Funnel Math, Talent Drama, and the Tools I’m Actually Shipping With

Funnel Math, Talent Drama, and the Tools I’m Actually Shipping With

Feb 24, 2025

Jul 22, 2025

Jul 22, 2025

Jul 22, 2025

Jul 22, 2025

|

4

min read

Here’s a new episode of The Reddy Rundown, crafted so you don’t have to frantically follow everything in the AI news space wondering what you’re missing as an exec in 2025 trying to keep up. I’m Shawn Reddy—CEO who still thinks like a marketer. My edge isn’t just taste, it’s systems. The AI tools I wire together are the real marketing asset, and this week’s news made that painfully obvious. 1. Google’s “Official” IMO Gold: Cool Flex, But What’s the Funnel Play? DeepMind’s Gemini + Deep Think version just bagged an officially graded gold at the International Math Olympiad (35/42, 5/6 problems). OpenAI says it matched the score but didn’t run through the IMO’s formal grading process. My take: Math flexes sell “trust” and “competence” to enterprise buyers. But trophies don’t auto-convert. If I were Google’s PMM, I’d be running: • Credibility retargeting: Serve case studies to technical buyers who watched the IMO clip. • “Reasoning = ROI” bridge content: Translate benchmark wins into tangible ops wins (fraud models, pricing engines, QA bots). Benchmarks are the headline; ROI diagrams are the CTA. • Sales enablement kits: Give reps 1-slide “why reasoning matters” visuals so they stop dumping 40-page PDFs. For everyone else: don’t chase “we’re smart too” messaging. Show where your stack already lowers CAC or speeds creative iteration. 2. Alibaba’s Qwen3: Open-Source Power = Marketing Moat (If You Package It Right) Alibaba split its “thinking” vs “instruct” models and dropped a 235B-parameter, 256K-context open model that beats Kimi K2 and nips at Claude Opus 4. Why it matters to marketers: • Zero-cost sandboxing: Run massive prompt sweeps for positioning tests without burning API spend. • Context-heavy personalization: 256K tokens means whole-customer-histories in a single prompt. That’s lifecycle email copy with real memory. • Thought leadership angle: “We built on the world’s best open model” is better press than “we fine-tuned GPT again.” If you adopt Qwen3, ship a public teardown: how you integrated it, the speed delta, the cost delta. That content ranks, recruits, and reassures buyers. 3. Build-Your-Own Email Drafter Agent (Grok 4 + n8n): Stop Manually Polishing Cold Outreach I wired an agent in n8n using xAI’s Grok 4 to draft emails directly into Gmail drafts. Triggered by a chat message, stored with lightweight memory, QA’d before send. Use it like a marketer: • Auto-draft follow-up emails after webinars using event metadata. • Generate variant copy for A/B tests, push the winner to the CRM automatically. • Pipe sentiment tags back into your audience segments. Key: keep it in draft mode until your tone library is tight. One rogue email can torch a deal faster than it saves you time. 4. Sapient’s Brain-Inspired HRM (27M Params): Small Model, Big Implications Sapient Intelligence dropped a 27M-parameter model that beats giants on ARC-AGI 2 puzzles—no chain-of-thought crutches, no mega-pretraining. Marketing lens: • Edge deployments: Low-parameter models can live closer to the data source (POS systems, kiosks, in-store displays). That means hyper-local, real-time creative. • Data-light niches: If you’re in a vertical with thin data (healthcare, climate, niche B2B), these architectures might finally let you build smart agents without 10M rows of logs. This is a signal: efficiency is a message. “We don’t need your data or your GPU farm” is a privacy and cost story clients will bite on. Quick Hits & Marketing Reads • Cohere Catalyst Grants – Free model access for impact startups. → Pitch: ‘We built our attribution engine on Cohere—here’s the case study.’ • Pika’s AI-only social video app – If TikTok taught us anything, early mover creatives win. → Prototype vertical-specific channels (legal, SaaS, medtech) and own the niche. • Runway Act-Two via API – Motion capture in your own app. → User-generated motion templates for ads without a studio budget. • Netflix “10x faster” VFX via genAI – They’re already marketing the speed story. → Borrow the narrative: “We cut production time by X using Y model” isn’t bragging; it’s proof. • OpenAI’s 2.5B daily requests – Volume ≠ value. → Ask: how many of those requests become measurable lift? Track your own agent “success events,” not just usage. • xAI poaches Nvidia’s Ethan He – Talent flows tell you where frontier R&D is headed. → If your vendor’s brain drain is public, reevaluate roadmap risk. See content credentials Tools I’m Actually Testing (and How You Could Use Them) Qwen3 (open source LLM, 235B params / 256K context) What it actually does: Long‑context, high‑performance model you can run cheaply/on your own infra. Marketing use: Bulk message/angle sweeps, persona chatbots that remember full histories, cost-controlled experimentation. Link: Qwen Chat / GitHub. n8n + Grok 4 Email Agent What it actually does: Workflow that drafts Gmail emails on command, with light memory and human review gates. Marketing use: Auto-draft nurture follow-ups, variant cold outreach, push winners to CRM, tag sentiment back to segments. Links: n8n / xAI API docs. Runway Act-Two API What it actually does: Motion-capture generation you can call from your own app. Marketing use: Spin up product demo animations or UGC-style ads without a studio; templatize motion for fast creative swaps. Link: Runway API. Sapient HRM (27M param hierarchical model) What it actually does: Tiny, brain-inspired model that out-reasons bigger ones on certain puzzles—no giant datasets needed. Marketing use: On-device personalization, kiosks/POS copy tweaks, low-data vertical agents (healthcare, climate, niche B2B). Link: Sapient GitHub. Cohere Catalyst Grants What it actually does: Gives impact startups free access to Cohere models + support. Marketing use: Launch a low-cost POC, then market the grant + results as social proof and PR fuel. Link: Cohere Catalyst. The Meta-Lesson This Week Marketing in 2025 is controlled by whoever owns the ingestion → reasoning → creation → deployment loop without killing speed or attribution. Benchmarks, poaching, open-sourcing—none of it matters if your stack can’t turn news into leverage. 🛠️ Trending AI Tools • 📝 Gemini Code Assist - Google’s AI coding assistant, now with agent mode • 🤖 SOLO - Trae’s all-in-one Context Engineer for full software development • 🧠 Composite - Turn your existing browser into an AI agent • ⚙️ GEN - Create AI characters that build social media audiences end-to-end 💼 AI Job Opportunities • ⚙️ Sanctuary AI - Mechanical Engineer, Automation • ✉️ UiPath - Email Marketing Manager • 📊 Meta - Data Scientist • 🏙️ Metropolis - Market Leader Looking for a community of like-minded individuals who are interested in AI and Entrepreneurship? Join our free community here to get started: The AI Advantage Community Thank you for reading! -Shawn

Subscribe To Out Newsletter

Subscribe To Out Newsletter

Subscribe To Out Newsletter

Subscribe To Out Newsletter

Subscribe To Out Newsletter

Share It On: