About
Building FinalLayer - AI agents for professional presence. Our first agent is a LinkedIn…
Activity
13K followers
-
Vidya Narayanan shared this𝐍𝐚𝐢𝐥𝐞𝐝 𝐢𝐭: 𝐜𝐨𝐠𝐧𝐢𝐭𝐢𝐨𝐧 𝐢𝐬 𝐭𝐡𝐞 𝐛𝐨𝐭𝐭𝐥𝐞𝐧𝐞𝐜𝐤. I'm drowning. And I'm only running a few agents in parallel. Not the 100x that Jensen Huang is talking about. Our brains have capacity limits. That's reality. And we're trying to distort reality at the moment. There is a lot of performative content claiming superhuman powers. 𝐖𝐞'𝐫𝐞 𝐣𝐮𝐬𝐭 𝐞𝐱𝐩𝐞𝐧𝐬𝐢𝐯𝐞 𝐛𝐚𝐛𝐲𝐬𝐢𝐭𝐭𝐞𝐫𝐬 𝐟𝐨𝐫 𝐯𝐞𝐫𝐲 𝐜𝐚𝐩𝐚𝐛𝐥𝐞 𝐭𝐨𝐝𝐝𝐥𝐞𝐫𝐬. And I'm glad to see some people agree! This was originally discussed in a podcast between Lenny Rachitsky and Simon Willison. === 📌 Topic curated by FinalLayer; inputs are mine. 🗓️ Let's discuss your brand, book a time with me!
-
Vidya Narayanan shared thisDoes AI have a "rising tide" or a "crashing waves" impact? We're reacting as if it has a crashing waves impact, but research says otherwise. New MIT research analyzed 17,000+ worker evaluations across 3,000 labor tasks. The findings challenge the prevailing narrative about AI automation: ✨ AI capabilities improve gradually across ALL task durations simultaneously ✨ Not sudden breakthroughs in narrow areas ✨ Success rates went from 50% (3-4 hour tasks) to 65% in just over a year 𝐓𝐡𝐞 𝐝𝐚𝐭𝐚 𝐬𝐡𝐨𝐰𝐬 𝐀𝐈 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐢𝐬 𝐬𝐮𝐫𝐩𝐫𝐢𝐬𝐢𝐧𝐠𝐥𝐲 𝐟𝐥𝐚𝐭 𝐚𝐜𝐫𝐨𝐬𝐬 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭 𝐭𝐚𝐬𝐤 𝐜𝐨𝐦𝐩𝐥𝐞𝐱𝐢𝐭𝐢𝐞𝐬. A 10x increase in task duration only drops AI success by ~8 percentage points. This isn't a steep "difficulty cliff". What this means: 🔹 Workers get more visibility into automation trends (less blindsiding) 🔹 Broad capability improvements across multiple job functions 🔹 80-95% AI success rates projected by 2029 for text-based tasks 🔹 𝐁𝐮𝐭 𝐧𝐞𝐚𝐫-𝐩𝐞𝐫𝐟𝐞𝐜𝐭 𝐚𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐨𝐧 𝐬𝐭𝐢𝐥𝐥 𝐲𝐞𝐚𝐫𝐬 𝐚𝐰𝐚𝐲 The implication? 𝐀𝐈 𝐢𝐬 𝐜𝐫𝐞𝐚𝐭𝐢𝐧𝐠 𝐧𝐞𝐰 𝐫𝐨𝐥𝐞𝐬 𝐰𝐡𝐢𝐥𝐞 𝐠𝐫𝐚𝐝𝐮𝐚𝐥𝐥𝐲 𝐥𝐢𝐟𝐭𝐢𝐧𝐠 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐯𝐢𝐭𝐲 𝐚𝐜𝐫𝐨𝐬𝐬 𝐞𝐱𝐢𝐬𝐭𝐢𝐧𝐠 𝐨𝐧𝐞𝐬. It's less about sudden job displacement and more about evolving how we work. But the industry is reacting wrong and we're going to see corrections. The rising tide metaphor fits: all boats lift together, but at a pace that allows adaptation. === 📌 Topic suggested by FinalLayer 🗓️ Book a time with me to build your brand
-
Vidya Narayanan shared thisNo is the answer. Many incumbents have a 𝐃𝐍𝐀 𝐩𝐫𝐨𝐛𝐥𝐞𝐦. And they are too afraid to 𝐜𝐚𝐧𝐧𝐢𝐛𝐚𝐥𝐢𝐳𝐞 𝐭𝐡𝐞𝐦𝐬𝐞𝐥𝐯𝐞𝐬. 𝐶𝑎𝑠𝑒 𝑖𝑛 𝑝𝑜𝑖𝑛𝑡: Adobe shoves AI into existing products. Existing products have complex UX and are impossible to democratize. Startups will build disruptively simple user experiences. Speed isn't the bottleneck here. Most large companies can adopt AI tools quickly when they want to. They have resources, talent, budgets. 𝐈𝐭'𝐬 𝐚𝐥𝐰𝐚𝐲𝐬 𝐚 𝐃𝐍𝐀 𝐩𝐫𝐨𝐛𝐥𝐞𝐦. === 📌 Topic curated by FinalLayer; inputs are mine. 🗓️ Book a time with me, let's build your brand!
-
Vidya Narayanan shared thisDistribution is simultaneously the hardest thing to crack and the most rapid when you do crack it. When it comes to GenAI, the product itself is transformative ✨ It satisfies two critical vectors - curiosity and productivity - and that's rare. Andrew Chen wrote about the death of network effects - 𝐰𝐞 𝐝𝐨𝐧'𝐭 𝐭𝐫𝐮𝐥𝐲 𝐫𝐞𝐜𝐨𝐦𝐦𝐞𝐧𝐝 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐬 𝐭𝐨 𝐟𝐫𝐢𝐞𝐧𝐝𝐬 𝐚𝐧𝐲𝐦𝐨𝐫𝐞. The bar is higher than ever. The product gap for adoption between incumbents and new products used to be "9x better". I wonder if it's even higher now. What's the next AI application that you think will make everything else feel obsolete? ======= 📌 Topic curated by FinalLayer; inputs are mine. 🗓️ Let's build your brand together; book a time!
-
Vidya Narayanan shared thisOpenAI 𝐚𝐜𝐪𝐮𝐢𝐫𝐞𝐝 TBPN. It's about narrative control. Storytelling is the new hot job. And community is the new moat. Every AI company is racing to become a media company. Apparently, a 1-year promo deal with TBPN was $1M. They are the talk of Silicon Valley when it comes to tech media now. 𝐓𝐡𝐞 𝐭𝐞𝐜𝐡 𝐢𝐬𝐧'𝐭 𝐞𝐧𝐨𝐮𝐠𝐡 𝐚𝐧𝐲𝐦𝐨𝐫𝐞 - you need to shape the conversation around it. When you can't quite win on pure tech merit, you build the platform that explains why your approach matters. Sam has decided he's going to amplify his hype machine. Makes sense. === 📌 Topic curated by FinalLayer; it can find yours too. 🗓️ Let us help you with your storytelling. Book a time with me.
-
Vidya Narayanan shared thisPerplexity just can't stay out of trouble. Less than a month after Amazon's court order over their shopping agent violations, now they're facing a class action lawsuit for allegedly sharing private user chats with Meta and Google. Perplexity has to be the most unserious company in the AI era. I quit using Perplexity after my Free Pro subscription ended. I also tried out the Comet browser when it first came out and gave up within a week. I'm actually pretty liberal when it comes to trying out new tech but Perplexity is one company that I wouldn't trust with sensitive data. ==== 📌 Topic suggested by FinalLayer 🗓️ Book a time, let's build your brand!
-
Vidya Narayanan shared thisOmg, 1000% this!! It's about time AI rescued us from administrative hell. Email Meetings Hiring loops Outreach All the 𝐞𝐬𝐬𝐞𝐧𝐭𝐢𝐚𝐥 𝐜𝐨𝐠𝐧𝐢𝐭𝐢𝐯𝐞 𝐨𝐯𝐞𝐫𝐡𝐞𝐚𝐝 𝐭𝐡𝐚𝐭 𝐛𝐮𝐫𝐢𝐞𝐬 𝐚𝐜𝐭𝐮𝐚𝐥 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐯𝐢𝐭𝐲. If I weren't drowning in process, I would be so unstoppable 🚀 This is exactly why I'm building FinalLayer 𝐈𝐧𝐛𝐨𝐱. 𝐁𝐞𝐜𝐚𝐮𝐬𝐞 𝐢𝐭'𝐬 𝐭𝐢𝐦𝐞 𝐰𝐞 𝐡𝐚𝐝 𝐚 𝐫𝐞𝐚𝐥 𝐀𝐈 𝐂𝐡𝐢𝐞𝐟 𝐨𝐟 𝐒𝐭𝐚𝐟𝐟 𝐭𝐡𝐚𝐭 𝐰𝐨𝐫𝐤𝐬. === 📌 Tweet curated by FinalLayer; inputs are mine. 📃 If you need Inbox , comment here 👇🏼
-
Vidya Narayanan shared this𝐓𝐡𝐞 𝐛𝐞𝐬𝐭 𝐭𝐡𝐢𝐧𝐠 𝐭𝐡𝐚𝐭 𝐡𝐚𝐩𝐩𝐞𝐧𝐞𝐝 𝐭𝐨 𝐦𝐞 𝐢𝐬 𝐀𝐈 𝐜𝐨𝐝𝐢𝐧𝐠. Probably the worst thing that happened to our engineering team is me having access to AI coding 😂 (Lakshminath wdyt?!) But one thing is for sure - 𝐰𝐞'𝐫𝐞 𝐧𝐨𝐭 𝐜𝐨𝐦𝐢𝐧𝐠 𝐛𝐚𝐜𝐤 𝐟𝐫𝐨𝐦 𝐭𝐡𝐢𝐬. The old handoff model between PM and eng is dead. 𝐖𝐡𝐲 𝐬𝐢𝐭 𝐭𝐡𝐫𝐨𝐮𝐠𝐡 𝐞𝐧𝐝𝐥𝐞𝐬𝐬 𝐬𝐩𝐞𝐜 𝐦𝐞𝐞𝐭𝐢𝐧𝐠𝐬 𝐰𝐡𝐞𝐧 𝐲𝐨𝐮 𝐜𝐚𝐧 𝐬𝐡𝐢𝐩 𝐚 𝐰𝐨𝐫𝐤𝐢𝐧𝐠 𝐩𝐫𝐨𝐭𝐨𝐭𝐲𝐩𝐞 𝐢𝐧 𝐡𝐨𝐮𝐫𝐬? ✅ Skip the translation errors ✅ Skip the vision dilution 𝐈𝐟 𝐲𝐨𝐮 𝐡𝐚𝐯𝐞 𝐠𝐫𝐞𝐚𝐭 𝐜𝐥𝐚𝐫𝐢𝐭𝐲 𝐨𝐟 𝐭𝐡𝐨𝐮𝐠𝐡𝐭, 𝐢𝐧𝐭𝐮𝐢𝐭𝐢𝐨𝐧, 𝐚𝐭𝐭𝐞𝐧𝐭𝐢𝐨𝐧 𝐭𝐨 𝐝𝐞𝐭𝐚𝐢𝐥, 𝐚𝐧𝐝 𝐭𝐚𝐬𝐭𝐞, 𝐲𝐨𝐮'𝐫𝐞 𝐚𝐧 𝐀+ 𝐛𝐮𝐢𝐥𝐝𝐞𝐫 𝐭𝐨𝐝𝐚𝐲. And that requires precision thinking most CS programs never taught us. Companies are still figuring out the new organizational API. But 𝐭𝐡𝐞 𝐝𝐚𝐦 𝐡𝐚𝐬 𝐛𝐫𝐨𝐤𝐞𝐧. === 📌 Topic curated by FinalLayer; inputs are mine. 🗓️ Let us grow your LinkedIn; book a time with me!
-
Vidya Narayanan shared thisI know this is only one breaking news of the day (what a day with both OpenAI and Anthropic contributing to extra major news 😅) ... but am I the only one finding a 14% dilution at this stage very strange? Also the flywheel described is so generic that it applies to just about any LLM. Don't get me wrong, I downloaded Codex after Claude just started giving me the dumber version more often. And I'm not hating Codex. It's actually pretty good. But I am not yet seeing the dominant moat. I have officially switched to Deepseek for many things on the API side. It's literally 50x cheaper and almost as good for many tasks. We're living in interesting times, I guess we'll find out 🤷🏻♀️ ================ 📌 For a change, this post was NOT written with FinalLayer :). This is an impulsive repost from the parking lot of a Lucky's... Sometimes, thoughts can't wait 😂Vidya Narayanan shared thisBREAKING NEWS: OpenAI raises a whopping $122 billion. Valuation of $852 billion. Here are the parts that I found interesting: * Now generating $2B of revenue...per month. * Over $3B of the investment was from individual investors. Nice! * They now have over 900M weekly active users. * Revenue from their ads pilot hit $100M in 6 weeks. I love flywheels, and thought theirs is well articulated: "The OpenAI flywheel is simple. More compute drives more intelligent models. More intelligent models drive better products. Better products drive faster adoption, more revenue and more cashflow. That gives us the ability to reinvest and deliver intelligence more efficiently to consumers, enterprises, and builders around the world." Congrats again to the OpenAI team (that's twice in two days). You're on a roll.
-
Vidya Narayanan liked thisVidya Narayanan liked thisWe reduced data review time by 60% in our clinical pilot. The technology worked perfectly. Then we hit week 13. Our mid-size pharma client asked about ongoing support budget. The pilot had been funded through Q2. No one had secured funding beyond that. The regulatory team loved the results. The data scientists were proud of the model. IT was ready to deploy. But there was no money for maintenance, no plan for scaling across departments, and no executive sponsor willing to own the transition. 𝐓𝐡𝐞 𝐩𝐢𝐥𝐨𝐭 𝐝𝐢𝐞𝐝 𝐧𝐨𝐭 𝐛𝐞𝐜𝐚𝐮𝐬𝐞 𝐢𝐭 𝐟𝐚𝐢𝐥𝐞𝐝, 𝐛𝐮𝐭 𝐛𝐞𝐜𝐚𝐮𝐬𝐞 𝐢𝐭 𝐬𝐮𝐜𝐜𝐞𝐞𝐝𝐞𝐝 𝐢𝐧 𝐢𝐬𝐨𝐥𝐚𝐭𝐢𝐨𝐧. Now we map stakeholders before we write a single line of code. Budget allocation conversations happen in week 1, not week 13. We identify who needs workflow changes and bring them into design sessions early. Because 𝐭𝐡𝐞 𝐛𝐞𝐬𝐭 𝐭𝐞𝐜𝐡𝐧𝐨𝐥𝐨𝐠𝐲 𝐢𝐧 𝐭𝐡𝐞 𝐰𝐨𝐫𝐥𝐝 𝐦𝐞𝐚𝐧𝐬 𝐧𝐨𝐭𝐡𝐢𝐧𝐠 𝐢𝐟 𝐭𝐡𝐞 𝐨𝐫𝐠𝐚𝐧𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐢𝐬𝐧'𝐭 𝐫𝐞𝐚𝐝𝐲 𝐭𝐨 𝐚𝐝𝐨𝐩𝐭 𝐢𝐭. #DigitalTransformation #LifeSciences #Innovation #AIGovernance #ProductStrategy
-
Vidya Narayanan liked thisVidya Narayanan liked this"How to connect with millionaires?" I was mentoring a startup founder, he asked me this interesting question. Sharing the gist of our conversation - READ every bit and SAVE it. 1. Solve a problem. 2. Look for the pain point. Rich people also have problems. Real ones. Most of us imagine money solves everything. If they are rich, they are lucky. Happy, they made it. But therapists who work with millionaires say the opposite. Many of the wealthy struggle with isolation, broken relationships, and even questions of meaning and purpose. Let’s unpack this. And more importantly if you want to connect with wealthy people, learn to let’s talk about their problems and how you can help solve these problems. So what’s the bigger picture? Wealth can create: - Distrust - Isolation - Empty family life - Fear of losing out - Pressure from fake relationships The lesson? Money changes circumstances, not emotions. The ultra-rich don’t need more products. They need people who understand their struggles, and offer real solutions. If you want to connect with rich people, stop selling luxury. Start solving their problems. (or at least start talking about these problems) Share one or two problems Rich People don't talk much about.. (Use Claude to find your answer and add in comments.
-
Vidya Narayanan liked thisVidya Narayanan liked thisSales reps hate scripts. They hate hunting for scripts even more. ALT+Tabbing between Notion and your CRM during calls? Painful. I moved our sales scripts directly into the CRM. Now reps choose their script at call start. No tab switching. The master document stays with leadership for updates. But the daily workflow? Seamless. Over time, script discipline became our sales foundation. Scripts are boring but closes the sale. I piloted this at one location first. The other locations stopped arguing and started following. #SalesOps #RevOps #CRM
-
Vidya Narayanan liked thisVidya Narayanan liked thisNicolas and I constantly Airdrop .md skill files back and forth to keep our Claude Code skills in sync. So we built a lil something... Our old workflow: I update a skill. Nick's still running the old version. We get different outputs. 10 minutes of "wait, which version are you on?" followed by an AirDrop & a prayer. The alt was version control with Github but it was annoying to push and pull with every file change. So we built SkillTree: a desktop app that detects every skill on your machine, puts them in one view across tools (Claude Code, Cursor, Codex), and syncs changes to a shared workspace so you're always using the same version. It was a fun, quick project we built because we needed it. But if your team shares skills across roles, you've probably hit the same wall. So we're putting it out there in case it's useful! Link in comments & let us know if you want to learn more :)
Experience & Education
-
FinalLayer
View Vidya’s full experience
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Publications
-
RFC 5295 - EAP EMSK
IETF
The Extensible Authentication Protocol (EAP) defined the Extended
Master Session Key (EMSK) generation, but reserved it for unspecified
future uses. This memo reserves the EMSK for the sole purpose of
deriving root keys. Root keys are master keys that can be used for
multiple purposes, identified by usage definitions. This document
also specifies a mechanism for avoiding conflicts between root keys
by deriving them in a manner that guarantees…The Extensible Authentication Protocol (EAP) defined the Extended
Master Session Key (EMSK) generation, but reserved it for unspecified
future uses. This memo reserves the EMSK for the sole purpose of
deriving root keys. Root keys are master keys that can be used for
multiple purposes, identified by usage definitions. This document
also specifies a mechanism for avoiding conflicts between root keys
by deriving them in a manner that guarantees cryptographic
separation. Finally, this document also defines one such root key
usage: Domain-Specific Root Keys are root keys made available to and
used within specific key management domains.Other authors -
-
802.11 Authentication – Introduction and Concepts
CCCT04 Conference Proceedings
-
Activity and Device Position Recognition in Mobile Devices
ACM UbiComp 2011 Poster and Demo Paper
-
Enabling Semantic Understanding of Situations from Contextual Data in a Privacy-Sensitive Manner
AAAI 2011 Workshop Paper
-
Qualcomm Context-Awareness Symposium Sets Research Agenda for Context-Aware Smartphones
IEEE Pervasive Computing, Wearable Computing Section (to appear)
Patents
-
Methods and apparatus for event distribution and routing in peer-to-peer overlay networks
Issued US 8,996,726
Methods and apparatus for event distribution and routing in peer-to-peer overlay networks. A method is provided for event distribution and routing in a peer-to-peer overlay network that comprises a plurality of nodes. The method includes identifying a plurality of buckets on the overlay network, wherein each bucket includes one or more nodes, respectively, identifying bucket groups, wherein each bucket group includes a selected number of buckets, respectively, distributing events based on the…
Methods and apparatus for event distribution and routing in peer-to-peer overlay networks. A method is provided for event distribution and routing in a peer-to-peer overlay network that comprises a plurality of nodes. The method includes identifying a plurality of buckets on the overlay network, wherein each bucket includes one or more nodes, respectively, identifying bucket groups, wherein each bucket group includes a selected number of buckets, respectively, distributing events based on the bucket groups, and updating a routing table based on the events. A node includes a transceiver and a processor coupled to the transceiver and configured to identify a plurality of buckets on the overlay network, wherein each bucket includes one or more nodes, respectively, identify bucket groups, wherein each bucket group includes a selected number of buckets, respectively, distribute events based on the bucket groups, and update a routing table based on the events.
Other inventors -
-
Systems and methods for key management for wireless communications systems
Issued US 8,948,395
Other inventors -
-
Methods and apparatus for merging peer-to-peer overlay networks
Issued US 8,825,768
Other inventors -
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content