm happy to announce that Vanderbuild is a Forge Expert! Why did SF stand out? - you pay per action, not per seat - usage based pricing in outbound - Krzysztof Szyszkiewicz would be proud - LinkedIn senders are unlimited (on the same credits pool) - better for usage management and scale once you have your PMF validated - multichannel without a 2nd tool - especially usefull when you want to transfer custom variables for the message in 2 separate channels: email + linkedin And also, costs scale with output, value and not with more accounts connected more senders = more sends = more potential leads = more bacon 🥓 Can’t wait to expand our coop with Frank and his Forge team!

Mateusz SektaSep 16, 2026

Think about your last three CLOSED deals. (Not the pipeline. The ones that actually closed.) My guess: at least two came from a referral, or from someone who already knew who you were before you pitched them. Now think about what you spent on lead generation last year and how few of those deals felt like that. That gap is four truths about how B2B buying actually works: 1. Only 5–10% of your market is buying right now. And you realistically reach about half of them. So one campaign talks to a few percent of your market. One campaign is mathematically not enough. 2. 1,000 companies → ~80 in-market → ~40 reachable → 2–4 closed. The other 996 didn't leave the market. They'll be in it in 3, 6, 12 months. The question is whether you're still there and how visible you are to them. 3. Email → Gmail. CRM → HubSpot. Nobody evaluated those. The name was already stored next to the problem. Which word should pull up yours? 4. Association gets you remembered and trust gets you paid. 8-14 touchpoints for a smaller company. Closer to 38 for enterprise. A referral is just trust somebody else built on your behalf. So the fix is never a bigger campaign. It's a recurring one. Same market, 12-24 months. I recorded a full breakdown of this. It's the first video I'm sharing here from Vanderbuild's new YouTube channel (link in the comments). The channel is new for us, and now it gets a huge boost with Krzysztof Kurpiewski. The edits are his work - let us know if we do good job :D We'll be recording more content like this. What would you change? Any tips for us?

Mateusz SektaSep 11, 2026

Jumping straight into automation is a trap. Building complex workflows for unproven triggers usually ends in a mess. At Vanderbuild, we mapped out countless data flows over the years. The real game-changers always share the same foundation: - proving the concept manually - structuring the database - automating only what actually works That’s how our internal process was born: The Signal Validation Framework 🛠️ Here is the exact 5-step path we follow: 01 Signal hypothesis Mix your internal HubSpot, Pipedrive, Salesforce data with external triggers. 02 Manual test Let your reps send messages directly without any tech leverage. 03 Read the result Export data to CSV or JSON or connect sequencer via MCP and let Claude analyze the replies. 04 CRM ready Clean up the base, enrich profiles using e.g. Clay, and push back to CRM. 05 Recurring system Connect AI agents from @Codex ChatGPT or Claude Code and unified enrichment and research API to run the validated flow on autopilot. The main takeaway here is simple: You shouldn't write a single line of code until your team manually gets positive replies. If someone else told you that it will work ask for confirmations in numbers. It is always cheaper to spend hours or days on confirm that something does not work or works and will be fine tuned and implemented.