Sign in to view Prathmesh’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
San Francisco, California, United States
Sign in to view Prathmesh’s full profile
Prathmesh can introduce you to 8 people at MCPJam
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
3K followers
500+ connections
Sign in to view Prathmesh’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Prathmesh
Prathmesh can introduce you to 8 people at MCPJam
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Prathmesh
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Sign in to view Prathmesh’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
About
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
Activity
3K followers
-
Prathmesh Patel shared thisThanks to everyone who came out to "Fix My Agent" at Hogpatch last night! I talked about how we test & debug failures in MCP servers at MCPJam across the User-Value Chain and across clients like ChatGPT, Claude, Cursor, Muse. We covered running swarms of agents acting as your users in isolated client environments, human user testing across clients, and running evals across clients + models, and automating checks against best practices. We had a "Live Fixes" portion where people brought their own agent and MCP workflows that were breaking, and we opened them up together in the room- biggest takeaways I had from attendees: - MCP cross-client failures are top of mind- one attendee put it: "I thought MCP was a standard, but then every client decides to support capabilities and not and I'm left questioning what really is standard". - MCP is key to distribution & software discovery, but knowing how to do this well is hard. Especially as more and more capabilities (like ChatGPTs Plugin Extensions as part of OpenAI Dev Day) are built on MCP. Thanks to AI Valley, PostHog, and GMI Cloud for hosting, and to Courtney Ko, Thierry Damiba (Arcade.dev), Hendrik Krack (CodeRabbit), and Mingjun Sun (GMI Cloud) for great talks! If you missed it and have an MCP server that's misbehaving across clients, send me a DM and we'll help take a look! #mcp #modelcontextprotocol
-
Prathmesh Patel reposted thisNorth Carolina School of Science and Mathematics
North Carolina School of Science and Mathematics
6dPrathmesh Patel reposted thisWe are proud to announce that NCSSM has once again been named the #1 Best Public High School in America by Niche. This is the fourth time NCSSM has been honored with this title since 2023. NCSSM has also been named the #1 Best Public High School in North Carolina for the seventh year in a row. We are honored by this recognition of our commitment to our mission in Durham, Morganton, and across the state of North Carolina. We continue to strive to accept the greater challenge and provide opportunities to our phenomenal students. -
Prathmesh Patel shared thisI'm speaking Tuesday, Sept 29, at "Fix My Agent: An AI Agent Clinic" at Hogpatch in SF 📍 with AI Valley, PostHog, and GMI Cloud! At MCPJam we build testing and debugging tools for MCP servers, so we see a lot of agents that fail to deliver value when connected to MCP servers in production. The cause is usually one of the same few things: - Tool discovery issues: descriptions the model misreads, so it calls the wrong tool or passes bad arguments - Connection & OAuth issues: connection & OAuth flows that pass locally and break with real AI clients - MCP servers that work well enough for users with one client+model configuration, and fails with another - No continuous evals practice, so every prompt or schema change is a guess In my talk I'll go through how we find these problems. That means inspecting an MCP server directly, tracing tool calls, debugging OAuth step by step, and running evals across models before your users find the bugs. If you have an MCP server or agent workflow that's misbehaving, bring it. I'll be there all evening and happy to open it up in MCPJam with you! Also look forward to hearing from Thierry Damiba at Arcade.dev, Hendrik Krack at CodeRabbit, and Mingjun Sun at GMI Cloud! Oh, and free food, drinks, and swag 👀. RSVP: https://lnkd.in/g2uZH89h
-
Prathmesh Patel shared thisChatGPT now lets people connect more than one account to the same plugin. MCPJam supports it today: you can connect multiple accounts to one MCP server and test how your server behaves. tl;dr on why this matters Agents are becoming an always-on assistant- that needs all your accounts. Users would benefit from connecting multiple accounts to their server/plugin- top of mind: Gmail/GCal How it works: → Connect a second account to your server/plugin, and ChatGPT adds a `link_id` enum of your connected accounts to each tool. → It also adds the account list to the end of the tool description, so the model knows which account it's picking. → When the agent calls the tool, ChatGPT takes `link_id` out of the arguments and puts that account's token in the request header. Your server never sees `link_id` → Optional: if your server already has a tool that returns the current user (e.g. `whoami`), set `"openai/profile": true` in the _meta tag. ChatGPT calls it with each account's token and uses the returned name, email and stable ID to label the account. Should be fairly seamless- handled on the client side and your server doesn't need any changes. An ask from me: major AI clients should settle on one standard here- awesome for ChatGPT to kick this off. Test it out today! Connect multiple accounts and test your server: https://mcpjam.com OpenAI's guide: https://lnkd.in/gtz4QtAM Connect your own accounts in ChatGPT: https://lnkd.in/g_sGpGtd #mcp #modelcontextprotocol #plugins #chatgpt
-
Prathmesh Patel shared thisToday, we're launching MCPJam as the first testing & evaluation platform for MCP servers! Your users often aren’t in your product anymore. They're in AI clients like ChatGPT and Claude using your MCP server. You can’t see what they prompted for, how the agent interpreted it, or whether your server actually delivered value. Most MCP servers FAIL to deliver value to their users. Over 106,000 developers and +300 enterprises use our solution to see how their servers behave locally across major AI clients. Now, MCPJam helps teams answer a hard question: what does "good" look like for MCP? Improve experiences for your users across AI clients, and build continuous reliability into every deployment, open-source! - With Swarm, use agent personas to simulate your users and uncover user prompt patterns and failures across AI clients - With User Testing, let real users deliver feedback on their experience so you know what to fix - With Evals, turn user workflows into durable tests cases to run across every major AI client and across models - With CI/CD, gate every change and ensure you’re delivering value & catching failures before they reach your users Check us out on Product Hunt today: https://lnkd.in/gWw6pEXW We can't wait to see what developers build with this, knowing the software that reaches their users is reliable. #mcp #modelcontextprotocol
-
Prathmesh Patel shared thisThanks to Pamela Fox, Liam Hampton, the MCP Community Connect organizers, and GitHub for hosting us in San Francisco! I spoke about 🎤 “Effective MCP: Building Reliable Stateless MCP Servers” and a question that should guide how we build: How do you know your MCP server is delivering value to users? We covered: • Patterns from testing leading MCP servers: where agents fail going from user intent to user value, and how those failures vary across AI clients • Where context belongs in a stateless MCP server: across the agent/client, your server, the requests • How to build a testing ground where you can see the user’s intent, the agent’s actions, and the outcome across AI clients • How to turn user journeys into repeatable evals across clients, so you can continuously catch breaking user outcomes before you ship Thanks to everyone who joined! I’d love to hear how you’re testing your MCP servers and what you’re learning. Slides down below 👇, recording out soon! MCPJam #modelcontextprotocol #mcp
-
Prathmesh Patel shared thisI'm speaking TODAY at MCP Community Connect at 📍 GitHub HQ in San Francisco @ ⏲️ 3:30PM! How do you know your MCP 2.0 server is delivering value to your users? 🎤 "Effective MCP: Building Reliable Stateless MCP Servers" In 20 minutes, I’ll walk through: • Patterns from testing leading MCP servers: where agents fail going from user intent to user value, and how those failures vary across AI clients • Where context belongs in a stateless MCP server: across the agent/client, your server, the requests • How to build a testing ground where you can see the user’s intent, the agent’s actions, and the outcome across AI clients • How to turn user journeys into repeatable evals across clients, so you can continuously catch breaking user outcomes before you ship Thanks for having us Pamela Fox! Come join us! Excited to share and would love to hear what you're building! MCPJam #MCP #modelcontextprotocol
-
Prathmesh Patel shared thisMCPJam launched WebMCP Inspector. 🌐 We’ve integrated a browser into our chat client so you can test agent workflows that require browsing. On top of that, we’ve added support for WebMCP debugging. Some context: - Browser interactions let an agent navigate pages and interact with a website’s UI. - WebMCP is a web standard that lets websites expose structured tools for agents to discover and call in the browser. - ChatGPT & Codex released support for WebMCP For example, a travel website could expose a “search_flights” tool with explicit inputs for destination and dates. An agent can use that tool as part of its browsing workflow. For developers, that means testing how browsing and tool use work together. Can the agent find the right tool, use it correctly, and complete a user’s task? With WebMCP Inspector, you can test these workflows and debug your WebMCP integration alongside the conversation in MCPJam. If you're building with WebMCP - give it a try and let us know what you think! #mcp #modelcontextprotocol #webmcp
-
Prathmesh Patel shared thisExcited to share I'll be speaking at AGNTCon + MCPCon in San Jose this October! My talk: "Rethinking CI/CD Release Gates for Agent-native Software" How do you know your software is "good" and ready-to-ship for agents consuming it? Agent-native software needs expanded testing standards, but most MCP server and tool-based interfaces ship without them. In the session I'll share how to build an automated release gate built for agent-facing software that combines deterministic + probabilistic checks across reliability, security, protocol conformance all across the "User-Value Chain" to ensure your software is ready for production. These gates also provide a feedback loop needed for self-improving software to autonomously optimize for prompts and patch code in your CI/CD pipeline. Hope to leave you with a concrete CI/CD checklist for safely shipping agent-facing tools, grounded in the patterns we've seen across thousands of companies and servers at MCPJam! See you there! Talk: https://lnkd.in/d3Q6YzK4 #mcp #modelcontextprotocol #cicd #quality #reliability
-
Prathmesh Patel reacted on thisPrathmesh Patel reacted on thisI'm hiring Data Platform Engineers to join Mercor's Data Platform team in San Francisco. You'll own the infrastructure every data-driven team here runs on: Snowflake, Fivetran, Airflow, and Hex, building clear ownership, access, and governance models across the company. Looking for someone with hands-on experience migrating live production systems reliant on data, without breaking dependent workflows. Huge plus if you can also build out internal tooling for the data platform team from 0-1. If this sounds like you (or someone you know), please apply!
-
Prathmesh Patel reacted on thisPrathmesh Patel reacted on thisStill thinking about Fix My Agent 🩺 We brought builders together to talk about the unglamorous but important work of getting agents beyond the demo: reliable MCP workflows, secure agent actions, code review, model routing, and more. Also, had a lot of fun demoing a cute Doctor Hog game I made with GMI Cloud MCP 🦔 Big thank you to our co-hosts PostHog and GMI Cloud for making the event happen. And thank you to our amazing speakers: • Prathmesh Patel, MCPJam • Thierry Damiba, Arcade.dev • Hendrik Krack, CodeRabbit • Mingjun Sun, GMI Cloud I loved seeing people bring real questions, projects, and production problems into the room. These are the conversations that make the AI builder community so fun. More to come from AI Valley ✨
-
Prathmesh Patel liked thisPrathmesh Patel liked thisSometimes changing someone's future starts with changing what they see. Ben Carson's mother understood this before anyone knew what her son would become. She didn't have much formal education herself. She worked as a maid and raised her children with very limited resources. But while working in someone else's home, she saw something that stayed with her. Books. Lots of them. She saw a world of knowledge that she hadn't had access to herself. And she wanted her children to have access to it. So she changed a simple thing at home. Less television. More reading. She made her sons read books from the library and write reports about them. At first, Ben didn't like it. But slowly, something changed. He started learning about people, places, ideas, and possibilities that were much bigger than his immediate surroundings. His grades improved. His confidence grew. And eventually, the boy who struggled in school became a neurosurgeon. But the part that stays with me isn't what Ben became. It's what his mother did before she had any proof it would work. She changed his inputs. And his inputs changed how he saw the world. I think that's true in business too. The people we listen to shape how we think. The things we read shape the questions we ask. The environments we spend time in shape what we believe is possible. Sometimes we focus too much on changing the person. When changing what surrounds them can be enough to start the process. So here's the part worth remembering: You don't always need to give someone more. Sometimes you just need to show them a bigger world. What book, person, or experience changed the way you saw what was possible? ♻️ Repost if this landed. ➕ Follow Kanan Alibayov for more on building, growing, and learning. #Leadership #PersonalGrowth #Education #Parenting #Success
-
Prathmesh Patel liked thisPrathmesh Patel liked thisWell that’s a bit different color gray If you cross-examine the anatomy “Bias and Motive are always Relevant.” United States v. Abel. 469 U.S. 45 (1984) https://lnkd.in/gcmhMVzx
-
Prathmesh Patel liked thisPrathmesh Patel liked thisKindness Is The One Thing AI Cannot Fake Into Meaning. We live in a world where people see something extraordinary and immediately ask: “Is that AI?” But not everything beautiful needs an algorithm behind it. Sometimes it is simply a human being choosing to show up for another human being. An 11-year-old boy turns grief into purpose, raising money for the hospice that cared for his mother. Then someone with an enormous platform walks into the room, shakes his hand, listens to his story, celebrates his courage, and creates a memory his family will carry forever. That is not artificial intelligence. That is human intelligence. Human warmth. Human empathy. Human connection. Technology will continue changing everything around us. AI will become as commonplace as the internet, smartphones, and automobiles. But technology should advance the human experience, not replace the humanity inside it. No model can substitute sincerity. No algorithm can replace affection. No automation can manufacture what it feels like when another person genuinely sees you. That is why leadership still comes back to something beautifully simple: Lift others in service. Give back. Stay humble. Use whatever platform you have to make somebody’s day a little brighter. And show the next generation what good looks like. Because the future absolutely needs more AI. But it needs even more HI. Humanity. And that remains our greatest technology of all. #Leadership #Kindness #Humanity
-
Prathmesh Patel liked thisPrathmesh Patel liked thisThis video shows how psychopathical fathers destroy the things their children enjoy (that are usually an escape from the abuse), in order to push their buttons to make them appear mentally ill. I am guessing that most people who come on this scene will side with the father, in his portrayal of the son as a no-go layabout with anger-management issues. But look at what is going on here. This represents a source of enjoyment and probably escape from the abuse he has at home. This is the bit people don’t see. So what does the father do? Destroy or spoil it! People do not realise what the treehouse or video game represents – it is an escape from the abuse and the child is being deprived of that source of escape – so naturally it will provoke a trauma response, and the father knows this. That is why he is doing it. My grandfather’s way of escaping the psychological abuse of his son (my father) and his wife (my grandfather), was to visit his sisters (my great aunts), and local pubs. He also escaped into alcohol and tobacco, often visiting local pubs. So how did my father portray this – as the victim of an alcoholic father. That as a child he never had anything because of his father’s (my grandfather’s) drinking. My grandfather did not live to see 60 and my father keeps his gold shirt studs as a souvenir of the man he drove to his death. Similarly, when my sister helped to finish off my mother (using similar abuse techniques), my father gave her my mother’s and her mother’s (my grandmother’s) wedding and engagement rings (promised to me), as a reward for helping to abuse and finish her off. Similarly, with their abuse of me. At the age of four I was sent to a convent school because they had a reputation for ‘discipline’ (see previous posts about Ruby Frankie for analogous examples). To this day, my father does not boast of my achievements (I have three degrees, including an M.Phil. from Cambridge) – see this profile for details. Instead he portrays me as a waster and my sister (who has a cleaning company) as the golden child (see post on toxic family roles) Portraying decent people as scum and vice versa is a common tactic of the psychopathical parent – they know it upsets and winds people up. When I mentioned this to the police they said ‘people are entitled to their opinion’. The chances are the young man in this video would not come across in interviews, especially if triggered, or confronted by Jobcentre staff who take a bullying approach similar to the boy’s father. (see also my post on Jobcentre staff) The same is true of politicians and others, who portray young men like this as flakes, layabouts, wasters…and the problem with this society (Gen Z) As I have said a number of time, society needs to vilify and separate the abusers, not dog-pile on the victim, in an endorsement of the abuser’s tactics.
-
Prathmesh Patel liked thisPrathmesh Patel liked thisWhat if dyslexia changes how you see words, but never limits what you’re capable of seeing? This scene from Pearl Harbor (2001) has always hit hard for me. Rafe can see the letters. He knows what’s there, but dyslexia changes how his brain processes what he sees and a system designed around one definition of ability nearly overlooks the person standing in front of it. I know that feeling personally. I have dyslexia. I wasn’t diagnosed until later in life, but once identified it explained so much. As it turns out, it runs in my family. My mother, brother, nieces, nephew, and daughter all understand this feeling. Growing up, I misread words, pronounced things differently, and often took three times longer to read something than others. People corrected me - sometimes kindly, sometimes through teasing. Even as an adult, I once had a manager berate me for repeatedly mispronouncing a team member’s name because I genuinely struggled to hear the difference between the J and G sounds. Experiences like that stay with you. Twenty-five years after this film was released, we’ve come a long way in understanding dyslexia and learning differences, but understanding requires more than accommodations - it requires compassion. Before correcting someone, consider how they may be hearing, seeing, or processing the world around us differently than you do. Dyslexia has never defined my intelligence or potential, but it has shaped how I lead - with greater patience for how others think, learn and communicate … and I’m in good company. Some of the most creative, innovative problem-solvers I know have dyslexia, perhaps because we grew up learning to find work arounds quickly. Great leaders don’t mistake different for less capable. They create room for people to think differently - because sometimes the person who sees the problem differently is exactly the person who sees the solution no one else does. Psalm 139:14 reminds us that we are, “fearfully and wonderfully made.” Dyslexia isn’t something that makes a person less capable … our minds simply process and see the world differently. Perhaps leadership begins with having the patience to understand those differences rather than judging them, and great leadership begins by recognizing ability that a traditional system might miss. #Dyslexia #Neurodiversity #Leadership #InclusiveLeadership #LearningDifferences #Empathy #FaithAndLeadership
Experience & Education
-
MCPJam
***
-
*****
********* ***** *** * ********* ********
-
******
******** ********
-
**** **********
******** ** ******* * ** ******** ******* undefined
-
-
***** ******** ****** ** ******* *** ***********
-
-
View Prathmesh’s full experience
See their title, tenure and more.
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
or
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Courses
-
Applications in Entrepreneurship
IE450
-
Computer Architecture
CS250
-
Database Systems
CS316
-
Design and Analysis of Algorithms
CS330
-
Intro to AI
CS270
-
Linear Algebra
MATH218
-
Modeling with Differential Equations
MA454
-
Multivariable Calculus
MATH212
-
Operating Systems
CS310
-
Research in Chemistry
CH442
-
Strategies for Innovation and Entrepreneurship
IE352
-
Web Application Development
CS290
Projects
-
GameDash.AI
-
Chrome extension that brings real-time NBA scores, highlights, and stats directly to your browser, without leaving your stream
Honors & Awards
-
1st Place - Award of Distinction - Creative Communication Challenge
Singapore International Science Challenge
-
Best Presentation - Creative Communication Challenge
Singapore International Science Challenge
-
Summit Diplomat and Semifinalist
Conrad Spirit of Innovation Challenge
Semifinalist and Summit Diplomat for the creation of an app called Gas Pass.
Gas Pass is an app that revolutionizes the purchase of gas.
Test Scores
-
ACT
Score: 35
View Prathmesh’s full profile
-
See who you know in common
-
Get introduced
-
Contact Prathmesh directly
Other similar profiles
Explore more posts
-
Noel Moldvai
Augment • 9K followers
Thanks Ryan Lawler for covering our raise. “Investors are clamoring for new ways to acquire pre-IPO shares as high-growth startups stay private longer… Augment focuses on high-demand, late-stage names like OpenAI, SpaceX, xAI, Databricks and Canva, but has roughly 300 private companies on its platform… ‘You can go from seeing an offering to owning the shares in five minutes,’ Moldvai says.” Read the full Axios article here: https://lnkd.in/gn-K8-h7
44
6 Comments -
Afraem Ahsan
Khulna University • 1K followers
San Diego-based Manifold (AI Detection and Response platform) has raised $8 million in seed funding; lead backer Costanoa Ventures, along with several institutional investors including Cherry Ventures, Rain Capital, Modern Technical Fund, and angel investors.
4
1 Comment -
Hadley Harris
ENIAC Ventures • 23K followers
In my experience, ~17 people is the tipping point where an org stops operating as a single atomic unit and starts fragmenting. Fragmentation kills speed. AI lets you push scale without pushing headcount. With all the tooling we're building, I don't see why Eniac Ventures would ever need to exceed 17 That's the future of VC
43
10 Comments -
Enke Bashllari, Ph.D.
Arkitekt Ventures • 9K followers
Major drugmakers are partnering with AI x Bio startups at breakneck pace. And these aren’t small deals. They’re multi year, high stakes bets on AI as core infrastructure! - Fresh this week Insilico Medicine Medicine expanded its partnership with Eli Lilly with $100M in upfront and milestone payments (Nov 2025) - Accipiter Biosciences came out of stealth with a collaboration and license deal with Pfizer and with Kite, a Gilead Sciences company (Nov 2025) - Manifold Bio signed a $55M upfront deal with Roche with up to $2B in potential milestones (Nov 2025) - Nabla Bio struck a $1B deal with Takeda (Oct 2025) - Algen Biotechnologies announced a $555 million deal with AstraZeneca (Oct 2025) - Variational AI announced a deal with Merck with the potential value of $349M (Sept 2025) - Superluminal Medicines Inc. Medicines inked a $1.3B deal with Eli Lilly (Aug 2025) More and more keep showing up on my feed each week. At this point, partnering with big pharma looks less like a potential go to market path and more like *the* business model for AI in life sciences. But is it the right one? If AI platforms depend on back loaded payouts, pharma still holds the leverage. If upfronts grow, we may finally see both operating as true discovery peers, sharing risk, data, and value creation on equal footing. Anyone have insight into how these deal terms are actually shaking out? Image Credit: Fierce Biotech
72
7 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Prathmesh Patel in United States
-
Prathmesh P.
San Francisco Bay Area -
Prathmesh Patel
United States -
Prathmesh Patel
Cherry Hill, NJ -
Prathmesh Patel
Dallas-Fort Worth Metroplex
8 others named Prathmesh Patel in United States are on LinkedIn
See others named Prathmesh Patel