
Sign up to save your podcasts
Or


AI best practices have evolved beyond writing better prompts. In Episode 38 of the In-Secure-Ity Podcast, Drew Thompson explores context engineering, durable knowledge management, and practical approaches to working with AI agents.
Drawing on firsthand experience with AI-assisted development and a custom-built podcast manager, Drew discusses using project instructions, decision logs, and reusable documentation to improve AI output and preserve context between sessions.
He also explores how these practices apply beyond software development, including AI workspaces, reusable skills, enterprise knowledge management, and workflows that remain portable across models and platforms.
The episode closes with two practical recommendations: keep important knowledge outside an AI platform's conversational memory, and document project decisions so agents can build on established requirements.
Stay secure. Stay connected.
We've been predicting the death of passwords for decades. Are passkeys finally going to make it happen?
In Episode 37 of the In-Secure-Ity Podcast, I explore what multi-factor authentication actually means, how public and private keys work, and why passkeys could finally reduce our dependence on traditional passwords.
We cover the three categories of authentication factors, the principles behind HTTPS, how passkeys authenticate users, and why they're resistant to phishing.
But using passkeys isn't always seamless. I share my frustrations with Microsoft Edge and its built-in password manager, explain the differences between device-bound and synchronized passkeys, and discuss what happens when you lose access to the device holding your credentials.
We also explore what passwordless authentication actually means, why passwords may stick around for recovery, and the practical challenges of using credentials across different devices and platforms.
My take? Passkeys are worth adopting, even if the experience still needs work.
And there's another question worth exploring: How will we handle sharing passkeys with family members or accessing accounts after someone dies?
**Technical clarification:** I briefly misspeak about public and private keys during the episode. The public key is shared with the website; the private key remains protected.
For more episodes and show information:
https://www.beeyondlearning.io/in-secure-ity/
In-Secure-Ity Podcast
Stay Secure. Stay Connected.
AI has made custom software dramatically cheaper to build, but that doesn’t automatically mean SaaS is dead.
In this episode of the In-Secure-Ity Podcast, I dig into how AI is changing the build-versus-buy equation, why bloated SaaS platforms are often the result of years of reasonable product decisions, and why cheap code still doesn’t eliminate deployment, maintenance, security, ownership, or long-term operating costs.
I also get into the part I find most interesting: what happens when AI becomes the interface to SaaS. If tools like Salesforce can be accessed headlessly through APIs and MCP, the interface may matter less while the underlying platform becomes even more valuable. That could also make it easier for smaller, purpose-built tools to compete with individual pieces of larger platforms instead of replacing them outright.
The future probably isn’t build or buy.
It’s figuring out what’s worth building, what’s worth renting, and which risks you want somebody else to own.
For more episodes, articles, and updates from the In-Secure-Ity Podcast:
https://www.beeyondlearning.io/in-secure-ity/
In-Secure-Ity Podcast
Stay Secure, Stay Connected.
Stop calling people the “human firewall.”
Firewalls operate on rules. People operate on knowledge, judgment, and habits. And turning knowledge into capability takes more than an annual security-awareness course or a briefing telling someone what they should do.
In this episode of the In-Secure-Ity Podcast, I talk about why information transfer is not the same as capability development, why practice matters, and why even capable people will sometimes make imperfect decisions when urgency, authority, consequence, and the pressure of real work enter the picture.
Training matters. Practice matters. But neither turns people into deterministic security controls. Organizations still need layered technical and procedural defenses that reduce the likelihood and blast radius of inevitable human mistakes.
If your security model depends on people consistently remembering and perfectly applying information you gave them in a briefing, you don’t have a human firewall. You have a design flaw.
In-Secure-Ity Podcast
Stay Secure, Stay Connected.
Vibe coding is here. The question is no longer whether developers should use AI. The question is how security has to change when software development starts happening at machine speed.
In this episode of the In-Secure-Ity Podcast, I look at vibe coding as the latest compression of the software development lifecycle.
We have seen this kind of compression before. Agile shortened development and feedback loops. CI/CD compressed integration, testing, and deployment. DevSecOps and intelligent orchestration pushed security toward continuous, context-aware controls throughout the lifecycle.
AI-assisted development compresses things again, especially implementation.
That does not make the SDLC disappear. It means security has to operate at the new speed.
I also talk about my own experience using AI-assisted development to build the tooling behind this podcast and an Android app for my daughter in roughly two hours of development time. This is not a theoretical shift. The development model is already here.
In the episode, I cover:
- What vibe coding changes about the SDLC
- New attack surfaces introduced by AI coding agents
- Prompt injection, poisoned context, excessive permissions, and unsafe tool use
- How threat modeling changes when AI is helping shape the architecture
- Why AI’s tendency to expand scope can work against minimal, secure design
- How security requirements can be built into prompts and architecture up front
- Verification strategies when line-by-line human review no longer scales
- Where SAST, DAST, SCA, CI/CD policy checks, and runtime validation fit
- Why deterministic guardrails and environment-level permissions matter
The central idea is simple:
Every time software development gets compressed, security has to learn how to operate at the new speed.
Vibe coding is the most aggressive compression we have seen yet.
For more episodes and the official home of the In-Secure-Ity Podcast, visit:
https://www.beeyondlearning.io/in-secure-ity/
In-Secure-Ity Podcast
Stay Secure, Stay Connected.
IN-SECURE-ITY PODCAST is back after a ~15-month hiatus, and Drew explains exactly why the show paused—and why it’s returning with a new, sustainable Q&A format.
In this episode, Drew breaks down what didn’t work in the old solo format (the “talking at a wall” / “talking into a void” feeling and the dead air it created), and why questions change everything. The goal isn’t a tightly scripted show or an artificial interview vibe—Q&A is used as a lightweight outline that lets Drew think in real time, pivot when needed, and keep the conversation moving naturally.
You’ll also hear how questions will be sourced going forward: listener submissions, live Twitch chat, and (when needed) AI-assisted prompts using tools like ChatGPT. Drew is direct about it—AI-generated questions aren’t meant to fake audience participation; they’re placeholders that provide structure until more listener questions arrive.
Operationally, the episode explains the real bottleneck that caused the hiatus: recording was “dirt simple” (mic on the desk, fire up OBS, hit record), but publishing tasks like newsletters, advertising, and other overhead made the old workflow hard to sustain. The new approach is live-first: record on Twitch and ship it as-is—no edits and no pickup outro.
Cadence and expectations are intentionally reset. IN-SECURE-ITY will publish every two weeks, with topics announced ahead of time so listeners can submit questions, and with flexible episode length—30 minutes is a guideline, not a requirement.
Question boundaries are also clarified: keep submissions roughly on the announced topic, avoid company-specific/internal material, and feel free to ask adjacent questions—good “off-topic” ideas may seed future episodes.
Key conclusions:
- The hiatus wasn’t about losing interest; it was about format fit and operational burden.
- Q&A provides structure without scripting, making episodes more conversational and easier to sustain.
- AI-assisted questions are acceptable prompts, with real listener/Twitch questions prioritized.
- Success is redefined: “Success criteria is a mouse click every two weeks.”
- Consistent publishing beats perfect production—Episode 33 ships live as-is.
Takeaways:
- Want your question answered? Submit it via Twitch chat or ahead of time, keep it broadly useful, and avoid internal/company-specific details.
- Expect live thinking: Drew may explore ideas, pivot, miss a point, and return later—by design.
- The show is optimized for sustainability and enjoyment first.
Stay Secure, Stay Connected.
In this episode the hosts of the In-Secure-Ity Podcast will provide some background about themselves.
In Episode 7 of the "In-Secure-Ity Podcast," we dive into the impact of high-profile breaches like Snowflake and Target. Learn how these incidents drive changes in company policies and government regulations, and get practical tips on maintaining work-life balance. Join us for an insightful discussion that blends technical expertise with personal growth strategies.