In partnership with
1,000+ Proven ChatGPT Prompts That Help You Work 10X Faster
ChatGPT is insanely powerful.
But most people waste 90% of its potential by using it like Google.
These 1,000+ proven ChatGPT prompts fix that and help you work 10X faster.
Sign up for Superhuman AI and get:
1,000+ ready-to-use prompts to solve problems in minutes instead of hours—tested & used by 1M+ professionals
Superhuman AI newsletter (3 min daily) so you keep learning new AI tools & tutorials to stay ahead in your career—the prompts are just the beginning
We track Product so you don't have to. Top Podcasts summarised, the latest AI tools, plus research and news in a 5 min digest.
Hey Product Fans!
Welcome to this week’s 🌮 Product Tapas.
New here? We're your shortcut to staying sharp. Essential stories, practical tools, real insights.
For the best reading experience, check our web or app version and sign up for future editions here.
What’s cooking this week? 🥘
The layoff dominoes keep falling - Meta's planning to cut 20% of its workforce to fund AI while its own model "Avocado" can't pass internal tests (nothing says confidence like firing the humans to pay for robots that don't work). xAI's on its second rebuild in a month (nine of eleven co-founders gone - at what point do you just start a new company?), OpenAI's ditching side projects ahead of an IPO, and Nvidia's projecting $1 trillion in chip sales at GTC. Meanwhile Google shipped five major products in a single week while nobody was looking and Anthropic's hiring a chemical weapons expert (responsible AI gets very literal).
📰 Not Boring → Meta's layoff yo-yo, Nvidia's $1T bet, xAI's rebuild, OpenAI's IPO pivot, Google's five-product week
⌚️ Productivity Tapas → AI-native CRM, task-breaking toolkit, plain-English BI dashboards
🍔 Blog Bites → LinkedIn tactics that actually work, messy docs beat rigid systems, self-compounding Claude workflows
🎙️ Pod Shots → Jason Lemkin: 1.2 humans + 20 AI agents = same sales performance
Let's go 🚀
📰 Not boring
The Big Reshuffles
Meta planning 20% layoffs to offset $600B data centre spend - while its AI model Avocado flops on internal tests and China calls in executives over the Manus deal. Meanwhile companies not seeing AI revenue gains may be oversized for their actual market.
Elon Musk says xAI "wasn't built right" - Nine of 11 co-founders gone; hired senior Cursor engineers; staff say constant upheaval is killing momentum
OpenAI pivoting - cutting side projects ahead of IPO - Fidji Simo refocusing on coding and business customers. Atlassian slashes 10% for same reason. Adobe CEO stepping down.
Meta has churned through more employees since 2019 than the company had in 2019. This is the quarter where tech collectively decided AI spending has to come from somewhere - and "somewhere" is payroll.
Nvidia's Universe
Nvidia GTC 2026 - $1T in chip sales forecast through 2027. 88-core Vera CPUs with 6x throughput (CPUs becoming the agentic bottleneck). DLSS 5 neural rendering. Space-1 orbital data centres. Jensen calls it "agents-as-a-service".
NemoClaw enterprise agent platform - safely build autonomous agents in contained environments. Disney's Olaf robot trained on Nvidia's stack debuts Disneyland Paris 29 Mar.
ASML lithography as real AI bottleneck - ~100 EUV machines/year by 2030 caps industry scaling regardless of capital
The buried lead is CPUs - if agentic workflows are CPU-bound, Nvidia's Vera chips put them against AMD and Intel in a market they weren't in two years ago. ASML's 100 EUV machines per year caps how fast this industry can scale.
Anthropic's Week (New Developments Only)
Anthropic $100M partner network for consulting firms, plus chemical weapons policy manager job posting - building enterprise distribution and domain-specific safety simultaneously
1M-token context window now GA with doubled usage limits. Cowork dispatch lets you message Claude from phone. One-person marketing operation ran all of marketing for a year.
Anthropic Skills internal playbook - hundreds across nine categories; how teams use Claude Code details the lessons learned
$100M partner network = enterprise distribution without a sales team. One person running all marketing for a $380B company is either inspiring or a red flag depending on which side of the headcount you're on.
The Agent Wars
Perplexity Enterprise Computer now connects to Snowflake, Salesforce, HubSpot via Slack; claims $1.6M saved. Manus AI "My Computer" takes over your desktop directly. Both face trust problems: judge blocked Perplexity's Comet from ordering on Amazon, and China warned OpenClaw is insecure.
How Stripe's Minions actually work (ByteByteGo deep dive) - hybrid orchestration, curated context feeds, fast feedback loops. Open SWE framework offers the open-source version: isolated sandboxes, subagent orchestration.
"Night Shift" agentic workflow - developer day shift, AI agent night shift. Perplexity Agent API makes the model layer swappable. Aaron Levie argues agents will be the primary user of all software. Anthropic says the bottleneck moved to long-horizon tasks in regulated settings.
Agents work, but only with deterministic guardrails bolted on. The companies winning are building safety rails, not promising magic.
Money Talks
GPT-5.4 Mini and Nano - Nano at $0.20/$1.25 per million tokens. Morgan Stanley says GPT-5.4 scores at or above human expert level.
Replit raises at $9B, launches Agent 4 - multiple sub-agents in parallel with infinite canvas and auto-merging. McCormick: "first good AI-native product."
Britannica suing OpenAI over ~100,000 articles. Kled AI raises $5.5M for human data marketplace. Google Fiber sold to Stonepeak.
GPT-5.4 Nano at $0.20 per million input tokens makes "AI is expensive" a legacy argument.
Consumer & Platforms
Google's five-product week: Maps biggest redesign in a decade ("Ask Maps" via Gemini, 3D navigation), Gemini Personal Intelligence free for all US users, Stitch AI design canvas, Vibe Design (UI from natural language), and Android developer verification closing the open ecosystem.
Design tools surge: Midjourney V8 (5x faster, native 2K), Gamma nearing 100M users and embedding into ChatGPT/Claude, Figma integrates into Anything for design-to-code.
Four major product moves in a single week - Google is on a quiet tear. Android's developer verification is Google slowly closing the openness that differentiated it from Apple.
The Human Side
Invisible job displacement: Tom Tunguz nails it - AI takes jobs one unposted job at a time. Karpathy published a 1-10 rubric. Meta, Netflix, Domino's using AI avatars for 100% of initial screenings. jobloss.ai tracking it all.
Hidden costs of AI speed: Comprehension debt is the gap between code that exists and code anyone understands. Every layer of review makes you 10x slower. Shopify Liquid got 53% faster via AI micro-optimisations on a 20-year-old codebase - but who understands the changes?
AI and health: ChatGPT downplays 51.6% of emergencies while 230M use it weekly for health advice. Meanwhile Copilot Health and Amazon both launching health AI platforms.
Thought pieces: AI productivity sweet spot is 50 min/day (only 3% hit it). McKinsey AI bot hacked exposing 46.5M messages. AI models are the new rebar - commodity. Ethan Mollick on AI hitting markets, jobs, and governments simultaneously.
AI isn't firing anyone - it's preventing the next hire from ever being posted. And ChatGPT downplaying 51.6% of emergencies while 230M use it weekly should be front-page news.
Bonus: Notable Launches & Signals
Rogue AI agents that lie and self-preserve (Guardian investigation). Big Tech energy pledge as data centres approach more electricity than aluminium + steel + cement + chemicals combined.
Databricks Genie Code doubled coding agent success from 32% to 77%. Okara AI CMO deploying marketing agents (6M+ views). Runway Labs pushing video beyond filmmaking. DeepMind AGI scorecard: 10 cognitive abilities, $200K hackathon.
Databricks doubling coding agent success from 32% to 77% is the kind of quiet improvement that matters more than any flashy launch.
Go from AI overwhelmed to AI savvy professional
AI will eliminate 300 million jobs in the next 5 years.
Yours doesn't have to be one of them.
Here's how to future-proof your career:
Join the Superhuman AI newsletter - read by 1M+ professionals
Learn AI skills in 3 mins a day
Become the AI expert on your team
⌚ Productivity Tapas: Time-Saving Tools & Workflow Automation
Lightfield - AI-native CRM that builds itself from your
email/calendar/Slack. From the Tome founders. No data entry, just ask
questions about your relationships.Goblin Tools - Free AI toolkit that breaks overwhelming tasks into
manageable steps, estimates time realistically, and turns brain dumps into
structure. Originally for neurodivergent users, useful for everyone.Supaboard - AI-native BI with 600+ integrations. Ask data questions in
plain English, get instant dashboards. Hit #1 on Product Hunt in Feb.Remember. Product Tapas subscribers get our complete toolkit - 550+ personally tailored, time-saving tools for PMs and founders. Your shortcut to efficiency and what's hot in product management 🔥
Check the link here to access.
🍔 Blog Bites - Essential Reads for Product Teams
Strategy: Elevate Your LinkedIn Game with Tested Tactics
Will Leatherman shares 21 practical strategies for optimising LinkedIn posts to generate meaningful engagement and drive sales pipeline growth. The article emphasises actionable tactics over generic content, advocating for specific approaches that have shown results in real B2B contexts. Read the full piece here
Key Takeaways:
• Craft Compelling Hooks: An engaging hook can significantly improve post visibility. Write three variations for each hook and choose the most attention-grabbing, reinforcing the need to treat it as a headline in a paid ad campaign.
• Use Specificity in Content: Concrete numbers and relatable stories are more impactful than vague advice. Sharing personal experiences enhances credibility and connection with the audience.
• Visual Authenticity Matters: Using real photos instead of stock images creates a human feel that resonates with audiences. Showcasing authentic team moments can lead to greater engagement.
• Optimise for Distribution: Successful posts require as much focus on distribution as on content creation. Creating Thought Leader Ads for top-performing organic posts can significantly amplify reach.
• Engage with the Community: Thoughtfully tag relevant experts or brands to foster genuine connection and expand your post’s reach. The goal should be meaningful engagement rather than seeking attention through spam tagging.
• Embrace Trial and Error: Implement a mindset of constant testing and iteration. By tracking what resonates with your audience, you can refine your strategy and improve your content effectiveness over time.
Strategy: Embracing Confusion to Enhance Team Performance
John Cutler discusses the power of "messy documents" in facilitating effective team collaboration at Dotwork. By favouring real examples and dynamic information sharing over rigid systems, teams can adapt and thrive in the complexity of product development. Read the full article here.
💡 "The work is messy, and that’s no surprise. That’s product work."
Key Takeaways:
• Messy Documentation: High-performing teams utilise freeform documents that reflect real-time insights, connections, and ideas, allowing for a richer understanding of evolving projects without the constraints of rigid structures.
• Externalising Cognitive Load: By transferring thoughts from individual minds to shared documentation, teams alleviate pressure and maintain clarity, ensuring that critical information is captured and easily revisited.
• Frequent Integration: Regular updates and iterations of documents help teams avoid stagnation and keep critical discussions alive, preventing information from easily falling through the cracks.
• Survivorship Bias: The ability to maintain messy, evolving documentation isn't necessarily what makes teams high-performing; rather, it may indicate that they possess the patience and adaptability necessary to thrive amidst ambiguity.
Productivity: Unlocking the Power of Self-Improving Systems
You know how much I love Dex; these self-compounding systems are 100% the future. Recently Paweł Huryn shared insights from building a self-improving agentic system using Claude, reducing his workload from four hours to just 30 minutes a day. His self compounding system led to significant audience engagement, illustrating the architecture's value over mere content. Read the full article here.
💡 "The more you use Claude, the more it compounds. Structure emerges."
Key Takeaways:
• Compounding Growth: By using Claude strategically, Huryn experienced a 5.2 million impressions and a 7.2% engagement rate in just three months, showcasing how a structured approach to content can significantly boost visibility and interaction.
• Real-World Learning: The system learns from actual data and continuously adjusts by identifying patterns, making it applicable for various uses such as customer research and market intelligence, where insights can surpass conventional methods.
• Knowledge Architecture: Huryn developed a file-based knowledge graph that simplifies knowledge management and streamlines task execution, providing a cohesive structure that Claude can navigate effectively.
• AI as Assistance, Not Replacement: Claude enhances efficiency by handling research and data organisation, but crucial decision-making and creative input still rely on human judgement, preserving the value of expertise in the process.
• Progressive Disclosure: The design allows Claude to access only the necessary information for each task, preventing overload and focusing on relevant data when needed — a strategy that can be replicated in various professional contexts.
🎙 Pod Shots - Bitesized Podcast Summaries
Remember, we've built an ever-growing library of our top podcast summaries (120 or so). Whether you need a quick refresher, want to preview an episode, or need to get up to speed fast - we've got you covered.
Check it out here
🤖 Can AI Agents Replace Your Sales Team?
Jason Lemkin replaced his 10-person sales team with 1.2 humans and 20 AI agents. Same business performance. He's not hiring humans in sales anymore - unless they're truly great. Here's what the future of go-to-market actually looks like.
"We're done with hiring humans in sales." Jason Lemkin, founder and CEO of SaaStr, shares a rare glimpse into the future of sales and go-to-market. His company went from 10 full-time salespeople to 1.2 humans plus 20 AI agents - with the same business performance. He's got desks labelled with agent names: Repli for Replit, Quali for Qualified, Arty for Artisan.
In this conversation with Lenny Rachitsky, Jason explains why AI is replacing the jobs people don't want to do, why mediocre performers are being displaced first, and why the future isn't no salespeople - it's $250,000/year SDRs managing 10 agents instead of 10 people. He'd still hire two great humans tomorrow, but he won't hire someone who doesn't know what SaaStr does after three months.
🎥 Watch the full episode here
📆 Published: January 2026
🕒 Estimated Reading Time: 8 mins. Time saved: 52 mins! 🔥
Key insights from the full article:
🤖 10 humans to 1.2 humans + 20 agents — Same business performance, dramatically different cost structure. The transformation happened faster than anyone expected.
⏰ Agents work all night, weekends, and Christmas — The always-on advantage compounds quickly when multiplied across 20 agents handling different functions.
📉 AI displaces the mid-pack and mediocre first — Great salespeople remain irreplaceable. But the third-month rep who still doesn't understand the product? Gone.
💰 $250K SDRs managing 10 agents — The classic junior SDR role is dead. The future is fewer, higher-paid humans orchestrating agent fleets.
✉️ No more humans for "contact me" leads — Inbound qualification, email follow-ups, and lead scoring are entirely agent-driven now.
🎯 Still hiring - but only truly great humans — If two exceptional salespeople applied tomorrow, Jason would hire them. The bar just got much higher.
🔮 This is just the beginning — Current agents handle discrete tasks. Next generation agents will handle entire sales cycles.
That’s a wrap.
As always, the journey doesn't end here!
Please share and let us know what you would like to see more or less of so we can continue to improve your Product Tapas. 🚀👋
Alastair 🍽️.