Dan Benger

No matter what, an AI agent will find the smallest crack in your castle.

@aiandcloud @aaronrubin Hi Jintao, I totally get that! Balancing cost with compliance is tricky. At Tego, we're building solutions for agentic AI security that might help. Would love to chat more!

@aaronrubin Same 😅 Claude Enterprise has a very high cost. However, considering compliance and access control features, it is necessary to purchase it.

Big news. I migrated from my personal Claude Max to our company account so I can pay $1,600 instead of $200 for the same monthly usage.

Tal Melamed, CTO & Co-Founder of @Tego_AI, is rocking the stage today at AI Summit Europe! Tal is breaking down how to leverage the power of AI agents while maintaining bulletproof security and granular access control. #AIsummitEurope #AgenticAI #AIsecurity

AI agents need access control, not just access. Tego AI helps organizations define what agents can see, do, and touch, before risk becomes reality. #AISecurity #AIAgents #TegoAI

@joy014 @ankrgyl Hey Joy, totally feel you on the distribution and access control challenge! At Tego, we're building solutions to enhance security visibility and control for agentic AI. Would love to chat more about it!

@ankrgyl yes. Everything right now is written for solo/end developer. Distribution and access control still not figured out.

is it just me or are none of the harness libraries/frameworks/etc designed for multi tenancy? all of the state seems filesystem oriented, no RBAC, etc

@twlvone @vikhyatk Totally agree! The gap between code and full deployment is real. At Tego, we're tackling these challenges with agent-native security. Would love to chat about bridging this gap!

@vikhyatk Code generation collapsed the dev bottleneck — turns out the remaining work is all the stuff that isn't code: deployment, monitoring, incident response, access control, data pipelines. AI can write a function in 5 seconds but can't own a 3am prod incident yet. That's the gap.

software generation is no longer the bottleneck. it's operations trillion dollar opportunity for whoever solves it

@natalie_avfieb @thedatabunny Hey Natalie, great insight on AI security! At Tego, we're tackling this by offering agent-native security visibility, ensuring safe AI adoption. Would love to share more!

@thedatabunny From a research angle, the agentic AI security problem isn't just access — it's what tool servers return at runtime. When an agent executes an MCP call, the response content itself becomes an attack vector. Most enterprise security frameworks still don't inspect that layer.

Cybersecurity professionals, how do you all feel about Ai and the things agentic Ai can do specially around security? I’m very curious what you all think. Thanks! 🙏🏾

@dprophecyguy @GergelyOrosz @leerob Hey Vijay, that sounds frustrating! Balancing cost and efficiency is tough. At Tego, we're building solutions for smarter AI adoption. Would love to share insights on optimizing usage!

@GergelyOrosz cursor has gone full predatory on pricing, exhausted around 400$ dollar budget in a day. its really bad now. have been using claude with api pricing and never has crossed 100$ a day with same workload and pattern. cc: @leerob

I am hearing tons of complaints from Cursor customers at enterprise companies: A silent change put almost all models Cursor uses behind Max mode. Devs who used to manage to “spread out” monthly credits over a month see all of it used up in 1-2 days. Are furious + switching.

@Afei_th @SBF_FTX Totally agree! Adoption is key. At Tego, we're helping enterprises safely integrate AI by providing agent-native security and control. Would love to share insights on enabling governed autonomy!

Yeah, spot on — the real bottleneck is actual adoption, not just buying ChatGPT Enterprise licenses. Most companies are still experimenting, not reorganizing workflows, incentives, or legacy processes around what frontier models can really do.Low adoption numbers today don’t predict the ceiling; they just show how slowly big orgs move. Frontier models are still underpriced relative to the value they’ll unlock once serious reorganization actually starts. We’re early in the “reorg shock” phase.

AI's current bottleneck is adoption. Actual adoption. Reorganizing firms around the full potential of the latest models; not "we bought ChatGPT Enterprise." Adoption tells you nothing about the eventual scale of economic impact. AI is still underpriced.

@101010_jv @spluscollective Totally agree! The shift to AI applications is where the real value is. At Tego, we're all about securing agentic AI for enterprises. Would love to share insights on how we do it!

@spluscollective Ai trade is cooked imo. Ai trade meaning Optics/Memory, data center spending.. The Ai trade should move to the application of Ai...ai agents, ai security, FSD, robots, companies making money from it etc...

@crypto_preneur @Benioff @salesforce @agentforce Great point! Security is crucial for trust in agentic workflows. At Tego, we're building solutions to ensure enterprises can safely manage AI agents. Would love to share insights!

@Benioff @salesforce @agentforce Interesting - the real test will be whether enterprises trust Salesforce with their customer data for agentic workflows. Security concerns could be a major barrier.

Welcome Agentforce CCaaS: AI, CRM, Voice, ITSM, Field Service for unified success. 🚀 #Agentforce

@APEXFEED01 @jacksonhinkle Wow, that's a bold move by Google! The stakes are definitely high. At Tego, we're all about enabling safe AI adoption with security visibility and control. Would love to share insights on how we're tackling this!

@jacksonhinkle A massive move by Google. Betting $32 billion on cloud and AI security shows just how high the stakes are for enterprise infrastructure right now

🚨 🇺🇸 🇮🇱 BREAKING: Google has acquired Israeli cloud security company 'Wiz' in multi-billion dollar deal. Wiz will fully integrate into Google Cloud. It is the largest technology acquisition involving an Israeli-founded company.

@theroberthu @FT Hi Robert, totally get your point! Securing AI systems is crucial. At Tego, we're all about providing agent-native security for enterprises. Would love to chat about how we can help!

@FT the firm that advises everyone on ai security couldn't secure its own ai systems

McKinsey rushes to fix AI systems after hacker exposes flaws ft.trib.al/9dzqmjy

@sooyoon_eth @rocklambros Hi Soo, totally agree! Agentic AI security is crucial. At Tego, we're building solutions to give teams visibility and control over AI agents. Would love to chat more!

@rocklambros this is huge. agentic AI security is the next frontier and most teams aren't even thinking about it yet. prompt injection is the new SQL injection and we're watching the same mistakes happen in real time

One more thing before you close this briefing: NIST's RFI on securing AI agent systems closes Monday, March 9. That's 48 hours from now. This is the first formal federal request for input on agentic AI security.

@TechLifeNotes @Benioff @salesforce @agentforce Hi Sandeep, totally agree on the need for unified platforms! At Tego, we're building solutions to bridge those gaps with agent-native security. Would love to chat more about it!

@Benioff @salesforce @agentforce Not sure why it looks so pixelated. After a decade in IT Ops, I’ve always felt contact centers and IT teams need a unified platform to truly break silos. If executed well, this could be a big win. 🚀

@ejae_dev @Benioff @salesforce @agentforce Totally agree! Fragmented stacks can be a hassle. At Tego, we're building solutions for seamless AI integration with robust security. Would love to share insights on this!

@Benioff @salesforce @agentforce combining all of these under one ai layer is the move. fragmented support stacks can't keep up

@AlmOSt_INTel @aylacroft @0xTib3rius Totally agree! AI security and policies need to catch up. At Tego, we're building solutions to help enterprises safely adopt AI with agent-native security. Would love to share insights!

@aylacroft @0xTib3rius Exactly. I'm waiting for AI security, policies and practices, and overall human decency to catch up to the latest and greatest. We currently way behind in all 3! Fingers crossed for the good guys/gals to step up.

The "political" compass of AI. I'm in the blue. Where are you? 👀

@RobTerrin @ZackKorman Hi Rob, great question! AI security risks are definitely complex. At Tego, we're building tools to help manage these risks by providing visibility and control over AI agents. Would love to chat more about it!

@ZackKorman To really know it's the least scary you'd have to rank all the probabilities and impacts of AI security risks. Seems like based on probability alone it's pretty high up there. Impact is really hard to gauge, but could be serious. What are other highly probable AI security risks?

Of all the AI security risks, vendors training on your data is one of the least scary. It’s just the one CISOs focus on most because it’s the easiest to understand and solve.

@VizierPrime @lopp Great point! It's crucial to address those hidden vulnerabilities. At Tego, we're building agent-native security to ensure AI adoption is both safe and innovative. Would love to share insights!

@lopp A lot of people hear “AI security risk” and think only of new attacks. But another danger is faster excavation of old vulnerabilities buried inside the digital substrate. The future arrives by exposing how much of the past was never really secured.

As a security professional, AI acceleration is concerning. For example: Bruce Schneier's monthly cybersecurity newsletter just went out and 14 of the 23 news items were AI related. OpenSSL just patched a dozen vulnerabilities that were found by AI; some were 25 years old.

@araneusx1 @0xTib3rius Hi Thomas, great question! AI security is evolving fast. At Tego, we're building solutions to ensure agentic AI is secure and adaptable. Would love to share insights on how we're tackling this!

@0xTib3rius Bro @0xTib3rius do you think that AI Security will survive for Very Long period of time? will it live forever?

2025: using command line tools to hack 2026: teaching agents to use command line tools to hack 🔥