Tarush Aggarwal

What I learned from Jerry Yang & Larry Page TL;DR: Technology companies launching disruptive products should religiously focus on end-users above everything else, including partners. Solve end-user problems and everything else follows– adoption, revenue, and partners. Yahoo! & Jerry Yang (2006) In 2006, phone calls were expensive and based on time & distance pricing. Skype changed all that with VoIP, which bypassed telco networks and made super cheap calls using the Internet. They already amassed 34 million users and were growing at insane rates. At the same time, Yahoo Messenger had over +110 million users worldwide. We could leverage this userbase and create a VOIP product to challenge Skype. Yahoo had all the advantages on paper: the brand, the users, and the technology. So, with the leadership of Brad Garlinghouse, we built a competing product with a very disruptive pricing plan – 1 cent/min; Skype's pricing ranged from 2-3 cents/min. However, a week before the launch, Jerry Yang paused the project at a partner’s request. ATT-SBC was concerned our pricing would cannibalize its own voice business. Yang did not want to upset that relationship. Ultimately, Yahoo missed its opportunity, and Skype was off to dominate the VoIP market. Google & Larry Page (2008) Before the iPhone, companies had to negotiate with mobile carriers to get their apps on phones. Google was struggling to secure deals with Verizon for Google Maps and a new phone concept, Android. Around the same time, the FCC planned to auction 3G spectrum. Google pushed for open application and open device requirements on the carriers who won the auction. This infuriated the carriers. They argued that such conditions would decrease the spectrum's value below the $4.8 billion auction floor set by the US government. The FCC was leaning towards the carriers. The agency did not want to risk the auction not meeting the government's financial goals. Google's only option was to participate in the auction to ensure the minimum bid. Inside Google, the decision to bid was intensely debated. Android founder Andy Rubin argued the carriers would hate us and never partner with Android; Chris Sacca argued that bidding was the only way to guarantee Google's future in mobile. Page sided with Sacca. Page’s rationale: “. . . the carriers already hate us. If we don’t bid, they would hate us and not respect us.”  We bid, the openness provisions were triggered, and we secured a path for Google apps on 3G. Sixteen months later, Andy got his deal Verizon –  Verizon Droid! Page was willing to sacrifice mobile carrier deals to ensure Google’s products were universally accessible to its users. He played spectrum poker with $4.8 billion to achieve that result. (cont. below)

29

11 Comments

What do 4 Ascend.io data 🤓s talk about on a Hackathon hike? 🤖 multi agent orchestration with #LangGraph 🦆 #DuckDB benchmarks and CSV parsing 👷streaming with #Confluent It's an exciting time in data & AI!

60

1 Comment

Had the privilege to join a Fireside Chat with Sudivya from the Google Cloud team at IIMA Ventures. Key takeaways: LLM democratization: Making custom LLMs accessible to non-tech users. AI in action: Impactful use cases like RAG and domain-specific automation. Build vs. leverage: Pre-trained models offer speed and flexibility; fine-tuning ensures precision. Tech challenges: Managing data at scale, ensuring compliance, and balancing the economics of GPU-intensive workloads remain complex challenges. Balancing data efficiency with cost-effective GPU scaling via distilled LLMs. Why GCP?: Powerful AI tools, stellar startup support, and global scalability. #DataNeuron #AI #FiresideChat #LLM #GoogleCloud #Innovation #Startups

41

Old way of enriching data: (using 1 provider) - Missing out on 50-80% of your TAM - Reliance on a single 'source of truth' - Restricted to provider's capabilities - No external data verification layers New way of enriching data: (aggregating 20+ providers) - Maximized coverage and data quality - Data reliability with multiple verification layers - Bigger pipeline due to data competitors can't find - Saving costs through one consolidated subscription Do new data enrichment in 2024. Are you still doing it the old way? 👇

30

6 Comments

1/ Uncomfortable $10B question for $DELL 😳 Analyst Toni Sacconaghi asked a simple yet brutal Q: Why didn't operating profit change when $DELL added $1.7B worth of servers for #AI services? #earnings #tech 2/ The implication? $DELL's operating margin on #AI servers is effectively zero! 😲 A single Q led to a $10B drop in market cap. Ouch! #investors #stockmarket 3/ Are chipmakers the real #winners of the #AI hype? 🤔 While tech giants chase AI, semiconductor companies rake in profits from high demand. #analytics #bigdata 4/ $DELL's situation reminds us that not everyone benefits equally from new tech trends. 📉 The road to #AI domination is paved with costly investments. #riskvreward #innovation 5/ A $10B question indeed! 💸 One analyst's query exposed potential cracks in $DELL's #AI strategy. Will they course-correct or get left behind? 🔭 #disruptor #futuretech 6/ In the world of finance, simple Qs can have massive impacts. 💥 $DELL's case shows why public companies must be transparent & prepared. #accountability #shareholdervalue 7/ The AI wave is rising, but who'll be the real surf champions? 🌊 Tech titans or component suppliers? An exciting race to watch! #competition #disruption 8/ Perhaps $DELL's AI conundrum serves as a wake-up call. 💡 Pivoting to new tech requires careful planning & execution. No room for rookie mistakes! #strategy #leadership 9/ While some rejoice at $DELL's $10B stumble, others see an opportunity. 👀 A shake-up could spur innovation & realign priorities. #transformation #pivot 10/ In the end, the $10B Q highlights the complexities of the #AI revolution. 🧩 Not every player will emerge a winner, but the game's just begun! #futureisnow #techbuzz

1

Temu Founder Actually Shopping Like a Billionaire: Has Amazon Worried Interesting story from Bloomberg last week reported that the former Googler and Pinduoduo founder, Colin Huang. Started Pinduoduo in 2015 selling extremely cheap products with incredible promotions. Grew quickly and became his net worth became $71.5B in 2021 at COVID's peak. From there it came crumbling down 87% in a year. But the company created the Temu brand to expand out of China, and expand it has. Now Mr. Huang is China's richest person according to the Bloomberg Billionaire's Index. Jack Ma is 5th - who everyone knows in the West. But hardly anyone knows PDD or Colin Huang. The goal of Temu? Essentially good deals. In a 2018 interview, the founder stated: "The goal is not to be cheap, but to make users feel like they got a good deal." The downside of Temu continues to be how they treat employees and suppliers (Bloomberg reports that employees must work 11 a.m. to 11 p.m., six days a week, plus overtime). Ouch. On the supplier front, global marketplaces have never been given a gold star for treating suppliers well. Also given that Temu is in the EU, it is under regulation by an entity which isn't exactly seen as friendly to big tech companies. I note that many in the industry seem to expect Temu will just go away, it will spend itself out of existence like Wish. I do not share the same view. The difference? Wish raised $1.6B. PDD has $33.3B cash on hand. Obviously Wework proved you can spend any amount of money if you try REAL HARD, but to think a deal-oriented site needs that much capital even with advertising costs is a stretch in my mind. Temu's market is white-label goods. Items you don't care what the name is. Incidentally, 77% of product searches on Amazon are unbranded. Not to say that much of Amazon is white-label goods, but especially in an increasingly unaffordable America (child care, healthcare, college, real estate) the value trend is not going away. Consumers are cheap, and getting cheaper. So if you think that a somewhat trusted Temu isn't at least some threat to Amazon you are kidding yourself. In a world where missing 5% of your growth is a big deal, that's all it would take for Amazon to lose share. Combined with loss of first eCommerce searches from social shopping platforms, all is not completely rosy in the Amazon universe.

113

24 Comments

An interesting thing has started happening in #AI -- the use cases have started expanding from "personal productivity" and "cost cutting" to directly DRIVING REVENUE. For example:  March 2023: "Hey guys, check out how I used ChatGPT to write a Linkedin post" to Nov 2023 : "Hey guys, we deployed a CustomGPT.ai bot on our website and dramatically cut down on customer support costs" to July 2024 : "Hey guys, we directly integrated AI into our product offering and have started driving incremental revenue" We are seeing lots and lots of such use cases emerging: 1. Integrating AI into your products and services and charging extra for them (just look at Microsoft!) 2. Using AI to drive leads and sales --> direct impact on business. 3. Using AI to directly improve conversion funnels (for SaaS businesses) All these examples are speaking directly to the CFO -- rather than to cost cutting. Hint: If you are thinking on these lines, speak to Eli Simon

11

2 Comments

Increasing revenue is a much easier internal sell than "We need to replace Bob with AI". We are now seeing a lot of action around the "increasing revenue" value proposition -- so things like: 1. Incorporating AI straight into the product (and charging extra for it) 2. Using AI to drive the revenue driving processes like sales and lead gen. 3. Using AI to expedite product roadmaps, research and processes directly tied to driving revenue. For service providers like accountants, tax lawyers, consultants, agencies, etc building a custom GPT and using that as a differentiator is driving new clients and incremental revenue. For example: Instead of being "just another tax lawyer", you can be a tax lawyer who comes with a special custom GPT in hand -- this gives you an edge over other tax lawyers.

6

Catch my latest interview with AbdulMajedRaja (1littlecoder) where I discuss my thought process about identifying and building AI SaaS apps as bootstrapped and resource-constrained builders! Most of last decade's startups were built on the premise of the problem as the core. With the advancements in AI, one can focus on the traditional problems that are already being paid for and actively work on a 10x improvement delivered with AI! I discussed how most GPT Wrappers (including mine) have a monetizable window and how it extend it by grounding it in a business vertical! I also shared my thoughts about how to keep up with the trends in AI by context switching on topics every 2 months, whether it be Multimodal LLMs, RAG, AI Agents, etc. #llms #artificialintelligence #saas #startup #ai https://lnkd.in/gnTzWTr6

139

7 Comments

As we near the US presidential election, prediction markets have been surging. So, what are prediction markets and where does this budding industry go post election? As prediction markets gain mainstream traction, platforms like Kalshi, PredictIt, Polymarket, Betfair, and Smarkets are transforming how we leverage event-based insights for betting purposes. While regulatory frameworks evolve, these platforms are setting the stage for data-driven market intelligence. The industry's growth is primarily supported by: ◽ Increased Interest in Real-Time Data: Prediction markets serve as valuable tools for organizations and individuals looking for near real-time information and market sentiment on key events. This shift is encouraging broader use cases, from economic forecasting to risk management for bettors. ◽ Expansion of Regulatory Frameworks: Platforms are navigating complex regulatory environments with increasing success, providing legitimacy and attracting new users. Regulatory acceptance is anticipated to pave the way for more mainstream platforms, leading to greater user adoption. ◽ Advances in Blockchain Technology: Some platforms leverage blockchain for decentralized prediction markets, allowing for transparent and secure transactions, which appeal to users interested in global, crypto-based markets. There is even a new startup aggregating content across the prediction market industry, Prediction News. I have also seen that some prediction market platforms are going to start offering sports betting... What role does prediction markets play in the overall betting ecosystem? Will it grow faster than sports betting due to it's broad range of events? https://lnkd.in/g3R9Sgpu #PredictionMarkets #Fintech #FutureOfFinance #Kalshi #EconomicInsights #Blockchain #DataDriven #EventTrading

17

2 Comments

Amazon is upgrading #Alexa by replacing its in-house AI with the #Claude model from #Anthropic. I've always been a fan of Claude #AI, particularly for its conversational and engaging responses, and its stronger emphasis on ethical features. It seems like a great fit for powering conversational smart assistants like Alexa. Originally, Alexa has always been powered by Amazon's own AI model, but testing has shown that it takes longer to respond to users than expected. The Claude model apparently performed significantly better in this regard. This collaboration between the two tech companies is not sudden. Last year, Amazon invested $1.25 billion in Anthropic, and earlier this year, they added an additional $2.75 billion investment. Amazon promises that the upgraded Alexa will offer more complex, context-aware interactions and multi-step task handling. The new version sounds exciting, but it comes with a subscription fee of around $5-10 per month. Alexa has always been free to use for a long time, and the enhanced features of this new version might justify the cost for frequent users. However it's always tricky to ask casual users to start paying for a service that was initially free. Would you be willing to pay for it? What do you usually use Alexa for? __________________ I share my learning journey here. Join me and let's grow together. For more on AI and learning materials, please check my previous posts. Alex Wang #artificialintelligence #technology #machinelearning #innovation #bussiness

6

You do realize that 2GB of ram means 2B bytes? For 3B params? And you need multiple bytes for each token? You can quantize all you want - LLMs will never be reliable in limited memory save for narrow purpose-build cases, that’s just physics. (Some would say they will never be reliable, period. )

3

3 Comments

Google Cloud at a ~$45B run rate growing 35% (not constant currency). Google cloud includes GCP and GSuite Quarterly YoY growth trends and quarterly net new ARR added below. Starting to see more reacceleration

58

2 Comments

Wondering if anyone is getting traction selling AI consulting and implementation services to SMBs or enterprise? Gregory Kamradt asked this question on twitter, got a lot of great responses from industry source and put together a must read doc summarizing them. Some highlights 1) Bigger companies are adopting AI faster than SMBs. Normal companies (outside the gen AI echo chamber) are still trying to understand the tech. More interest in training/education than deployment/development. Allie K. Miller has been doing a great job providing educational content and if you are not already following her, I highly recommend you do. 2) Most companies don’t have the data infra or basic practices in place to utilize AI well so a lot of it is using automation as a forcing mechanism to digitize or consolidate data so AI can be used with it. This mirrors my experience and is something StepChange.work can help with Link to doc in comments below.

19

4 Comments

Ever wondered how blockchains maintain such incredible efficiency while guaranteeing data integrity? The secret lies in a fascinating concept called Merkle trees. Imagine a vast library where every book's accuracy needs verification. Checking each book individually would be tedious and time-consuming. Merkle trees offer a magical solution. They allow us to verify the integrity of the entire library by simply checking a single hash, called the root hash. Prior to Merkle trees, various methods like full data verification and cryptographic hashing existed, but they lacked efficiency or granularity. Merkle trees revolutionized the game by combining the efficiency of hashing with the ability to verify individual data points within a larger dataset. This perfect balance between security and scalability makes them a cornerstone of blockchain technology. Merkle trees find diverse applications within blockchain. They play a crucial role in sidechains, layer-2 solutions, zero-knowledge proofs, and decentralized storage offered by dApps like Filecoin and Arweave. These applications leverage Merkle trees to ensure secure and censorship-resistant storage solutions.

2

https://lnkd.in/gwVZPStp - the largest model ever. - open source-ish. - multi-modal, vision (images and video) and text. - whispers of improved reasoning #ai news.

3

1 Comment

We’ve already seen that software eats the world. Ali Ghodsi predicts that, in the next phase, AI eats software. During a conversation on the B2BaCEO podcast, Ali, the Co-founder & CEO of Databricks, said AI will not just be a feature; it will be the foundation of most platforms, as we move toward a world where AI is seamlessly integrated into nearly every tool. Imagine every piece of software evolving from static code into an autonomous decision-maker, as intelligence and automation are built into all data architecture underlying enterprise software. Our full conversation on the B2BaCEO podcast here: https://lnkd.in/gXvSSf_C

56

4 Comments

Turns out, Sword Health has an interesting org structure where they hire ex-founders/wannabe founders as GMs, invest $1 million as a seed round, and give them 12 months to work. If the product performs well after a year, they follow up with a Series A investment. We do something similar at Docquity, and each time, the products have generated quite a bit of value for the organization in a fraction of the time it usually takes. In my humble view, this is the only way to create urgency and build value from 0 to 1 as a company matures. h/t Nikhil Krishnan

45

10 Comments

Use OpenAI batch API effortlessly, reliably and at scale. Realtime inferencing is not needed for numerous usecases. With this launch and integration you can use the best model on the planet at 50% off and yet get reliable executions. Also spend almost no money waiting for these results - let Union and Flyte handle this painlessly for your usecases. Try out today, coming soon to Union serverless! Example usecases: Evals, large scale generations, summarizations and lowering the cost of agentic style workloads.

12