14 of 14 posting · 139 posts · 1 Jul to 1 OctSince 1 Jul6 weeks4 weeks2 weeks

Hari Srinivasan

Chief Product Officer, LinkedIn EcosystemCompletescanned 1 OctLinkedIn profile
14Posts
13Own words
254Median reactions
22Median comments
300Median interactions

Mostly theirs

Claims

Posts

A few months after joining LinkedIn, I was struggling to adjust from entrepreneur to pm at a larger company. I went to Ryan’s office (a fast-rising exec with at the time with this killer profile) and let him know it was best for everyone if I moved on. The… moreA few months after joining LinkedIn, I was struggling to adjust from entrepreneur to pm at a larger company. I went to Ryan’s office (a fast-rising exec with at the time with this killer profile) and let him know it was best for everyone if I moved on. The next morning, I came into work and there was Ryan, sitting right next to me in our little shared cubicle. He told me he had decided to change desks. I learned more watching him in those next few months, and had more fun, than I ever have at work. And thanks to his act of kindness, I ended up not quitting. Ryan Roslansky, you’re one of a kind leader and friend.  Thank you for always setting the bar on doing well & doing good for each other.
145 reactions · 3 comments · – repostsOpen on LinkedIn
With 20,000+ customers, our little Hiring Assistant is growing up fast. With today's launch, we've made it even better. It's now: -Smarter, using better reasoning models, persistent memory, and improved personalization from past searches. -Better integrated,… moreWith 20,000+ customers, our little Hiring Assistant is growing up fast. With today's launch, we've made it even better. It's now: -Smarter, using better reasoning models, persistent memory, and improved personalization from past searches. -Better integrated, allowing seamless workflows across the hiring stack, new voice calls & more. -More trusted, leveraging our fast growing verified identity & skills evidence graphs. All these features are based on our belief that nobody is taking a job without talking to another person. We aim to reduce the overhead in recruiting, so hirers can focus on what really matters: the talent. https://lnkd.in/gmtwNxts
77 reactions · 7 comments · 3 repostsOpen on LinkedIn
"Who is real" has become the most important question on the Internet. You can't talk, let alone do business with someone, when you don't trust who they are. We’ve made good progress here. 75 people now verify their profile every minute, adding to the 115M+… more"Who is real" has become the most important question on the Internet. You can't talk, let alone do business with someone, when you don't trust who they are. We’ve made good progress here. 75 people now verify their profile every minute, adding to the 115M+ members who have already verified. Along the way, we have learned a lot too.  Every government manages identity differently, there is a cost to validating if a person or company is real, not all companies have work email addresses and more. Overcoming these challenges at scale is hard, but critical to get right. We have a few new features going live to help: 1. We're enabling network verification, promoting verified members to verify connections that they worked or studied with. Obviously there are many edge cases here. Thus, we are building a series of guardrails around this, including ensuring the verifier is already verified, collecting multiple confirmations, time of connection limits, two-factor authentication requirements and more. 2. Company admins can now remove accounts that falsely claim employment from their page (fyi - this will take some time to roll out more broadly) . 3. Rather than requiring everyone to undergo separate, often lengthy, ID checks, you can choose to display your "Verified on LinkedIn" status directly on a growing list of partner platforms, now including Truecaller & PeerSpot
314 reactions · 36 comments · 32 repostsOpen on LinkedIn
Andrew Chimka
LinkedIn's Hiring Pro continues to get better and better. One of my favorite new features is the ability to work together to find the applicants you're the… moreLinkedIn's Hiring Pro continues to get better and better. One of my favorite new features is the ability to work together to find the applicants you're the most excited about hiring. Hiring Pro has the full context on your entire workflow and works side-by-side. You can ask to re-evaluate all of your applicants in seconds if you think of a new requirement, and even put your top applicants into a table to easily compare. Try it - we'd love to hear your feedback! https://lnkd.in/ghrt_PKC
We're keeping at it on reducing AI slop. -There's a new proofread feature. Many people go to AI to double check their work, but leave with a fully rewritten post. The new addition helps proofread and clarify while keeping your voice. -We continue to take… moreWe're keeping at it on reducing AI slop. -There's a new proofread feature. Many people go to AI to double check their work, but leave with a fully rewritten post. The new addition helps proofread and clarify while keeping your voice. -We continue to take what we’ve learned from your AI slop reports to improve AI slop detection and update our classifiers. This includes improving the ways we identify low-quality AI slop images (a key content type associated with AI slop). -We’re also giving members more places to tell us when they see AI slop beyond the feed, including messaging.
649 reactions · 116 comments · 28 repostsOpen on LinkedIn
A few observations after a couple weeks with the new set of personal agents (Muse, Instinct, OAI, etc.): - These agents are going to save us time by knowing a lot about us. Trust is earned, but even without handing over access to my email, it's clear an… moreA few observations after a couple weeks with the new set of personal agents (Muse, Instinct, OAI, etc.): - These agents are going to save us time by knowing a lot about us. Trust is earned, but even without handing over access to my email, it's clear an agent can simplify booking a restaurant with basic context on my family, what time we eat, our favorite places, etc. - Our agents will know each of us, but will not know everything about the world; for example, which restaurant has a table. So they will need many connectors to the world. - The most important 'ux' element for any site that connects with personal agents will be agent management. When thousands of personal agents call the local sushi restaurants as soon as reservations open, who gets the table, how do you manage no-shows, and how do you even know which agent is real? - There is a whole series of agent management mechanisms that will be developed over the next few weeks/months to solve these problems. - Many places without these mechanisms are going to get flooded by these agents. They can opt out, but will lose demand. So there will be a healthy market for solutions that successfully use AI to manage AI.
254 reactions · 45 comments · 1 repostOpen on LinkedIn
We’re testing a new design on the top of the feed: -The Network tab helps you stay connected to your network, ie the people you have explicitly connected or followed. We all want more real connection with real people, and we hope this plays a small part in… moreWe’re testing a new design on the top of the feed: -The Network tab helps you stay connected to your network, ie the people you have explicitly connected or followed. We all want more real connection with real people, and we hope this plays a small part in rebuilding that world. -The For You tab helps you stay informed. We’re all trying to make sense of the world, and we hope this plays helps more people from real practitioners inside and outside your network. There’s more to come here too, but wanted to provide some context if you're seeing something new this weekend :)
536 reactions · 80 comments · 34 repostsOpen on LinkedIn
One more follow up on AI Slop. -First, thank you. Over a million people have now clicked ‘seems like AI slop,’ since we launched it two weeks ago...helping us better understand how our community experiences low-quality content. As a reminder, you can send… moreOne more follow up on AI Slop. -First, thank you. Over a million people have now clicked ‘seems like AI slop,’ since we launched it two weeks ago...helping us better understand how our community experiences low-quality content. As a reminder, you can send this feedback in the three dot menu on any post & comment -This week, we're taking another step. If a post receives enough community feedback, you may see a message in your Post Analytics to help you understand how your content is being received. We approached this assuming good intent; I know I'm increasingly conscious on how to not sound like AI & the goal is to provide helpful feedback. -Importantly, no single piece of feedback determines how content is distributed.  We look at many signals together, and we've built safeguards to help prevent individual feedback from being used to unfairly target other members. I hesitate to give numbers, as everyone has a different network and feed, but overall members are now experiencing 40% less views on what we classify as AI slop from just a few weeks ago. Despite the progress, we know we have more to do to ensure LinkedIn remains a place where you can find real people & real perspectives... this all remains very top of mind.
1,013 reactions · 194 comments · 74 repostsOpen on LinkedIn
AI slop is a top priority for all of us. We really care about this. People come to LinkedIn to connect with real people and share their real perspectives, ideas and expertise. Here are a few more changes to keep it that way. - We continue to improve and… moreAI slop is a top priority for all of us. We really care about this. People come to LinkedIn to connect with real people and share their real perspectives, ideas and expertise. Here are a few more changes to keep it that way. - We continue to improve and invest in our automation defenses. On comments alone, everyday we are now catching hundreds of thousands of automated comment attempts, and have blocked billions of other automation attempts (posting at scale, slop) in the last couple months alone. - We are ramping up a series of new and improved classifiers that identify if a post is AI-slop or generally low-quality content. This will reduce the amount of AI slop you might see in suggested content and content from outside your network. - We are ramping the ability for members to tell us if they believe a post or comment seems like AI slop. Slop is hard to define and the definition changes; this lets us tune our models and make better feeds. - For anyone who shares content, we will test a way to privately flag, in your analytics dashboard, when members feel your post may have come off as inauthentic or heavy use of AI. This approach is based on two learnings.  First, AI and slop are not the same thing; many people refine thoughts with AI, and we believe they want to know when they sound inauthentic. Second, we want members to get feedback from real humans on what sounds authentic - not just have an AI detector review it and get it wrong. - We asked ourselves why do people post with AI anyway?  The answer is LinkedIn isn’t a one-word kind of place and they feel more confident running their posts through AI.   We’re taking this learning, removing the ‘enhance your post’ feature you see when you write a post or message & replacing with a feature that proofreads your words, but does not change your voice. - Finally, we continue to expand profile & pages verification so members know who they are connecting with and who is writing the content they are reading. You’ll also be able to block any comments from Company pages you no longer want to see. There is more to come.  Will do a better job keeping you all updated too.
2,466 reactions · 497 comments · 350 repostsOpen on LinkedIn
To anyone in the Midwest & Canada struggling with the air quality, just a reminder that our family charity, Gift of Air, sends air purifiers to any school or day care. We've shipped to locations across the globe & can usually get them to you in 1-2 days.… moreTo anyone in the Midwest & Canada struggling with the air quality, just a reminder that our family charity, Gift of Air, sends air purifiers to any school or day care. We've shipped to locations across the globe & can usually get them to you in 1-2 days. https://lnkd.in/gCenKrkq You can't learn if you can't breathe. Just message me if we can help.
342 reactions · 14 comments · 28 repostsOpen on LinkedIn
First AI / humanoid related strike I at least have heard of. Workers are asking for: 1. Salary vs hourly pay, to protect against automation. 2. Job-security guarantees (unclear what this means, but includes increasing the retirement age), to protect against… moreFirst AI / humanoid related strike I at least have heard of. Workers are asking for: 1. Salary vs hourly pay, to protect against automation. 2. Job-security guarantees (unclear what this means, but includes increasing the retirement age), to protect against reductions. 3. Bonuses tied to stock & net profits, to get some of the AI upside. https://lnkd.in/gT9_qhRC
51 reactions · 3 comments · 2 repostsOpen on LinkedIn
Hari Srinivasan14 JulRepost with commentaryAI and the future of work
This feels like a great product. Clear pain points, clear time savings & clear respect for the practitioner. Feels rare in AI products; it's not just our benchmark now shows we can do your whole job. Nice work Drew Bent & team. Good for all of us if this really helps teachers teach.
Drew Bent
Today we’re launching Claude for Teachers, bringing the power of premium Claude and Cowork to every US teacher, for free. It works in the background, while… moreToday we’re launching Claude for Teachers, bringing the power of premium Claude and Cowork to every US teacher, for free. It works in the background, while teachers can focus on what’s most important: time with their students. Teachers have been experimenting with AI for some time, but we at Anthropic kept hearing from teachers about the need for something that was aligned to curriculum and evidence based. I think what the team has built is special in a few ways: 1) A foundation in curriculum and learning science -- Claude for Teachers comes with the Learning Commons connector: academic standards in all 50 states, the prerequisite skills beneath each one, and curriculum context from high-quality providers like Illustrative Math and OpenSciEd. Ask for a lesson on constructing linear functions and Claude pulls the standard, notes the misconceptions students typically hold, and builds the lesson around them. 2) Teaching skills, co-built with nonprofit Learning Commons -- These enable teachers to build lesson plans and differentiate instruction. Importantly, they are built to work well in Cowork, allowing teachers to harness the full agentic capabilities of these models. Teachers can control what they share (past lesson plans, student assessment data, etc) and have Claude automatically work for them each evening. It will help them prepare materials for the next day, tailored to their exact students. 3) Privacy first -- We don’t train on teacher conversations and student information is protected by our terms that are written to comply with FERPA. We worked closely with the American Federation of Teachers to align it with their new gold-standard privacy principles. Randi Weingarten, their president, calls it “a tool designed by and for educators... to give them more time for the human relationships at the heart of learning.” 4) Ecosystem approach -- Nine education connectors are available at launch, connecting to tools loved by teachers like TeachFX, MagicSchool, and Brisk Teaching. We’re also releasing a new free AI Fluency for Teachers course with Teach For America. And we’re taking everything we learned and releasing our Skills and evals as public goods, so everyone can leverage them. There is a whole team to thank, including our partners and teachers. I especially want to call out Sofia Wilson and Alex Kasavin, who co-led this. This started as Sofia’s onboarding project, and Alex was her spin-up buddy. In typical Anthropic fashion, we like to give employees overly ambitious spin-up projects (Claude Code was itself one). This was no exception. From the start, the two of them were laser focused on building the highest quality tools for teachers. And it was Elizabeth Kelly and Daniela Amodei who empowered our team to build a product focused on teachers and learning outcomes, first and foremost. If you’re a teacher, you get Claude for Teachers for free starting today: https://lnkd.in/gfRYVmjj
44 reactions · 1 comment · 1 repostOpen on LinkedIn
Really seeking to understand, what's holding people back from 'acting now' to understand the economic impact of AI? Is it a lack of specific data, research funding, forums to discuss, etc? I think everyone agrees there could be change ahead, but not sure… moreReally seeking to understand, what's holding people back from 'acting now' to understand the economic impact of AI? Is it a lack of specific data, research funding, forums to discuss, etc? I think everyone agrees there could be change ahead, but not sure what the people who wrote this (https://lnkd.in/gaT_ZFfX) are asking for here.
47 reactions · 22 comments · 1 repostOpen on LinkedIn