Ivan Xu

The most useful AI prompt I use for B2B outreach: "You are writing to [Name], [Title] at [Company]. They recently [specific trigger — funding/hire/post]. Their likely pain: [1 sentence]. Write a 3-sentence cold email opening. Voice: direct, no adjectives, one specific number. Do NOT mention our product yet." Output quality = prompt specificity. Most people skip the context. That's why their AI copy sounds generic.

Most companies waste their trade show budget. Not because trade shows don't work. Because they prepare wrong. THE PROBLEM: Teams show up to events reactive. → Booth set up the week before → Prospect list built the morning of → Outreach starts on the show floor By then, competitors who prepped 4 weeks out have already booked the meetings. THE MATH: 4 weeks before the show → high response rates (prospects planning agenda) 1 week before → medium (inbox getting noisy) Show floor → low (everyone is overwhelmed) 80% of ROI is decided in that 4-week window. THE PRE-SHOW PLAYBOOK (what top exhibitors do): WEEK -4: Intelligence → Pull full exhibitor + attendee list → Identify ICP matches (title, company size, industry) → Research each: recent news, job posts, LinkedIn activity WEEK -3: Personalized outreach → First touch: reference something specific about them → NOT "hope to see you at the show" → YES "saw you just expanded to EU — we help teams do X there" WEEK -2: Follow-up + calendar push → Send calendar invite with clear agenda → Offer specific time slots, not "let me know when works" WEEK -1: Confirm + prep briefs → Confirm all meetings → Brief your team: who are they, what do they care about RESULT: Walk in with 15 booked meetings instead of 0. This is exactly what we're building at @Lensmor_ai — the pre-show intelligence layer that makes this repeatable. If you exhibit at 3+ shows/year, what's your current prep process?

Unpopular opinion: most B2B cold outbound is just noise. Not because outbound doesn't work. Because 95% of it ignores timing. Sending a cold email to someone who: → Just raised a Series A → Just hired a new VP of Sales → Just posted about switching tools ...is not cold. It's perfectly timed. The difference between spam and pipeline is context.

Something I've learned building @Lensmor_ai : The features users ASK for are rarely the ones they actually need. In 15+ discovery calls, every event marketer said: "We need better AI matching." What the same people couldn't do without once we shipped it: Bulk export with custom field mapping. The unsexy feature became the stickiest feature. When users describe symptoms, dig for the workflow underneath.

Most B2B teams use LinkedIn wrong. They post content and wait. What AI-powered teams do instead: → Monitor when prospects get promoted → Track when target companies post new job reqs → Alert when a competitor is mentioned in their posts These are buying signals. React to signals. Don't broadcast into the void.

B2B buying cycles average 6-12 months. But here's what most teams miss: 70% of that time is spent in the "dark funnel" — prospects researching without filling a form, talking to peers, lurking in Slack groups. By the time they contact you, they've already decided. Win the dark funnel, win the deal.

Counterintuitive take: the faster things move, the more long-term thinking matters — just at a different layer. Plan in 90-day cycles, sure. Tools, tactics, and execution should be that fluid now. But your principles should stretch 10+ years: what problems you care about, who you serve, how you sharpen judgment and taste.

The last 18 months changed more than the previous 10 years. The world is moving at an unprecedented speed I stopped making plans longer than 90 days in a world where new AI models/tools are dropping like rain That's the pace of AI/tech right now I think it's only accelerating

I replaced 60% of my B2B marketing workflow with AI. Here's the exact stack + prompts + the money I saved. BEFORE AI: → 1 BDR doing 50 cold emails/day → 3 hrs/day on prospect research → 1 content writer for blog → $12K/month total cost AFTER AI STACK: → 1 person managing 500+ personalized touches/day → $3K/month total cost → 4x output THE STACK — 4 TOOLS, NOTHING FANCY: TOOL 1: Clay ($349/mo) For: prospect enrichment + AI-personalized first lines Uses: scrape LinkedIn, enrich with 50+ data points, let Claude write opening lines per prospect TOOL 2: Claude API (via Cursor) For: email sequences, landing page copy, blog drafts My daily prompt template: "You are writing to [persona] at [ICP]. They feel [pain]. Write in [voice] — short sentences, one specific number, no adjectives." TOOL 3: HubSpot Breeze (or free: n8n + OpenAI) For: lead scoring + nurturing automation Setup: score on behavior (not demographic), auto-send different sequences per score band. Result: response rate +180% vs generic nurture. TOOL 4: Apollo + 6sense For: intent detection Rule: prospect visits pricing 2x + competitor mentioned → instant Slack alert → human rep takes over. AI sets the table. Humans close. THE META-LESSON: AI is terrible at strategy. AI is amazing at execution at scale. Keep humans for: → Positioning → Deal qualification → High-stakes conversations Delegate to AI: → Research → Drafting → Follow-ups COMMON MISTAKE FOUNDERS MAKE: Ask AI "write me a cold email" → get garbage → give up. Correct approach — feed AI: → Your ICP doc → 5 winning email examples → Specific prospect context Output quality = input specificity. MY ADVICE FOR B2B FOUNDERS UNDER $1M ARR: Don't hire a marketer. Hire an AI operator. Someone who can: ✅ Wire up Clay + Claude + HubSpot ✅ Write prompts, not essays ✅ Measure reply rates, not vanity metrics They're worth 3 traditional marketers. I'm building Lensmor using this stack — sharing what I learn along the way. Follow if you want the templates.

After 20+ user interviews for Lensmor, one pattern is crystal clear: B2B users don't want "smart AI features" They want friction removed from workflows they already own. Case in point — we scoped "AI smart matching" as a flagship feature. Interviews revealed: they just needed reliable CSV export with custom fields. Shipped the boring thing. Users loved it. Build for the unsexy workflow. Innovate their time back.

Most B2B SaaS companies burn $50K+ on "content marketing" and wonder why leads don't come. Here's the 2026 framework that actually works: THE OLD MODEL IS DYING Publish 4 blogs/month → hope for SEO → wait 6 months → maybe get leads. Problem: Google's AI Overview now answers 60% of queries without sending clicks. SEO alone is dying. Fast. THE NEW MODEL — 3 LAYERS LAYER 1: GEO (Generative Engine Optimization) Get cited by ChatGPT, Perplexity, Gemini when buyers ask: "best [category] for [use case]" How to win: → Publish structured comparison pages ("X vs Y vs Z for [persona]") → Get 3rd-party citations (G2, Reddit, review sites) → Schema markup for product data → Clean H2/H3 hierarchy (AI crawlers love it) LAYER 2: Signal-Based Outbound Stop blasting cold emails. Trigger on intent: → New funding raised → Key hire announced → Competitor mentioned in job posts → Tech stack change detected Stack: Clay + Apollo + Ocean LAYER 3: Community Presence B2B buyers research in Slack, Reddit, LinkedIn DMs — NOT your blog. Show up where they are: → Answer 5 niche questions/week → Share 1 original framework/week → Zero links, zero pitches THE COMPOUND EFFECT Month 1: nothing happens Month 3: people tag you Month 6: "saw your post about X" DMs Month 12: inbound > outbound RUNNING THIS FOR LENSMOR: → 22 SEO articles indexed in 2 months → DR 0 → 16 → Early inbound signals starting to come in Still early. The flywheel is spinning. TL;DR — THE 2026 B2B GROWTH STACK ✅ GEO (not just SEO) ✅ Signal-based outbound (not spray-and-pray) ✅ Community presence (not just publishing) Companies still running 2020's playbook will die in 2026. If this helped: → Follow @Ivan808225 for daily B2B + AI growth breakdowns → RT the first line to help a founder who needs it

How Notion went from $0 to $2B without a sales team: 1. Built for individuals first (bottoms-up) 2. Let teams form organically around personal users 3. Only then added enterprise features The lesson: PLG works when the product creates visible collaboration. If your B2B product is used alone, PLG will fail.

Most B2B teams are still sending generic cold emails. Here's what AI-powered outreach actually looks like: → AI crawls prospect's recent LinkedIn posts → Detects their tech stack + recent funding signals → Writes a hyper-personalized first line Reply rate: 3% → 12%. The unfair advantage isn't budget. It's timing.

12 weeks into building Lensmor. 3 patterns from conversations with event marketers + sales teams: 1/ The 4 weeks BEFORE a trade show decide 80% of the ROI 2/ Most teams still build exhibitor/attendee lists manually (8-12 hrs per show) 3/ By opening day, top prospects are already booked by competitors who prepped earlier We're building for #3 — pre-show intel that turns "who's attending" into booked meetings before the doors open. If you exhibit at 5+ shows/year, when do you start prospecting?

Unpopular opinion: your CRM is lying to you. Lead scoring based on "downloaded a whitepaper" is useless. What actually predicts intent in 2026: • Competitor research behavior • Niche community questions (Reddit, Slack) • Job postings showing pain points Signal-based > score-based. Always.

GEO just became measurable. Bing Webmaster Tools launched AI Citations Report—the first official AI search analytics with queries, top pages, and citation data. This changes everything. You can't scale what you can't measure. GEO is now a real growth channel, not an experiment.

@Jarodxu7 happycapy

Excited to launch happycapy, an agent-native computer on your browser. Claude Code → Clawdbot → Happycapy > Private sandbox for everyone to run Claude Code anytime anywhere - no Mac Mini needed > GUI built for everyday user, visualizing skills and its output

Ideally, we want to drive the car, but habitually, we keep walking in front of it with a red flag. The transition from 'copilot' (bicycle) to 'agent' (steam/steel) is exactly where the friction lies today.

The barrier to creating content also dropped. The internet is flooded with fragmented information. Doing manual competitor research today feels like trying to drink from a firehose. It's exhausting. It's inefficient. And it scales terribly.

As founders, PMs, and PMMs, we're all asking the same questions: "What makes us different?" "How do we win without a price war?" To answer this, you need Competitive Intelligence. But here's the problem I faced...