Articles & Other Blogs/ Articles / News

Sam Walton (Walmart): America’s Richest Man - Founders Podcast

  1. Sam grew up during the Great Depression.
  2. What Sam learned when he worked at JC Penney – he learned here that efficiency, cost-cutting, and customer satisfaction are what make or break a strong business. Opening a large number of stores (1600) in smaller towns rather than big cities (was the optimal business model for his own endeavor).
  3. Sam’s father-in-law (affluent / successful businessman) lent him 25K to purchase a franchise of a Franklin Five & Dime. Sam went on to make it the most successful five and dime in the country.
  4. Sam was a visionary:
    • From what he learned at JC Penney, Sam knew to focus on customer satisfaction – he placed an ice cream machine and a popcorn machine right in front of his store since he knew it might lure some customers inside.
    • He realized that driving between multiple stores to oversee operations was too time-consuming. He started chartering flights to cut down on travel time.
    • Guided by his team, he was an early adopter of the computer (massive investment at the time). This eventually enabled them to structure their supply chains to facilitate same-day restocking.
    • Walmart discovered an interesting correlation between store size and sales. They found that building larger stores even in towns with just 2K people was a way to increase sales.
  5. Sam was relentless:
    • At the height of his store’s popularity and revenue (he was 32), he was forced to shut it down because of a bad rental agreement, but the same day he went out and found a new location.
    • He wanted to expand his new store, so he tried to strike a deal with the neighboring barbershop. He ended up making them offers six times until he finally convinced them.
    • He approached David Glass to be CEO at Walmart several times until he finally agreed.
    • He pitched the idea of Walmart (a discount variety store) to top executives at JC Penney, but they scoffed him out of the room. He made it one of the largest business stores in the world.
  6. Sam re-used ideas that worked:
    • Walmart was built on the business strategy of JC Penney - customer satisfaction and small towns vs. cities.
    • He copied Erwin Chase’s (Ann & Hope) idea of discounting (one of the first to pioneer the concept).
  7. Other ideas:
    • No one is eager to fix a cash machine that isn’t broken – he shouldn’t have pitched Walmart to JC Penney. This is similar to how James Dyson tried to sell the bagless vacuum cleaner to a company making tons of money selling vacuum bags. We also see this today in vertical AI agents—you can’t sell to the people you’re trying to replace. You need to sell to higher and higher executives.
    • Speed is everything – Sam believed that a good idea executed now is » an excellent idea executed a week from now. He would find a new location for a store, and before the store even had ACs, he would put standing fans throughout the store and start sales immediately.
    • Similarity to Bezos - Sam was extremely tough on his top executives, but he was the exact opposite with his front-line workers.

How to sell like Steve Jobs - Founders Podcast

  1. Build backwards from customer experience

    • The best products aren’t built from features; they’re built from understanding exactly how customers think, feel, and behave.
    • Reverse engineer from what the user wants—not from what’s easy to build.

  2. Sales is the only thing that matters for business

    • A company without sales is just an expensive hobby.
    • No amount of brand, design, or tech matters if no one is buying.

  3. When talking to investors - explain what you’ll do with the money simply

    • Most rich people don’t understand technical jargon—explain like you’re talking to a smart 12-year-old.
    • Makes it easier for investors to repeat your pitch to their peers.

  4. How to sell:

    • Draw a verbal roadmap simply, using unusual words

      • A great pitch is a story, not a data dump.

      • Unique phrasing makes ideas stick.

      • The goal is instant clarity, not technical depth.

      • Example : In the market today things are sleezy, slow and tiring to use.

    • Repetition of the central idea is everything

      • Everyone needs to walk away with the same message.
      • List the 3 things you want people to remember.
      • Back them with easy numbers, metaphors, or dead-simple logic.
      • Example: A thousand songs in your pocket - Ipod

    • It depends on who you’re selling to

      • Selling to developers? They care about control, customization, and integrations.
      • Selling to consumers? They care about ease, price, and emotional pull.
      • Same product, completely different pitch.

AI wont replace jobs - Marc Andreseen

Distribution of expenses in the last 20 years.

Distribution of expenses in the last 20 years (inflation adjusted)

We’ve always had the discussion of technological advancements replacing traditional jobs. Even if people are right, technology as a whole is restricted in most economical sectors. In the heavily regulated sectors (red above), both the government and private players have posed large barriers for technological entry which is why expenses have only been rising in these industries. Today, it costs $100 to get a flat-screen TV, but $1M to get extremely high quality college education - no one yet knows how to fix this. As before, the sectors in which powerful modern technology can easily creep in will continue to see quality improvements with declining prices. One the one hand, as a consumer we are frustrated with rising prices, yet on the other as a producer were frustrated by the constant need to adopt new technology and drop prices.

Anthropic CAN and DOES read your LLM requests

  • They aggregate anonymized requests, but still have access to those requests. The main concern is not with the company itself but with data-security in general. Most software companies in the US are above average when it comes to data security.
  • Interesting usage distribution:
What is Claude used for?

What do people use Claude for?

Gemini Flash 2.0 - Game changer for document parsing

  • Todays end-end models: struggle with complex format (detect tables, parse tables, create markdown).
  • Todays open-source and proprietary solutions: use multiple models for layout detection, extraction, parsing etc. For datasets in the order of millions of documents this is unsustainable.
  • Todays problems with semantic chunking: Semantic chunking is using an LLM to determine optimal chunk points in a document rather than defining heuristics. This therefore takes multiple LLM calls meaning it is very expensive.
  • Gemini Flash 2.0: ~6000 pages per dollar (pdf –> markdown), nearly perfect table conversion (some formatting problems) and its very cheap (meaning you can do semantic chunking for very low cost).
  • What do we lose? bounding-boxes for referencing exact sections within documents from which text was used. Think document scans rather than text-document.
  • Interestingly, very simple prompt:

    OCR the following page into Markdown. Tables should be formatted as HTML. Do not surround your output with triple backticks.Chunk the document into sections of roughly 250 - 1000 words. Our goal is to identify parts of the page with same semantic theme. These chunks will be embedded and used in a RAG pipeline.Surround the chunks with html tags.`

WhatsApp Pay gets approval to role out to 500M users in India

  • NPCI made two important decisions in 2024 :
    • Postponed 30% transaction limitation on UPI facilitators to democratize the market (Till December 2026, when the decision will be made)
    • Removed restrictions on Whatsapp Pay from 100M users to 500M users. This is strategic to help them achieve the first goal.
  • 65% of UPI transactions come from around 25% of all UPI users, which means these folks have already selected the app they’re going to go with.
  • Whatsapp had 100M registered users for Whatsapp Pay, but only 10M active users. The two largest providers have 190M (PhonePe) and 140M (Gpay) users respectively.
  • PayTM has 100M MAU, but only 40% use it as their primary UPI app. Cred, Navi etc have entered the market but despite significant sales and cashbacks on UPI, they captured collectively 3% of the market so Whatsapp will face similar disruption challenges.