Ankita Gupta
@ankitaiitr
Followers
2K
Following
6K
Media
357
Statuses
2K
Cofounder and CEO @aktodotio | @iitroorkee @Dartmouth
San Francisco, CA
Joined September 2009
MCP + AI agents = speed. Also: a new attack surface with new risks. Over the last few months, in all my conversations with CISOs and Security leaders, a pattern kept repeating: Teams are racing to ship MCP servers, tools, and AI agents… The old security playbook is
1
1
5
Achieving extraordinary outcomes requires extraordinary effort. Working hard & smart > working smart
Last night I left the @xai office after ~36 hours of working with no sleep. Although I was dead, I was also super energized. Incredibly grateful to be a part of this team. Happy thanksgiving!
0
4
15
Now in private beta: Aardvark, an agent that finds and fixes security bugs using GPT-5. https://t.co/xwtJhfDM3X
298
717
5K
B2b marketers, Marketing in b2b Enterprise is as much about emotionally touching the buyer as it is in consumer. The difference is that emotion is related to career growth, learning, job success •“Will this project make me look like an innovator?” •“Will I learn something
0
0
0
With the speed at which AI is evolving...if you’re comfortable in your job, you’re in the wrong company or the wrong job. Your peers in AI are growing faster than you.
0
0
1
The biggest Agentic Security conference! 1000+ folks joining… Tomorrow
0
0
2
I am hiring for 5 roles right now. In the first 5 mins of every call, I say: “This is on-site.” So surprised by answers! Every single candidate I am talking to says “We prefer on-site. Some said they are leaving their current job because it’s remote” I didn’t see this in
1
0
2
T-13 days left Only 2 weeks until the first-ever MCP and AI Agent Security Conference. Learn from industry leaders, see real attack scenarios, and take home defense strategies you can use right away. 📅 Sept 30, 2025 | Virtual 👉 Register now: https://t.co/608MDAfiq1
0
1
1
17 MCP security attack types across five key MCP assets: prompts, tools, resources, metadata, and configuration - By @ankitaiitr 1. Prompt Injection – Malicious prompts change how the agent behaves. 2. Tool / Service Misuse (“Confused AI”) – The agent uses the right tool in
0
5
14
One of the most powerful messages I have listened to in weeks is by @BarackObama "Just learn how to get stuff done, and people will notice."
0
0
2
Current Trend in Bay Area: Leave Google, amazon, meta, Cisco, f500, some startup Join AI startup That’s it. Everything else is boring, only AI startup is hot!
0
0
5
Building agents…. Building working agents Two different things. Every single dev I talk to tells me how hard it is and how steep the learning curve is. If you are a dev, start building them now if not already.
0
0
4
“MCPs are here. So what?” That was the question someone asked me a few weeks ago. Here’s the "𝘀𝗼 𝘄𝗵𝗮𝘁" of MCP risks ⬇️ MCP fundamentally changes the risk profile of enterprises. Why? Because it turns ordinary text and data into live, executable instructions for AI
0
0
1
Every time someone is mean and you don’t respond, the urge to do so feels overwhelming. Wait, if you can. Sometime later, you’ll be glad you didn’t. Silence is a response too and sometimes the best one.
0
0
2
The enterprise world today desperately needs an AI agent to fill RFPs 🤔
1
0
1