
Sign up to save your podcasts
Or


Jason Lau (CISO @ Crypto.com) watched attackers stop going after code and start going after people. When his team used AI to block scam messages aimed at support staff, attackers switched languages, then sent images of text, then swapped in unusual characters to slip past the models.
On this episode of Early Adopters: CISO, Jason shares what nine years as Crypto.com's founding CISO taught him about defending at crypto speed.
We cover:
- Why crypto sees new attacks 15 months before others
- Putting IT operations inside the security team
- Insider threats that start with outsourced customer support
- How scammers rework messages to beat AI detection
- Why organizations give AI agents too much authority
- Hiring for curiosity and an incident response mindset
When Jonathan Aluveaux (CIO @ Legora) started building security at Ramp, one of his early calls was an insider risk tool, which most people don't do early. He bet that observability would let the company take on risks others were uncomfortable with.
Now he owns AI enablement at Legora, where he sees agents as a massive insider risk. On this episode of Early Adopters, Jonathan shares how he balances speed and risk as access shifts from humans to agents.
We cover:
- Why he bought an insider risk tool early at Ramp
- The AI harness that rates itself on his workload
- Turning team experiments into production-grade AI workflows
- Agent access, user attribution and just-in-time grants
- How he measures a security vendor's velocity
- Why founders should skip the story in a first call
Myke Lyons (Chief Information Security Officer @ Cribl) bought his first logging system for $1.5 million, then waited for the license key. A six-foot rack of hardware showed up in the mailroom instead.
Now he's back in telemetry at Cribl, arguing that AI SOC should move tier 1 analysts up into tier 2 work. On this episode of Early Adopters, Myke shares how he builds security teams and where he sees the SIEM heading.
We cover:
- Why an operator CISO needs a number two and three
- Where LLMs are changing appsec the most
- How AI SOC moves tier 1 analysts into tier 2
- Why detection engineering took a month per rule
- Four tiers of security data, SIEM to dev null
- Why every team needs an open-weight model alternative
When OpenAI published details of the Hugging Face attack, Sean Cassidy (Chief Information Security Officer @ Plaid) told fellow CISOs that worrying about it happening to them was only half the job. They also had to make sure their own agents weren't hacking other companies without anyone knowing.
On this episode of Early Adopters, Sean shares how that shift shapes his thinking on agent access, SIEMs and buy vs. build at Plaid.
We cover:
- Why agent access and identity is still unsolved
- Why agents make a central SIEM less important
- Making risk acceptance an informed, conscious decision
- How Plaid decides what security to buy vs. build
- What founders get wrong when pitching CISOs
- The LostPass attack and the browser viewport problem
Datadog's security team dropped its security questionnaire vendor and built the one piece of functionality it needed in two to three months. Zack Allen, Senior Director, Trust and Detection Engineering at Datadog, says the call came down to expertise, data and the harness.
On this episode of Early Adopters: CISO, Zack shares how his team decides what to build and what to buy, and what founders need to show before his team buys.
We cover:
- How Datadog's security team decides what to build and buy
- Expertise, data and harness as build vs. buy tests
- Why saying no to AI is how security loses
- What founders need to show a skeptical security team
- Moving past hand-written detection rules
- Security as portfolio management and detection health
Andrew Becherer, CISO at Socket, joins Early Adopters: CISO to talk about what security teams should build, what they should buy, and what founders get wrong when selling to them.
The conversation covers Andrew's path from a sociology degree to iSEC Partners, joining Datadog in 2015 as its only security employee, and how the 2016 breach reshaped Datadog's security program.
Andrew also breaks down why a tool that is quick to build still has to be maintained for years, how he evaluates early-stage vendors, the softball question he asks founders, and why securing agents is the threat CISOs aren't talking about enough.
We cover:
Jacob DePriest, CISO/CIO at 1Password, joins Early Adopters: CISO to talk about how security leaders decide what to adopt in the AI era.
The conversation covers Jacob’s path from the NSA to GitHub to 1Password, why not shipping is its own form of risk, and how 1Password uses a velocity-weighted risk model to decide when AI tools can move fast and when they need to wait.
Jacob also breaks down how his team pilots new AI tools while due diligence catches up, how AI is changing build-vs-buy decisions, why security teams cannot keep scaling linearly, and what founders need to prove if they want to become a design partner.
We cover: