Mohankumar

Co-Founder & CEO | Securing AI Agents

[Mohankumar @Vimokumar]

@AiraSecurity has been selected as one of the 35 startups globally in the 2026 cybersecurity startup accelerator. We are building to help teams get visibility and control over how AI agents behave in real time with the right context.
press:
https://t.co/8EpPs1Lo7K

[Mohankumar @Vimokumar]

Agentic traffic is now more than 70% of all inference traffic... crazy to see how much has changed in just the last few months.

[Dario Amodei @DarioAmodei]

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.

Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.

You can read the full post here: https://t.co/OGyPb7yaYt

[Mohankumar @Vimokumar]

6 months ago I posted this. @AnthropicAI dropped a blog post and cybersecurity stocks closed down ~9%.
earlier debate: does AI replace security vendors?
Today, the opposite. After @DarioAmodei @sama talked about slowing down frontier capabilities.. cyber stocks are up by 10%.

[Wickey @Wickey_WW]

https://t.co/jkIUWFiFU5

[elune @elune0x]

Anthropic engineer:

"Most people use only 10% of what Claude can do

The other 90% is loops and graphs"

Chat → Agents → Loops → Graphs

In 27 minutes, he explains how Anthropic actually uses Claude and why most users never reach that level

Most people open a chat, write a good prompt, and stop when it works

He closes the chat and builds a system that keeps running without him

Same model, same tokens, completely different workflow

This is worth more than most paid graph engineering courses

Watch it today

Then read the full graph engineering guide below and save it before everyone catches up ↓

[JS0N Haddix @Jhaddix]

Little tip from our hackbots course. Do this AND give your agents the full path for everything... always... no exceptions. Skills, tools, and full defined syntax.

You'd be surprised how often your hackbot will not find a skill or tool and make up its on solution... Write its own tool, select its own syntax, or skip the methodology step all together.

https://t.co/VNyDu8Dd03

[Mohankumar @Vimokumar]

aircrafts passed less than 50 feet above the roof of Cape Town’s DHL Stadium..
this is what good risk management makes possible.

[Mohankumar @Vimokumar]

Nvidia is buying Hugging Face for ~$12.9B.
Hugging Face makes only around $150M a year, but became part of the default workflow for developers building with AI... that kind of distribution and developer habit is more valuable than the revenue it generates today.

[OffSec @offsectraining]

🚨 CVE-2026-72898: A critical unauthenticated SQL injection vulnerability in Metabase is being actively exploited.

The vulnerability affects the password-reset flow and can ultimately allow an attacker to gain administrator access to a vulnerable Metabase instance.

Our latest technical breakdown covers:
→ How CVE-2026-72898 works
→ Affected Metabase versions
→ The exploitation flow and potential impact
→ What defenders should look for
→ Mitigation steps
→ How to safely validate the vulnerability in OffSec’s Offensive Cyber Range

Read the full analysis: https://t.co/jKO2ZRFroi
Explore the lab: https://t.co/w9oIUXPUbG

#CVE #Cybersecurity #Metabase #VulnerabilityResearch #OffSec

[Oren Yomtov @orenyomtov]

i just won $200k using this one prompt

[JS0N Haddix @Jhaddix]

I guess there is some discourse on watching YouTube videos if you are a practitioner? If we are including conference recordings, then I watch so many YouTube videos. If we are only talking about cybersecurity content creators, I still watch some too. You can usually tell by the title or the first few minutes of a video if it's deep enough for you. It's always been my personal philosophy that from any conference talk or video, you can probably learn one tip or trick from the author, either explicitly or by reading between the lines, and add it to your tool belt. 🫶

[JS0N Haddix @Jhaddix]

https://t.co/MUsOCCTjfS

[Mohankumar @Vimokumar]

that’s a wrap on @defcon .. until next time, Vegas

[Mohankumar @Vimokumar]

Finding bugs is getting cheap... fixing the right ones safely is the harder problem... speaking at @defcon @AppSec_Village

if you’re there, come say hi.

[Mohankumar @Vimokumar]

The best conversations start with one question: Why?

[Aaron Levie @levie]

Great piece and vision for AI from Zuckerberg

[Krishna Agrawal @Krishnasagrawal]

Google just dropped a 1-hour course on agentic engineering from scratch:

00:00 – How to build your first AI agent
08:24 – Build agent memory (short, persistent, long)
28:34 – Agentic loops, long-running AI agents
40:04 – How to build MCP (MCP vs API)
1:00:22 – Multi-agentic systems

This 1-hour watch will replace 10 paid agentic courses on the internet.

Bookmark this. Watch this weekend.

[Patrick OShaughnessy @patrick_oshag]

My conversation with Sam Altman (@sama), CEO of OpenAI.

We discuss:
- Kimi, distillation, and open source
- OpenAI's compute bets
- The Hugging Face incident
- What happens after AGI
- Raising kids in an age of abundant intelligence
- And much more

Enjoy!

TIMESTAMPS
0:00 Intro
4:10 The Race for Compute
14:24 A Sci-Fi Cyber Incident
16:14 The Promise and Risks of AGI
23:27 How AI Will Change Jobs
29:38 Sam’s Vision for a Personal AI
35:02 Robotics, ChatGPT, and What Comes Next
44:39 The Weight of Leading OpenAI
51:36 OpenAI’s Biggest Lessons

[a16z @a16z]

"That logo you’re chasing is costing you your market."

"Every AI founder believes they’re inventing the future. And they are! But your buyer doesn’t purchase the future; they purchase either proof or math."

"If they need proof, go win the logo that gives it to them. If they need math, get on a plane and show it to them before your competitor or the incumbent does."

"The founders who get this wrong won’t fail because they built the wrong product or picked the wrong strategy off a menu."

"They’ll fail because they never asked which game they were in, and in a market moving this fast, you only get to ask it once."

a16z's Joe Schmidt and Julian Marx on the two GTM playbooks for AI companies selling into the enterprise: https://t.co/IguoXFNZs1

[Mohankumar @Vimokumar]

Looking forward to it @BlackHatEvents