Peeyush Rai

I've recently been reacquainted with Balaji's "idea maze"; which turns (decisions) in the product journey maze lead closer to success versus death. A founder can demonstrate progress in the maze by answering difficult questions well. This is hard early on, where you're still proving/disproving hypotheses. VCs may not want to fund founders too early in this journey because they'll be paying for the trip through the maze. Part of fundraising may be convincing them you're further along in the maze by just knowing your shit.

2

AI should be vastly simplifying POC delivery for all your vendors. If they are asking for more than a couple weeks to deliver a POC, that is a red flag. The speed of delivery here really is indicative of the speed of delivery in production. If your requirements are not challenging (in terms of speed, functional depth, and configurability), you're unlikely to learn anything new about the vendors you're evaluating. It's on you to show up with well documented use cases; it's on them to impress you with the solution. We're delivering full-featured POCs for complex revenue motions in 5 days or less. That timeline is only getting shorter, and we're seeing a direct relationship with actual project delivery. https://lnkd.in/e6AvmUE3

16

Why Draper Startup House Ventures invested in Taubyte Ai workloads are concentrating compute, cloud, and control plane power into fewer hands. Most cloud platforms introduce structural dependency at the infrastructure layer. Taubyte is rebuilding the cloud stack differently with Git-native deployment, content-addressed releases, and distributed execution across infrastructure you control. At the network layer, services communicate peer-to-peer via libp2p, eliminating central coordinators and creating a mesh that auto-discovers, self-heals, and scales horizontally without bottlenecks. Free from hyperscaler control planes and single points of failure. Infrastructure may be the real Ai arms race. Learn more 👉 taubyte.com Giulianna Crivello Nick Martin

13

1 Comment

The power of distribution wins again 👑 Microsoft Copilot dominates in big enterprises. Smaller teams adopt emerging tools like Cursor, Claude AI & Zed. Cursor and Claude AI usage drops sharply as company size grows. Surprisingly, Google Gemini shows low, flat usage across all company sizes 🤔 The real innovation in AI dev tooling is happening outside big enterprises. If you're building the next-gen stack, your early adopters are in startups (and maybe medium) where there's less friction and more room to experiment.

44

13 Comments

Some of the most promising AI companies on my radar are targeting what my partners Jaya Gupta and Ashu Garg call “glue functions.” Historically, roles like RevOps, DevOps, SecOps, etc grew in popularity because someone needed to carry context across functions. Someone had to pull data from one system, check something in another, apply some judgment, and move the process forward. Traditional software captured state inside individual systems, but it rarely encoded the cross-functional logic that actually drives decisions. So companies created glue roles to bridge the gaps. In contrast, agents are perfectly suited for these roles - making glue functions a natural wedge for the next generation of AI companies. Check out their full p.o.v. here: https://lnkd.in/gV_u3ngU

67

6 Comments

Software vendors are cutting prices before most enterprises have deployed AI agents at scale. This defies the normal sequence where widespread adoption precedes pricing pressure. But software markets are repricing now because buyers can credibly threaten replacement, even if they haven't executed it yet. Enterprise software purchasing teams understand that routine tasks—data entry, report generation, basic analysis—will soon cost pennies per hour instead of thousands per seat. This shifts negotiating leverage immediately, driven by the credible threat of substitution rather than actual substitution. The repricing wave hits different software categories with varying intensity. Some absorb massive damage while others prove surprisingly resilient. The pattern isn't random—it follows specific economic logic about which functions agents can replicate and which they cannot. But knowing that repricing runs ahead of deployment doesn't tell you which software companies face existential pressure versus temporary margin compression. It doesn't reveal which categories offer defensive moats or how quickly pricing recovers in resilient segments. Link in comments 👇

1

1 Comment

I'm so excited about the Data Center Innovation Initiative that Elemental Impact is launching today. Four of THE most important tech companies on the cutting edge of modern AI are teaming up to pilot technologies in order to prove that the AI boom doesn't need to increase emissions or rates, and can instead be a catalyst to pull forward a future with less carbon and lower rates. Selfishly, I'm also looking forward to the dealflow 😉 Apply here! https://lnkd.in/gzkaqCfM

29

2 Comments

🎤 GCC + Startups: Collaboration or Competition? Amlan S.'s Perspective His perspective was sharp, layered, and rooted in how GCCs and startups are fundamentally built differently—and why that makes both collaboration and competition inevitable. GCCs vs Startups — Structurally Different, Not Substitutes Amlan started with a powerful reframing: 👉 “GCCs and startups are built for completely different purposes.” 🔹 GCCs: Enterprise-owned, long-term capability engines Focus on data protection, IP, and brand integrity Own mission-critical platforms Designed to institutionalize innovation at scale 🔹 Startups: Venture-backed, opportunity-driven Move fast with minimal constraints Built for rapid experimentation & disruption 👉 His key takeaway: They are not substitutes—they are complementary systems Where They Compete vs Where They Collaborate This is where Amlan went very granular—almost like a playbook. 🔥 Where They COMPETE Primarily in the experience layer: Customer-facing apps Direct-to-consumer platforms Hyper-personalization engines Revenue-driving digital experiences 👉 Why? Because this is where: Speed matters Innovation cycles are short Differentiation is visible to customers ⚙️ Where They COLLABORATE (or GCC dominates) In the core enterprise layer: Supply Chain (SCM) Pricing, promotions Loyalty systems Core digital spine 👉 His sharp insight: “Enterprises don’t want to dilute their core.” This leads to a clear boundary model: Startups → innovate on the edge GCCs → protect and scale the core 🔴 The Future — From Competition to Collaboration Amlan made a very important shift clear: 👉 Earlier → GCCs vs Startups (competition mindset) 👉 Now → GCCs + Startups (collaboration model) “They are moving away from competition towards a collaborative platform.” 🔹 Why Collaboration is Increasing For Startups: Access to enterprise clients & datasets Validation at scale Funding opportunities Global scalability For GCCs: Access to cutting-edge tech environments Faster innovation cycles High R&D efficiency Minimal constraints experimentation 👉 His striking observation: “Startup tech environments are often ahead—unparalleled in agility.” ⚡ The Most Powerful Insight: Pilot vs Scale Model This one line from Amlan almost summarizes the future: 👉 “Startups excel at pilots. GCCs excel at industrialization.” 🧑‍💻 Talent & Innovation: Competition with a Twist Amlan also addressed a common debate: 👉 Do they compete for the same talent and budgets? ✔ Yes, they do compete ❗ But they attract talent for very different reasons Startups → Risk-takers, builders, speed-driven GCCs → Deep problem-solvers, stability-focused 👉 His conclusion: Competition exists—but intent is different, so conflict is limited 🔥 One Line That Captures His Entire Perspective: 👉 “Startups create possibilities. GCCs make them real at scale.” Small Clips are available in comments section Full Video - https://lnkd.in/eg73z5qE Somnath Chatterjee Dola Pal

17

5 Comments

So what is Scala? Hear Ardie Sameti co-founder and CEO explain how Scala is changing the game by helping to unify fragmented data and surface up what actually matters. If your head is dizzy from complex data, and you need to clarify those signals, you should learn more! "The gold is the data" Hear Ardie walk us through his incredible time at Accolade, the impact of core values and mentors, data and operational intelligence, and more! Theres a lot to learn this week on the Digital Health Heavyweights Podcast, check it out here: https://lnkd.in/gHcwrfyD

18

1 Comment

The team at Emergent is developing AI-assisted software creation for the masses. The company continues to scale thanks to a new round of funding led by Khosla Ventures and SoftBank. 1. Very cool to see an AI-native startup come out of India and expand to 190 countries in less than a year. 2. The customer segment features a broad base of non-technical users that range from jewelry store owners to senior citizens, all 'vibe-coding' for the first time. VSC is happy to help shape the narrative. Mukund Jha, Madhav Jha, Vinod Khosla, Ryan Scott,  Sarthak Misra, Manav Garg, Hemant Mohapatra, Jon Silber, Nicolas Dessaigne, Prashant Sharma

92

9 Comments

AI has changed the economics of software. Models are already intelligent enough, and coding agents make building cheap and fast. Markets too fragmented for traditional SaaS become addressable. The bottleneck is adoption and deployment. Abhijay Rana writes about why this makes forward-deployed engineering a good strategy at mid-market scale.

10

2 Comments

Fal operates in a highly competitive space but carved out a unique niche by focusing on generative media. Effectively creating a category of its own   From model creators to inference providers, cloud platforms to packaged solutions, many players in the space are both competitors and potential partners to fal   Their strategy of building an abstraction layer became a powerful wedge, helping them scale. As the market continues to swing with each new wave of innovation, Fal has been smartly placing bets across the value chain   The real key is maintaining a sharp tech edge. Staying ahead of major shifts, moved by big tech, and the next generation of models. Fal has consistently won by leapfrogging the competition 👇Hear it from Gorkem at e2vc Summit 💙

47

2 Comments

1 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 𝐡𝐞𝐚𝐫𝐭𝐛𝐞𝐚𝐭𝐬. 95% 𝐟𝐚𝐬𝐭𝐞𝐫 𝐝𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬. 🎧 Full episode available on Build With Shreya: https://lnkd.in/gcQiM83w In BuildWithShreya's latest conversation with Dr. Philip Nölling (Co-CEO, dpv-analytics GmbH), he shared how their AI-powered ECG system is transforming cardiac care. Instead of doctors manually scanning endless heartbeat data, AI now does the heavy lifting and flags the areas where human expertise is truly needed. Key takeaways: ✔️ Why traditional ECG analysis is inefficient ✔️ How AI “heat maps” simplify diagnosis for doctors ✔️ Why certification & oversight matter in medical AI ✔️ How patients benefit from faster, safer cardiac monitoring This is what happens when AI and doctors work together: smarter, faster, and more secure healthcare. #BuildWithShreya #dpvanalytics #AIHealthcare #DigitalHealth #CardiacMonitoring #HeartHealth #HealthcareInnovation #MedicalAI #FutureOfMedicine #HealthTechGermany

15

The next frontier of AI isn't just about choosing the right model. It's about building moats through implementation. Across the Emergence Capital portfolio, companies like Bland and Federato are scaling faster by embedding Forward-Deployed Engineers. These engineers bridge the gap between AI and reality, ensuring real results in production. Read more from my colleagues Gordon and Kyle on why FDEs are the key to unlocking AI's full value. https://lnkd.in/gcsRRGX2

29

3 Comments

Sustained ~252–291 GiB/s of HBM memory throughput on H100s under decode load — essentially hitting the hardware roofline. This matters because HBM throughput, not FLOPs, is the real bottleneck in LLM inference. By keeping memory nearly fully saturated, I’ve unlocked far higher efficiency and throughput than standard engines. The result: 0.9–1.36M tokens/sec with ~0.1 ms first-token latency  #AI #LLM #GPURouter #H100 #Inference #CostEfficiency #Innovation #DeepLearning #AIInfrastructure #HighPerformanceComputing NVIDIA AMD OpenAI Google Shilpa Kolhatkar Keith Strier a16z speedrun

5

Why Guardrails Are Critical for Safe AI Adoption Generative AI has moved from a research novelty to an everyday reality. Tools like ChatGPT, LLaMA and Gemini are redefining how we write, code and create. But with great power comes great responsibility and that’s where guardrails step in. LLMs are brilliant, but they’re not flawless. They can hallucinate facts, amplify bias, leak sensitive data or even generate toxic content. Left unchecked, these risks can lead to misinformation, legal troubles or loss of user trust. So, what are guardrails? Think of them as AI safety layers programmable rules and systems that monitor and control what LLMs produce. They act during user interaction, filtering unsafe inputs and outputs in real time. A good guardrail doesn’t just block bad content; it ensures outputs are accurate, ethical and compliant. Here’s how the ecosystem looks today: ✅ Llama Guard (Meta) – Fine-tuned on the LLaMA architecture to classify content into predefined safety categories. Flexible for different use cases. ✅ NVIDIA NeMo Guardrails – Uses structured conversational flows with KNN-based intent matching and moderation to keep conversations on track. ✅ Guardrails AI – Adds type and format constraints (e.g., enforcing JSON structures), plus corrective prompting if outputs fail checks. ✅ TruLens – Focused on evaluation and feedback loops to improve context relevance, groundedness, and fairness in RAG-based apps. ✅ Guidance AI & LMQL – Programming approaches that combine logic and generation, letting developers enforce constraints using regex, control flows and real-time checks. Why does this matter? Because attackers are getting creative. Jailbreak attacks, prompt injections, and adversarial queries can still bypass basic safeguards. Some even exploit language patterns or use multilingual prompts to trick models into producing harmful content. Building strong guardrails isn’t just about adding filters it’s about system design. We need: ✔ Multi-disciplinary strategies combining neural and symbolic reasoning ✔ Continuous monitoring and evaluation ✔ Integration across the AI development lifecycle, like safety-critical systems in aviation or automotive (think ISO 26262) The ultimate goal is trust. Businesses, governments, and users will only adopt AI at scale if they believe it’s reliable, fair and safe. Investing in guardrails is not an afterthought it’s the backbone of responsible AI. What's Next? Guardrails need to evolve beyond simple filters. The future lies ✔ Neural + Symbolic Systems working together for deep reasoning ✔ Lifecycle integration-from design to deployment (think ISO 26262 for Al) ✔ Continuous monitoring to detect new attack patterns and biases #AI #LLM #ResponsibleAI #Guardrails #GenerativeAI #EthicsInAI #AITrust

1

Excited to share that Pipeshub has been featured in GeekWire’s Startup Radar. As enterprises move beyond chatbots, the next wave of AI is about execution. The real opportunity is building AI agents that understand company context, work within permission boundaries, ground every response in source data, and actually get work done across enterprise systems. Thank you Taylor Soper for the mention. https://lnkd.in/gVrfKZSg

30

2 Comments

❤️ Loved this from Harshit Omar Cloud outages aren’t anomalies; they’re warnings. Most enterprises are architected as if hyperscalers are “too big to fail,” but recent outages have proven the opposite. Resilience needs a new baseline: Can you run somewhere else when your provider falters? That’s the promise of #CloudCloning and true #multicloud optionality. Big thanks to MR Rangaswami Sir for the conversation. Worth the read.

30

1 Comment