What happens when a company rushes into AI adoption without understanding what it actually costs? Ankit Mishra joins Jack Smith to share a firsthand incident involving an AI-powered sales lead generation platform, shared API access, and thousands of dollars in unexpected usage charges. What started as a promising business automation project became an expensive lesson in API budgets, cost monitoring, and the importance of understanding the technology before putting it into production.

We expand into the challenges of enterprise AI adoption, from AI fluency and employee training to human oversight, unreliable outputs, and the pressure to deploy new technology faster than organizations can adapt. Also explored are the economics behind AI services, the importance of keeping humans involved in critical decisions, and why slowing down might be the smartest move when everyone else is racing to automate everything.

Find out more about Ankit Mishra on https://ankitmishra.ca

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/episode-29-the-cost-of-rushing-ai/

00:00 Introduction
01:20 AI Horror Story
05:58 AI Fluency Concept
08:07 AI Levels Explained
12:26 Certification Ideas
19:10 Token Costs & Efficiency
23:25 Hardware Costs
29:22 Critical Thinking
32:10 Lessons Learned
35:22 Speed of AI
37:36 Conclusion

Call Center Down : If Only AI Could Diagnose Network Traffic Problems

What happens when hundreds of call center employees suddenly find their systems grinding to a halt, only for everything to return to normal at 5 pm? And why does the same problem return at 9 am the next morning?

Wolf O'Rourc joins Jack Smith to revisit a network incident where an overloaded Cisco backbone brought customer service operations to a near standstill. Finding the cause became an exercise in troubleshooting, escalation, and detective work defining the culprit to be quite an unexpected source.

Find out more about Wolf O'Rourc at https://www.wolforourc.com/

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/episode-28-call-center-down/

00:00 Introduction
01:09 Networking Issues
05:41 Communication Breakdown
10:37 Unauthorized Streaming
19:29 Resolution and Learnings
35:11 Future of IT

Two critical PHP vulnerabilities. Emergency patches. Government websites going offline. And IT teams around the world suddenly discovering just how much forgotten infrastructure they're responsible for. But the real horror story isn't PHP itself. It's the years of technical debt, unsupported software, missing documentation, and Shadow IT that turn a security update into a business crisis.

Because you can't patch a server you don't know you have.

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/jacks-rant-php-panic-and-technical-debt/

00:22 Introduction
00:39 PHP Security Flaw
02:00 RCE Dangers
02:35 Patch Release Chaos
05:01 Identifying Vulnerable Systems
06:54 Widespread Impact
09:04 AI in Discovery
09:15 GitLab Joins Chaos
09:32 Addressing Technical Debt
11:22 Possible Exploits
11:58 Emergency Wrap-up
12:48 The Real Horror
14:26 Conclusion

Companies complain that they can't find the right technical people. Technical people complain that companies have no idea who they're actually looking for. This time, Jack crosses to the other side of the table with Ashley Randall to find out why both might be right.

We look at how organizations turn operational problems into job descriptions, why they search for impossible unicorns, and what happens when the person they hire has the right skills for a job that doesn't actually exist as advertised. From HR screening and mismatched expectations to KPIs, management workshops, organizational silos, and the eternal question of whether HR is really there for you, we explore what happens when technical specialists and the business speak very different versions of the same language.

Sometimes the problem isn't the person you hired. It's the job you thought you were hiring them for.

To find out more about Ashley Randall and Aspiring Redirection, visit:
Ashley Randall: https://www.linkedin.com/in/ashley-randall7/
Aspiring Redirection: https://www.linkedin.com/company/aspiring-redirection/

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/episode-27-wrong-person-right-skills/

00:00 Introduction
00:43 Business Perspective
02:03 Hiring Challenges
04:48 Evaluating Talent
07:02 Structural Issues
08:52 Role Mismatches
13:24 Miscommunication Effects
17:15 Improving Processes
24:00 Culture and Metrics
33:43 HR and Employee Relations
39:23 Conclusion

Jack talks with Harvey Scolnick about his years working with air traffic control technology and the failures that came with it. From construction work accidentally cutting a critical cable and knocking an air traffic control facility offline, with systems unavailable and aircraft still in the sky, to the people on the ground having to fall back on procedures, neighboring facilities, and some very human problem-solving.

In addition, we look at the redundancy behind air traffic control, what recovery looks like when safety-critical infrastructure fails, and why even carefully designed systems can still be brought down by something remarkably mundane.

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/episode-26-cleared-for-disaster/

Find out more about Harvey Scolnick : https://www.scolnickatc.com/

00:00 Introduction
00:55 Harvey's Journey
02:13 Becoming ATC
05:34 Airways and Radar
07:03 Separation Standards
11:21 Busy JFK
14:33 Iconic Aircrafts
21:12 Technology Upgrades
29:27 Preventative Measures
38:10 AI and Future
41:27 Weather Challenges
43:57 Conclusion

Someone asked us why we weren't on MeshCore yet. First we had to establish that it wasn't a new kind of thrash metal. Then we discovered an off-grid messaging network built from little radios, antennas, repeaters, and licence-free spectrum.

So we started digging. How does it route messages ? How private is it ? What happens when repeaters disappear ? And haven't we built something suspiciously similar before ? MeshCore turns out to be considerably more interesting than just another messaging app.

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/jacks-rants-meshcore/

MeshCore documentation used :

- FAQ : https://docs.meshcore.io/faq/
- Packet Format : https://docs.meshcore.io/packet_format/
- Payloads : https://docs.meshcore.io/payloads/
- Companion Protocol: https://docs.meshcore.io/companion_protocol/
- CLI Commands : https://docs.meshcore.io/cli_commands/

MeshCore Map : https://meshcore.co.uk/map.html

00:00 Introduction
00:46 Discovering Mesh Core
02:09 Radio and Repeaters
05:19 Free Radio Insights
10:35 Encryption and Privacy
14:26 Old Meets New
15:19 Infrastructure Challenges
18:16 Conclusion

IT may run on the same protocols everywhere, but the people operating it certainly don't. Put teams from different countries, companies, and working cultures onto the same project and apparently simple questions about responsibility, communication, deadlines, and escalation can suddenly have very different answers.

This episode looks at what happens when those differences become part of the technical problem. From assumptions that nobody realised they were making to communication styles that work perfectly well locally but collide spectacularly across borders, we discuss how international IT teams get themselves into trouble — and what years of working across cultures can teach you about avoiding it.

And no, the solution isn't another mandatory cultural-awareness PowerPoint.

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/episode-25-cultural-differences-in-it/

00:00 Introduction
00:41 Cultural Intricacies
03:12 Sales Challenges
06:02 Engineering Insights
10:32 Crisis Management
14:10 Data Center Differences
16:00 Luxembourg's Unique Edge
17:52 European Culture Wrap-up

Texas, 1978. Robert was working at a cotton co-op running what may have been one of the world's earliest online auction systems. Farmers received punch cards representing their bales of cotton - occasionally complete with enough stray cotton fibres to jam the card readers - while buyers placed bids remotely using leased IBM 3270 terminals and acoustic coupler modems.

After an upgrade, the mainframe refused to come back properly. An IPL produced a corrupted device address before dropping the machine into a hard wait. Single-cycling the processor indicated that an UNPK (Unpack) instruction was producing the wrong output from perfectly good input. The instruction itself had effectively broken.

And yes, this episode contains a lot of old IBM hardware and mainframe terminology. We know. That's part of the fun.

Want even more info ? Read our show notes and related blog post related to this episode : https://blog.ithorrorstories.eu/episode-24-when-unpk-broke/

00:00 Introduction
00:43 Robert's Background
02:14 IBM DOS
03:27 Cotton Auction System
05:37 Online Auction
07:14 SNA Protocol
09:05 Hardware Issue
13:20 Debugging Unpack
17:13 IBM Fix
21:23 Conclusion