Jay Singh

CEO @ Casper Studios | Ex-Linkedin: We help companies adopt AI and build AI-powered products.

I am a co-founder of Casper Studios, a product studio that develops and integrates AI solutions for companies. I previously was a business development and strategic partnership manager at LinkedIn. With over eight years of experience in these fields, I have a track record of creating value for customers, partners, and stakeholders.

At LinkedIn, I have helped lead the global expansion of identity verification capabilities to key markets, such as India, Europe, and LATAM, enabling over 80 million members to access trusted and secure services. I have also driven 20+ partnerships with leading companies and governments, such as Clear, DigiLocker, and Hyperverge, to enhance the user experience and the network effect of the verification project . 

At Casper, I have leveraged my expertise in AI and product development to create a popular chrome extension that assists users in content creation and summarization, attracting over 20,000 users in less than six months. I have also generated $x00,000s of revenue by offering tailored strategy consulting, product development, and design services to other businesses that want to integrate and leverage AI.

I am passionate about making products accessible, efficient, and user-friendly, and empowering people and businesses to unlock their full potential through technology. I have a strong background in trust, verification, and HR tech, and a keen interest in entrepreneurship and innovation. Outside of my work, I invest time in supporting the AAPI community, love meditating, and dancing with my friends.

Co-Founder & CEO

With a team of experts from LinkedIn and Microsoft, we help companies integrate AI into their workflows, from strategy and prototyping to complete product development.

Schedule a call with us to see how we can help.

Excited to welcome Martijn to Casper Studios as our Head of Growth!

Martijn was one of the first people to welcome me to LinkedIn's BD team back in 2020, bright energy, one of the most authentic humans I've met, instant friendship. I've looked up to him ever since.

His track record speaks for itself:
- Google: Led mobile app distribution
- LinkedIn: Built mobile app distribution that drove 500m+ member growth, plus led subscription growth initiatives
- Microsoft: Founding member of the BD team, partnering with AI startups that raised $100m+ in total funding

He's joining at the perfect time. We've been nurturing small experiments and partnerships for a few years that are ready for someone to operationalize into real systems.

Martijn's that person. He's our first GTM hire outside of me, cannot wait to see what we build together :)

Jay Singh 

CEO @ Casper Studios 👻 | Ex-Linkedin: We help companies implement AI

Just hit 20,000 followers on LinkedIn. The journey started with an essay I read back in college, Kevin Kelly's "1,000 True Fans."

The idea is simple: you don't need millions of followers. You just need 1,000 people who will follow you, support you, and buy what you create.

The math: 1,000 true fans × $100/year = $100,000/year.

I remember reading that and thinking, "I still want to go work at a company and learn skills", which I did at LinkedIn, "but it'd be great to start building a foundation toward this."

Do I have 1,000 true fans right now? I'm not sure! But what I do have is all of you, following, supporting, cheering us on, and connecting us with opportunities to help grow Casper Studios and help people implement AI. LinkedIn has driven 90% of our business to date. 

I'm really grateful for that. It's changed my life. 

People ask how to grow on LinkedIn. Here are three unintuitive things that have helped me:

1.) Time. I've been doing this for almost 10 years. Will Lawrence back at UBC said, "Hey, you should start posting on LinkedIn," and I just started and kept going. You're not going to hit a quick win. You'll probably want to give up. Having a longer time horizon helps. It eases the anxiety of posts that get 3 likes. Who cares? Just show up the next day and try again. You're building relationships with people. That takes time, don't rush it. 

2.) Think of it as a daily coffee chat vs needing to go viral. My frame for posting is to just let you know what I'm up to, the same way you would over coffee. And just like a real coffee chat, opportunities come from the conversation. People think of others to connect you with. They want to help. That leads to more opportunities of depth. I may not have a million people in the audience but there's some incredible people who do, I focus on that.

3. Be the same person online as you are in person. If we've chatted, hopefully you see the same vibe in this post as the person you ended up speaking with. I love that congruency. It means when you reach out, you know what you're getting. That builds trust. And trust leads to deals, relationships, and opportunities, faster. Really proud of Derek, Martijn, Max, Sebastian, Paolo, Leonard, Giorgio and others on the team who are all finding and sharing their own authentic voice on the platform now.

Thanks for the first 20,000. Honestly, I don't really care about growing this number much more haha. If it happens, great. If not, great. 

I'm grateful for everyone who's here and following the journey. Excited to keep building with you all :)

The Casper SF squad is getting strong! Martijn’s been here a week. Elizabeth and Erika joined in November. Derek’s been here from New York since Sunday.

Several of us have been working out of my Nopa apartment. Definitely not a startup garage, it’s a cute place (!), but it’s giving startup garage vibes at the moment.

There’s something about being in the same room. Context just flows. You overhear a call, you jump into a conversation, you learn something without even trying. 

Realizing at 20 people it’s so hard to have the team be up to speed on everything going on, even when I think we’re sharing relevant updates. Need more of this in person time; especially for new hires. 

Reminds me of Jeff Weiner talking about how leaders need to repeat themselves to their teams so many times that they’re sick of their own voice. That’s only when you’re just getting started with making sure that message is understood. 

We’ve already gotten noise complaints from neighbors. Apparently we’re a bit too fired up talking to prospects and customers about AI…

Might have to get an actual office soon. Hopefully not yet though :)

No alternative text description for this image

We just signed our first client of 2026. It made me really proud. They're a major firm with >$100m in revenue, but more than that, the deal hit on a lot of the narratives and themes that are important to us at Casper Studios.

1.) LinkedIn's working: Our GTM is ~90% inbound from LinkedIn. Some of those are direct clients, others are partners that intro us to clients. Ryan Kurt at The AI Lab, someone who previously sold his company to Salesforce, had been engaging with our content for a while. One day he just showed up and said, "I know what you guys do. I've seen your work. I need trusted partners." Over time, that led to him connecting us with a potential customer, and that became this project. He's now an advisor and partner of ours, which is amazing. He's been excellent to work with. He's also cut the BS and been so open and honest from the get go which is 100% the Casper vibe and so nice to work with. 

2.) We went head to head against a VC-backed AI services firm: This firm has raised over $100m. The customer had already been working with them but didn't like their commercial terms. They were focused on owning the IP of what they'd build for the client vs making sure the client owned the IP themselves. That opened the door for us, and we ended up winning the initial first phase. We still have a lot more to go and we may even work directly with this firm at some point because they do great work. But I think this is a tell of how their incentives can become misaligned with customers over the long term. To justify the venture money, they need a Palantir-like story: "We're starting with services, but we're building IP to sell to others, don't worry." That can work sometimes, but in other cases it's just not the best outcome for the customer. 

Good start to 2026.

Proud of the journey Martijn is on. He joined a few weeks ago. He’s already speaking with enterprise companies. Learning Claude Code, then teaching it. Kicking off partner calls with all the major AI platforms. Building agents to automate his work. Sharing his learnings in public.

Maybe most importantly, coaching me and the team on how to be better leaders.

Big year ahead.

Spent the last 3 days onsite with a client in Indianapolis with Kelly and Derek. Super fun and energizing.

I’ll share more about the client later, but the work itself is so interesting. Why?

1.) We’re working with wonderful people. Their Chief Strategy Officer literally picked us up from the hotel and drove us around a few times. They also invited us today to their 100+ person leadership meeting where they introduced us to the team. What a lovely moment. 

2.) The problem they’re tackling supports a population that’s deeply underserved. This is one of the more, if not the most, meaningful projects our team at Casper Studios has worked on.  

3.) The solution they’re building is one of the more innovative things we’ve built. It lives in both the digital and physical worlds. Requires tons of model training. Heavy ML focus. 

I also just love getting onsite. It’s one of my favorite -and most tiring - parts of the job. It’s the best way to spend real time with the team and get to know the clients we’re building with. 

I’m excited to keep spending time in cities outside the coasts and work with businesses there as they start implementing AI.

Over the past year alone, we’ve spent time in Boise, San Antonio, Ardmore (Oklahoma), Cincinnati, Sacramento, and more. I love our NYC and SF clients, but there’s something special about getting out of the bubble, meeting different people, and helping them solve very real problems.

Onwards :)

My team keeps telling me to stop using Claude Chat and switch to Claude Code. As a non technical person, that still makes me nervous but I need to figure this out. I wanted to learn this in public with you.

I’ve asked my team to reallyy hand-hold me and get me up to speed live over the next few weeks. If you want to follow along, join the meeting, and be on the same journey, comment Claude Code below and I'll send you a link.

Our GTM strategy for the next few years is simple: be best friends with everyone! AI platforms, services firms, large consultancies, PE firms. The good companies are all growing fast in this space. So much room to help each other out.

Ria bought me the new Meta × Oakley smart glasses. They are kind of amazing. Look at the video quality of Martijn and I biking last week. It got me thinking about a few things re AI & hardware. 

First, I think this creates a new content format. Walking around with a phone feels a bit old. Moving forward I wonder if people will just have smart glasses on, interviewing each other, capturing content, recording day-in-the-life stuff. 

As long as there’s consent (and I like that there’s a visible white light so it’s obvious when you’re recording), the privacy issues feels solvable. When everyone’s opting in, this just feels more authentic. It gets a bit weird of course if people don’t know they’re being recorded. 

Second, I think there are some interesting applications for our client work. We’re already talking with a large marketing agency about using these for an activation. There’s also a more enterprise angle; esp in healthcare. 

We’re working with a firm where the core application will be powered by a computer-vision model. Today we’re using cameras to capture what’s happening in a room. Over time, as these glasses get better and cheaper, I wonder if it becomes more economical to roll out smart glasses instead of installing fixed cameras. I think we’re still a couple of years away from that being common in practice, but the direction feels interesting.

I’d love for this year to be one where we’re building real AI applications and experiences on top of smart glasses! If anyone’s curious or experimenting in this space, would love to chat :)

Should we start a Casper Studios Bike Club? Can talk about AI. Sounds cringe but I’m excited about it nonetheless haha.

Last week, ~700k AI agents built their own Reddit, founded a religion, established a government with a draft constitution, and started developing private languages. They did this in 3 days. That's started from Openclaw (fka ClawdBot) and the AI world has been losing their minds over it.

I'm still processing what this means myself tbh. It might just be a weird internet thing that fizzles out. But I think it hints at where enterprise AI is heading.

The first wave was "build a chatbot." That was 2023. A conversational interface. Most never made it past pilot - and honestly, often shouldn't have. Chatbots weren't the right way to drive ROI.

The second wave - where most companies are now - is "build agents to automate workflows." That's 2024/2025, probably extending a few more years. Agents integrated into how work actually gets done. This is where real value shows up, and where our team at Casper Studios spends every day.

I think there's a third wave coming: "help us design a network of agents."
Not agents that automates workflows. A network where agents take on different roles and work together - debating, challenging assumptions, surfacing insights no single agent or human would find alone. 

This is probably a 2027 thing for most. Maybe longer for most F100s. Though I'd bet you'll see younger companies experimenting this year. We're seeing experiments of our clients and our team doing this as we speak.

Here's what this could look like. We're working with a hedge fund to build them agents. Imagine for them:
-A deep value agent focused on fundamentals
-A momentum agent tracking price action and sentiment
-A management quality agent analyzing leadership
-A contrarian agent that argues against emerging consensus in tech, one in manufacturing
-A synthesis agent that flags when something interesting is happening with earnings calls

They run for a few days, argue, and bubble up insights to humans. The output is an emerging thesis the team hadn't had the time to articulate, with the reasoning trail attached. You can do elements of this already with Claude Code.
That's one version of it, anyway. 

The safety concerns are severe. This isn't close to ready for production for most use cases. But the design questions - population architecture, agent interactions, culture, human interfaces - that feels like the interesting work ahead. 

If you're a company thinking about where this goes, would love to hear what you're seeing.

There’s that one idea you want to build but never quite get to. Not enough time. Not enough eng capacity. Not worth pulling a team off work. With Claude Code / Cursor I'm finally getting to those ideas.

I’ve had one stuck in my head since last year around Google’s Multimodal Live API. The idea is straightforward. 

You share your screen. The agent can see what you’re doing. You just talk. It’s like someone looking over your shoulder and walking you through a product in real time. Imagine connecting voice and computer vision models into one. 

I used something like this last year to learn Zapier. I opened it up, used the agent, and just walked through it live. Ever since, I kept thinking it would be perfect for learning tools like Cursor and Claude Code.

So I finally built it this weekend. Cursor did what it does best. It researched the Multimodal Live API, scaffolded the project, and wired things together. When it needed something from me, cloud setup, billing, API keys, it told me. I filled in the gaps and kept going.

Agents don’t just do the work for you anymore. They tell you to do the work for them. Weird year ahead.

The first version didn’t worked. Screen sharing was on. Voice was on. The behavior wasn’t great. So I debugged in a loop. Something breaks. Screenshot. Paste it back in. Tried again. Each pass got better.

What exists now is simple. You open it. You share your screen. You can then click to Cursor / Claude Code. The UI stays out of the way. The agent knows how to guide you through Cursor and Claude Code because it’s trained on the same resources I use myself.

I built it in about an hour. There’s a link if you want to try it! This is very much a POC. It might break or burn through API credits. That’s part of building in public :)

We’re finishing the course later this week, and this comes with it. But instead of waiting, you can just share your screen and learn by doing.

This post went a little crazy. ~50 people commented wanting to be part of the journey of me learning how to use Claude Code. We were going to do live sessions, but because of the demand, we're going to host a pre-recorded mini-course instead.

It's probably going to take another week or two to build out. Once the course is done we'll invite you to a Slack group to keep the learning journey going. This will all be free DW!

If you want us to keep you in the loop, just fill out this form in the comments. 

If you need something immediately - especially if you're a business looking for Claude Code implementation services - just DM me in the meantime.

YAY.

I think robots caring for the elderly will be one of the more meaningful uses of the technology in the next 20-30 years. 

My grandma in India has three caregivers and still needs a lot of support from my uncle. I’ve seen how much my parents have poured into supporting their parents. It’s noble but exhausting.

Was just talking with my mom about this. As she and my dad get older, I want to be there for them the way they’ve been there for their parents. But I also know how much it takes.

Robots handling the burdensome stuff means I can actually enjoy my time with them. It also means you don’t have to hopefully consider nursing homes / home care, etc. 

Part of me feels weird about it though. There’s something in me that says I should be the one doing everything, that’s what love looks like, especially in our Indian culture. A robot helping my mom feels cold if I think about it too long.

Then I remember watching my family stress out and be worried about the new caregivers they’ve brought on . The hard truth is that exhausted children aren’t fully present.

I’d rather have a robot handle the late night check-ins and the physical labor so that when I’m with my parents, I’m actually with them. Not stressed, just there.

I think that’s a net positive. What do you think?

There’s this vibe that you need to raise to be competitive in SF. I think raising often increases your competition. Once you take venture money, a “successful” outcome narrows fast; basically a $1B+ exit (if not more).

When you bootstrap and survive the early years, the surface area of successful outcomes increases a lot. You can generate cash flow. You can sell earlier or smaller. You can spin things out. If you want a venture-scale outcome later, you can still go for it, on your terms.

What’s been exciting for us is we’re starting to beat venture-backed AI services firms in deals. At that point, it’s like… who’s the competition? Yourself!

There's been a lot of hype around ClawdBot lately. So what is it, and what actually makes it different from other AI agents?

Three things stand out. First, it meets you where you are, it lives in WhatsApp, iMessage, Telegram, whatever you already text. Second, it remembers things better than most current AI products. Third, it takes actions: sending messages, managing files, and more.

People are calling it the "AI companion we've all been waiting for."

Here's an early take: If you've been using Claude Code or Cursor, you look at ClawdBot and think, "I could already do these automations myself." And that's true, I could rebuild most of these workflows right now if I wanted to. The difference is that ClawdBot does it on its own, without me directing every step.

There was a story on X that captured this well. A user wanted AI to book a restaurant reservation. ClawdBot checked OpenTable but couldn't find an opening. So instead of stopping there, it created a skill using an ElevenLabs voice agent, called the restaurant directly, and booked the table.

The user just described the problem in a chat interface, and the system figured out the solution. That's the nuance worth paying attention to.

Right now, Cursor and Claude Code use the terminal as their interface. You're building agents, configuring workflows, writing prompts. It works but it doesn't feel natural for mainstream consumer adoption. I think enterprises still have a lot of opportunity to implement these platforms FWIW.

As a consumer, this reminds me of when I used Playground before ChatGPT launched. Playground was doing almost the same thing under the hood. But ChatGPT's UI is what allowed everyday people to actually start using it.

The way I see it: ClawdBot (or something like it) is to ChatGPT as Claude Code and Cursor are to Playground. Same engineering, same model, different interface.

Security is a real concern here. You're giving an agent access to your entire computer. That can cause some damage. Some people are even buying Mac Mini devices to test this out. I do think that'll get solved over time.

Will ClawdBot be the product that takes over? Too early to tell. I sense that larger tech companies are working on similar offerings. But it's clearly hitting on a real problem: you want the functionality that exists within Claude Code, packaged into a more accessible chat interface. This is likely the first example of that.

If I had to bet, I'd put my money on this type of interface being the future: something text-based, something easier to use, something that looks more like a document over time. Not sure if I'd say it's going to be ClawdBot. Could just be coming from Anthropic, OpenAI , or another big tech co. 

We're still learning this ourselves. Maybe it's something we can help clients with over time. We'll keep a close eye on this as the Casper Studios team to see how it plays out.

MCPs are way more functional on Claude than ChatGPT or Gemini. This is why I use Claude exclusively now. They've got all my info, Fireflies, Gmail, Google Drive, and the responses are incredible. It looks through everything and generates the right answer.

Here's an example of a workflow: I get an email asking if we had experience building a specific AI automation for a specific niche / industry from a prospect. At 20+ clients it's hard for me to have context on the minutiae of every project across our team but Claude + MCPs into our data does. 

I just drop the question into Claude. It runs through all our documents, all our Fireflies notes, and gives me the right answer. Super crisp. That's what model capabilities look like when they're connected to your data.

The faster Antropic go on these integrations, the more personalized the responses, the stickier the product gets. I don't think people are thinking about this enough. It started with a small MCP protocol they created, but I think we'll look back at that as the most important invention they made for user adoption.

The evolution of the "Blank Page" for knowledge workers:
-2020: Open a Google Doc.
-2023: Open ChatGPT or Claude.
-2026: Open an IDE (Cursor, Claude Code).

It sounds wild, but my workday no longer starts in a chat interface. Whether I’m working on sales, marketing, or recruiting, I’m starting in a code editor. That habit has changed of Tuesday this week haha. 

I think we’ll eventually come full circle: a simple, doc-like interface with the raw power of an IDE hidden underneath. For now, the "dev tools" are simply better at getting the job done.

Excited to keep sharing these learnings over time. 

Claude Code implementation / education will become a key service area for Casper Studios. I.e., I think there's a lot of value in simply implementing Claude Code + Cursor into companies. 

If you're one of those people interested DM and lets chat.

Reflection: ~80% of our clients’ decision-makers - CEOs, Heads of AI, CTOs, MDs, Partners - are using some form of vibe-coding tool every day. It doesn’t matter how big the company is. Billions in AUM, funding, or revenue they’re in the tools themselves. 

That’s one key thing you need to cultivate as a leader to drive AI adoption. 

You see what’s possible. Then you go to your team and honestly, pretty sternly you push them to try it too. And later, when you notice a process that should be AI-first but isn’t, you call it out and push again. 

That’s the hardest part. It’s pushing against the identity of your team. It’s what they’ve likely spent years building their skills to do. You’re pretty much saying “thanks but try it a completely new way”. Over time those that have high agency and excitement will love this process too but it takes a lot more effort than you think. 

This has become a core part of a leaders job now. Why? Because it’s basically an extension of what they already do, hiring, culture management, setting the pace. Enabling your team to adopt this stuff is a leadership priority now. 

Call it the “AI-Native Executive” or whatever, but this should be in every CEO’s job description going forward.

This is exactly what I’m trying to do with our own team. Especially the non-technical folks.

If you’re a leader and you’re not there yet DM me. Let me get you up to speed. Even just a 30-minute screen share of what’s currently possible has been blowing people away.

Trying to write a newsletter again! Been doing it on and off since I was a freshman in college haha. This time I'm sharing my journey learning coding agents like Claude Code. This post is on I got set up, what surprised me, what's next.

Over time I think this will become more about general insights from our work in AI with Casper Studios. 

There's a lot of noise out there. My goal, as it always has been, will be to make the complex concepts feel more attainable and simple. It'll be fun to do that through a longer form forum again where I can expand on the quick hits I'm sharing here on LinkedIn.

For those interested in following along with the course our team is building for learning Claude Code, signing up here would be the best way to stay in touch (unless you've already filled out the form, then dw!).

I'll put the post in the comments for you :)

Simple and powerful use of Claude: connect it with your notetaker and your file storage. When someone asks a question, just drop it into Claude and say "hey based on everything you know, what do you think I should reply with - include specific mentions of the client work we've been doing."

At our scale of 10+ active clients running, I'm not in the weeds on the day-to-day learnings per project. Most times I kind of know the answer but I forget the details because I'm tired from the day. I'm not in every internal meeting either. But Fireflies is.

So when a prospect asks about a specific financial services use case, I can query Claude and say "how would you reply to this?" - and now more often than not, it gives me an 80% done draft with a super specific example of something we've actually delivered before. From there I do guide it and massage it based on my own knowledge / taste.

Honestly the prompting isn't that magical. The magic is just using Claude Opus and connecting Fireflies + Google Drive; or whatever your systems are too.

I'm endlessly curious and deeply motivated by how emerging technologies can shape a better future. I’ve worked across AI agents, AI developer tooling, web3, robotics, and cybersecurity bringing a strategy-driven, execution-focused mindset to building partnerships, driving adoption, and exploring new frontiers. I’m always learning, always experimenting, and driven by a belief in democratizing access of technology to all. That mission has shaped my work across startups, Big Tech, and nonprofit impact.

Maximilian Stein

San Francisco was buzzing last week, nonstop events, amazing people, and a ton of excitement in the air. I had an incredible time with the MongoDB team sponsoring and supporting their hackathon, attending .Local, and joining several of their curated gatherings for founders and investors. 

I’m excited about the work we’re doing together as MongoDB rolls out their startup program and deeply partnering with Fireworks for Startups. We're working closely to enable founders at every stage of their AI development. 

If you’re building and thinking about how to get the most out of open models including fine-tuning, check out the Fireworks for Startups program below. We’re excited to help you build, iterate, and ship faster! 

Last photo: didn’t realize I’d be coming face to face with a humanoid robot like this 🤖

Startup Partnerships at Fireworks AI • Ex-Microsoft • Co-Founder of empyre webs

taplio

The Battle

Jay Singh

Maximilian Stein

Followers

Posts

Engagements

Likes

Performance ScoreJay Singh's LinkedIn Performance

Performance ScoreMaximilian Stein's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Jay Singh

Maximilian Stein

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Jay Singh's LinkedIn Performance

Performance ScoreMaximilian Stein's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreJay Singh's LinkedIn Performance