Karina Canales

Visa’s new VAMP rules mean all chargebacks, both fraud and non-fraud, plus TC40 cases, are now counted. This can lead to fraud-coded chargebacks effectively counting twice, increasing risk exposure. Merchants still must proactively fight chargebacks - thankfully, chargeback alerts remain a powerful tool for this. However, fraud prevention demands even more attention. Solutions like Order Insight and CE3.0 are critical for tackling first-party misuse, though these tools require ongoing enhancements from Visa. Another key challenge is the growing volume of inaccurate TC40 cases, which significantly impacts VAMP rates. It’s crucial to watch how Visa addresses this moving forward.

23

𝗗𝗔𝗬 𝟭𝟯: 𝗣𝗲𝗿𝗺𝘂𝘁𝗮𝘁𝗶𝗼𝗻𝘀 (𝗕𝗮𝗰𝗸𝘁𝗿𝗮𝗰𝗸𝗶𝗻𝗴) 𝘠𝘰𝘶𝘳 𝘧𝘰𝘰𝘥 𝘢𝘳𝘳𝘪𝘷𝘦𝘴 𝘪𝘯 20 𝘮𝘪𝘯𝘴. 𝘛𝘩𝘦 𝘥𝘦𝘭𝘪𝘷𝘦𝘳𝘺 𝘨𝘶𝘺 𝘩𝘢𝘥 8 𝘰𝘵𝘩𝘦𝘳 𝘰𝘳𝘥𝘦𝘳𝘴. 𝘏𝘰𝘸? 𝗧𝗵𝗲 𝗣𝗿𝗼𝗯𝗹𝗲𝗺: 𝗣𝗲𝗿𝗺𝘂𝘁𝗮𝘁𝗶𝗼𝗻𝘀 (𝗕𝗮𝗰𝗸𝘁𝗿𝗮𝗰𝗸𝗶𝗻𝗴) Generate all possible orderings of a set of elements. 𝗥𝗲𝗮𝗹-𝗪𝗼𝗿𝗹𝗱 𝗜𝗺𝗽𝗮𝗰𝘁: You order food on Swiggy at 7:00 PM. Your delivery partner has 8 orders to deliver in different locations across the city. The Question: In which order should they deliver to minimize time and distance? This is the classic "Traveling Salesman Problem" - finding the optimal permutation of delivery stops. Example: Orders at locations: [A, B, C] All possible delivery routes (permutations): A → B → C A → C → B B → A → C B → C → A C → A → B C → B → A The algorithm calculates distance for each route and picks the shortest one. 𝗛𝗼𝘄 𝗕𝗮𝗰𝗸𝘁𝗿𝗮𝗰𝗸𝗶𝗻𝗴 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲𝘀 𝗣𝗲𝗿𝗺𝘂𝘁𝗮𝘁𝗶𝗼𝗻𝘀: Start with empty route, keep adding locations: -> Add A: [A] → Add B: [A, B] → Add C: [A, B, C] ✅ -> Backtrack, try C: [A, C] → Add B: [A, C, B] ✅ -> Continue for all combinations... Time Complexity: O(n! × n) - n! permutations, n time to build each Space Complexity: O(n) - recursion depth 𝗪𝗵𝗲𝗿𝗲 𝗲𝗹𝘀𝗲: -> Zomato/Swiggy route optimization -> Amazon delivery fleet management -> Uber pool ride sequencing -> Tournament bracket generation -> Password cracking (trying all combinations) 𝗪𝗵𝘆 𝗶𝘁 𝗺𝗮𝘁𝘁𝗲𝗿𝘀: With 8 orders, there are 40,320 possible delivery sequences. Checking all manually is impossible. Backtracking algorithms explore these permutations intelligently, finding the optimal route in seconds. Swiggy's algorithm doesn't just find any route - it finds the route that gets your food to you hot and fast, while the delivery partner earns maximum orders per hour. 𝗧𝗵𝗶𝗻𝗸 𝗮𝗯𝗼𝘂𝘁 𝗶𝘁: When your delivery arrives on time despite the driver having multiple orders, that's permutation algorithms working behind the scenes to optimize everyone's delivery experience. Day 14 drops tomorrow. #DSAMeetsImpact | Day 13/30

5

Delivery Hero H1 Financials 📈: * Accelerated GMV (total sales) to €24.6bn (+11% YoY) * Accelerated Revenue to €7.2bn (+27% YoY) * Grew EBITDA to €411m (+70% YoY) * Free Cash Flow to €-8m (+€96m) * Generated our first operating profit It was a fantastic half-year, and I’m particularly happy to see the improved growth in Asia, where much of our focus has been. 🚀 Looking into H2, we see significant headwinds from a weakening USD and Korean Won, impacting our EBITDA by approximately €110 million, reflected in reduced guidance from €975–1,025 million to €900–940 million. Without these FX headwinds, we would have achieved between €1,010–1,050 million. GMV and revenue are reported on a like-for-like basis and in constant currency.

1,647

42 Comments

Malgosia is speaking at AI Innovation Asia, an executive-level AI summit organized by Economist Impact (part of The Economist Group) She leads the "Core Tech" team at Match Group, which includes their global AI, data, and platform divisions. From that vantage point, she will share firsthand insights into how consumer applications are adopting AI at scale and what those changes mean for individuals & societies generally. But also, the impacts on LOVE AI is reshaping every aspect of life, including how people meet and form relationships. As the executive responsible for AI at the world’s largest dating company, Malgosia sits at the unique intersection of AI & love at this very interesting time in human history. You should hear her stories! https://lnkd.in/gHa8X48F

20

1 Comment

1/ 🔥 Aug ’25 AI Top 100 (mobile): creative tools are the growth engine. ByteDance Xinghui +834% MoM rev; Kuaishou KLING AI +330%. Why are creation apps winning? 2/ 📊 Scale is concentrating at the entry points: MAU leaders—Baidu AI Search 651M, ChatGPT 604M, Gemini 282M, Doubao 207M. “Entry” still captures daily reach. 3/ 💰 Revenue: Top100 >$6B/yr. ChatGPT ~$248M Aug IAP anchors the market. The “middle” is creative + edu/assistants (FaceApp, Remini, Photoroom, Speak, etc.). 4/ 📈 Growth pockets: • Music/audio—Splice (global MAU ~+52% MoM), djay • Images—ToonMe, SeaArt AI; in CN: Xinghui, Jimeng AI • Video—KLING AI downloads nearly doubled • Go-global—Seekee, Cici AI, Hypic, UpFoto 5/ 🧭 Takeaways: acquisition ≠ retention. Win with one-click outcomes, asset ecosystems (templates/style packs), and UGC loops. Full rankings + open data: 100aiapps.cn | by UNIQUE RESEARCH #GenAI #ChinaTech https://lnkd.in/g2wtHsm8

3

Excited to announce that Acceler8 has raised our seed round led by Andreessen Horowitz (a16z speedrun), with participation from Golden Gate Ventures, and incredible group of angels including Moses Lo (Xendit), Peng Ong (Monks Hill Ventures), Founders of Multiplier, senior leadership at GoTo, and investors from DST Global, among others. Acceler8 is real-time intelligence for workforce decisions inside large enterprises. Finance runs on real-time data. Sales runs on real-time data. Yet for the most valuable asset in any enterprise – its people, decisions happen based on calendar. Workforce planning, succession planning, and promotions are still largely calendar-driven, powered by spreadsheets, static HR systems, and periodic consulting despite over a $100 billion spent annually. In the AI era, that lag is no longer sustainable. We’re building the intelligence layer to close that gap — giving leaders continuous, evidence-based visibility into who to develop, who to promote, and where succession risk exists. We already work with Fortune 500 customers, with a strong enterprise pipeline. a16z speedrun team and the broader Andreessen Horowitz platform behind the program has been a powerful accelerator for Acceler8. Thank you to our customers who trusted us early, our team that executes at unreal pace, and our earliest backers & advisors. And lastly, grateful to be building this with my amazing co-founder Trisha Pathak, whose experience on workforce strategy with F500 gives us unfair advantage. In the generational shift driven by AI, workforce intelligence will become mission-critical. I couldn’t be more excited to build the platform that helps enterprises win. If this resonates, DM or chinmay@useacceler8.com Read full release here: useacceler8.com/about-us Marcus Segal, Sam Shank, Macy Mills, Lester Chen, Andrew Chen, Andrew Lee, Jonathan Lai, Marigold Vu, Tom Hammer, Jordan Mazer, Jordan Carver, Bella Nazzari, Emlyn Thompson, Emily Bennett, Joshua Lu, Samira Behrouzan

583

211 Comments

🚀 Just launched a full-stack production website from scratch. https://lnkd.in/gNgtFR4D is live — a luxury corporate + e-commerce platform for a premium noodle brand serving retail, B2B, and OEM clients. 🛠 Stack: Next.js 16 · React 19 · TypeScript · Tailwind v4 · Framer Motion · Supabase · Razorpay · Vercel 🎬 The homepage plays 192 JPEG frames on an HTML Canvas driven entirely by scroll position — a cinematic scrollytelling experience at 60fps. First-time visitors get auto-scrolling onboarding that cancels the moment they touch the screen. 🔥 Major problems I faced: 1. Blank Screen on First Visit 192 images = ~3.7MB. Auto-scroll fired before images loaded — broken animation. Fix: loading overlay until frames are ready + preloaded first frame in <head> + immutable cache headers. 2. Google OAuth Crashing the App Google Sign-In inserted Unicode characters into cookies. Next.js Server Components couldn't parse them — white screen. Fix: edge middleware that sanitizes every cookie value with a regex. 3. signOut() Hanging Forever Browser's Navigator Lock API caused signOut() to hang on some devices. Fix: Promise.race with 2s timeout + manual cookie/localStorage cleanup as fallback. 4. Payment Security Razorpay callbacks can be spoofed. Fix: server-side HMAC-SHA256 signature verification before recording any order. ⚡ Beyond the storefront: — Admin panel with analytics dashboard, product management, CMS, orders, leads, newsletter, coupons — Headless CMS via JSONB — admin edits go live instantly — Custom analytics tracker (no third-party scripts) — B2B lead gen pages for Distributors, Private Label, OEM — Row Level Security across 8 Supabase tables — E2E Cypress tests for the full purchase flow 📌 Building for production is a different game. Corrupt cookies, hanging promises, race conditions, payment forgery — none of this shows up in tutorials. Ship it. Break it. Fix it. 🔗 https://lnkd.in/gNgtFR4D 💻 https://lnkd.in/gt3BWAMJ #webdevelopment #nextjs #react #typescript #supabase #fullstack #buildinpublic #launch

17

2 Comments

Classical PM grew out of a world where simply understanding "what to build" carried enormous transaction costs. To formulate a hypothesis, you had to conduct dozens of interviews, manually synthesize insights, write a 20-page PRD, align it with five stakeholders, prioritize through RICE, compile a competitive analysis, and draft a roadmap. All of this was deeply artisanal work that demanded a very specific skill: the ability to translate the chaos of reality into structured artifacts. And it was precisely this ability that served as the main barrier to entry into the profession. Now LLMs are doing the bulk of this work: interview synthesis, user story generation, competitive analysis, spec writing, backlog prioritization. A startup founder who used to hire a PM because they couldn't structure their own product chaos now feeds their notes into Claude or ChatGPT and gets a perfectly decent PRD in minutes. The bulk of what junior and mid-level PMs did was essentially the work of a data entry operator. Packaging information into a polished format. Translating from the language of business into the language of engineering. Turning the obvious into Jira tickets. This layer is already visibly melting away. Why hire an associate PM to spend a week writing a spec when a GPT wrapper produces a draft in minutes and a senior PM polishes it in an hour? But there's a nuance. The real value of a product manager was never in the artifacts. It lay in three things: 1. Taste - an intuitive sense of what the market needs before the market itself realizes it; 2. Political navigation - the ability to push the right decision through an organization where everyone has their own agenda; 3. Ability to say "no," because 90% of product work is not about what to build, but what NOT to build. An LLM doesn't know how to say no, it's a yes-machine. It has no stance, no accountability for outcomes. And that's exactly why product management won't disappear, but it will thin out dramatically and transform — much like those camera operators who no longer need to understand the chemistry of film development, but still need to see the shot.

24

4 Comments

Thrilled to announce that Together Fund is investing in Sentra, alongside a16z speedrun! You track results in Jira. Decisions in Notion. Conversations in Slack. But the reasoning, the debates, trade-offs, and context behind why you chose A over B, disappears into what we call "Dark Matter." A decision made in March looks insane by July because no one remembers the constraints that made it smart. I lived this firsthand at Twitter scaling from 800 to 8,000 employees, and at Google while launching AI Overviews to billions at planet scale. The problem isn't process. Process is compensation for something deeper: organizational amnesia. An organization’s "Systems of Record" doesn’t solve this, they encode it. They store what happened, never why. That's why we are investing in Sentra. Sentra is the always-on collective memory that eliminates organizational amnesia by maintaining accurate context for all members and agents, functioning as an operational nervous system. It connects to every channel where work happens, meetings, Slack, email, code commits, docs, calendars, and treats them not as artifacts to search, but as living signals to synthesize. The fleeting and the permanent, unified into a memory that understands. The founding team is built for this: - Jae Gwan Park (CEO): Product-first founder, memory systems research at UofT and MIT - Ashwin Gopinath (CSO): Former MIT professor, created "Reflexion" (NeurIPS 2023), agents that learn from mistakes, 2x founder - Andrey Starenky (CTO): Early Vapi engineer, ex-IBM, built to process enterprise-scale data firehose Together is an operator-led fund. We invest in problems we've lived. This is one of them. Many congrats Jae, Ashwin and Andrey, we are so excited to partner with you! Read the full thesis: https://lnkd.in/gixj9cE4 Book a demo: https://www.sentra.app/ #OrganizationalMemory #AI #Sentra #TogetherFund #a16z #ContextGraphs

71

3 Comments

**Big Move in the Delivery Space: Delivery Hero and Grab** The rumors are now reality. Delivery Hero has officially agreed to sell its foodpanda Taiwan operations to Grab for $600 million in cash. This is a massive shift for the Asian tech ecosystem. Here’s the breakdown: - For Delivery Hero: This is a strategic "de-risking" move. By divesting a high-value asset, they are strengthening their balance sheet, paying down debt, and sharpening their focus on core global markets. - For Grab: This is a bold leap outside their Southeast Asian stronghold. By acquiring a business with a €1.5B GMV (2025), Grab is doubling down on its "Super-App" ambition in a high-density, high-spending market. - The Numbers: The $600M deal is set to close in H2 2026, pending regulatory green lights. The Bigger Picture: We’re moving away from the "growth at all costs" era. This deal signals a new chapter of market consolidation where the biggest players are prioritizing sustainable profitability and dominant market share over fragmented expansion. Will this spark a chain reaction of M&A in other regions? Time will tell. What’s your take? Is Grab’s entry into Taiwan a masterstroke or a risky expansion? Let’s talk in the comments! https://lnkd.in/gT9tTE6m https://lnkd.in/giYtfeTk

21

Kuaishou announced that Kling's annualized revenue run rate hit US$240mn in Dec 2025, and monthly revenue hit USD20mn. We estimate that 4Q25 Kling revenue is likely to be in the Rmb350-400mn range, above its guidance of Rmb300mn and our forecast of Rmb350mn. This would mean a re-acceleration of revenue growth in 4Q (QoQ incremental revenue was Rmb50mn in 3Q, and Rmb50-100mn in 4Q).

1

This a seductive idea; that organization's context fragments, decision logic gets lost, and an organization is prone to tackle collective amnesia. BUT this comes with a lot of caveats!! - MONOCULTURE: When one system becomes the authoritative interpreter of "why things are the way they are", it will collapse pluralism. An organization actually benefits from various "contested memories". An engineer and sales guy remember different things from a discussion and that helps. Employees can stop arguing with the system not because its correct but its "coherent". - OBSERVER's EFFECT: When people know that the reasoning within systems is being captured, they will change what gets recorded. The negotiations will become "safer" and will contaminate raw signals. - CENTRALIZATION: Interpretation gets centralized upwards. It may appear leadership or executives are gaining synthesis and becoming closer to the ground. But there is no comparison to living through the frictions, not even eclectic summaries. LEADERSHIP MUSCLE NEEDS TO PRACTICE LIVING THROUGH FRICTION OTHERWISE IT WILL TURN SENILE. - BIAS AMPLIFICATION: Model's inbuilt biases can combine with company's cultural quirks resulting in misinterpretations and distortions, worse they get reinforced over time. System will learn how the company thinks and then will optimize for continuity. AMNESIA IS SOMETIME A FEATURE. - SEMANTIC TRUTHS: Unlike objective truths this system will play on subjectivity and will build "causal" relationships that are based on semantics. This over time can form mythical decision norms with no external validity. The real question isn't "CAN IT BE BUILT?". Of course it can be. But "HOW IT SHOULD BE POSITIONSED". #llms #leadership #ai

1

Meta reported Q4 2025 earnings today. It was another strong quarter for our business and community, with total revenue of $59.9 billion, representing 24% year-over-year growth. Key Highlights: - 3.58 billion people use at least one of our apps every day, an increase of 7% year-over-year. - Facebook and WhatsApp both have more than 2 billion daily active users, and we are just shy of that on Instagram. - 🧵 Threads is on a tear, again benefiting from recommendation improvements, which led to a 20% lift in time spent on Threads. We also expanded ads on Threads globally this month. - 🤖 AI continues to enhance our services. In Q4, AI-driven optimizations led to a 7% lift in views of organic Feed and video posts on Facebook. - 🕶️ Our new Meta Ray-Ban Display glasses are a hit, with sales more than tripling last year, making them one of the fastest-growing consumer electronics products in history. We also shared further insights into how AI supported the ads system and drove significant improvements for advertisers: - In Q4, we doubled the number of GPUs used to train our GEM model for ads ranking. - Improvements to GEM and sequence learning drove a 3.5% lift in ad clicks on Facebook and more than a 1% gain in conversions on Instagram in Q4. - In Q4, we launched a new runtime model across Instagram Feed, Stories, and Reels, resulting in a 3% increase in conversion rates. - In Q4, we consolidated models for Facebook Stories and other surfaces into the overall Facebook model. This, along with a series of back-end improvements, drove a 12% increase in ads quality. with Trang Quynh Nhi Mai

90