Talsco Weekly: What happens when the last IBM i engineer retires?

Welcome to another edition of Talsco Weekly

IBM i Brief: 🏆 As AI scales, governance becomes the new battleground. IBM just won round one.
AI: 🛡️ One answer. Two words: protect your data. 🦾 AI is going physical, not just digital. 🪙 Your AI budget runs on a unit nobody can actually define.
Career: 🔄 You are not just being interviewed. You are interviewing them.
Development: 🟦 New VS Code extension tracks source history. 🔧 Another sign RDi’s days as the default IBM i IDE are numbered. 🛠️ No vendor tools needed for real-time CPU alerts.
Learning: 🪙 What is an AI token, and why should your project budget care?
Modernization: ⏳ The Cornish language died with one woman in 1777. Government IBM i shops are watching their own version play out. 🔓 Don’t confuse legacy with broken.
Security: 🕵️ Your team is already pasting company data into ChatGPT. You just can’t see it.
Vendor: 📈 More proof vendors are doubling down on IBM i and Power, not exiting it.

IBM i Brief

🏆 As AI scales, governance becomes the new battleground. IBM just won round one

Gartner named IBM a Leader in its first-ever Magic Quadrant for AI Governance Platforms. The category exists because enterprises moving AI from pilot to production need visibility into where AI is used, how risk is changing, and whether controls are actually in place, not just documentation after the fact. IBM is betting watsonx.governance becomes the standard layer for that oversight.

AI

🛡️ One answer. Two words: protect your data

When asked what companies should do today if they’re considering an AI strategy, a neuroscientist and AI strategist on the COMMON Europe keynote stage cut through the noise.

Before the models, before the tools, before the roadmap. Start with your data.

🦾 AI is going physical, not just digital

Genesis AI unveiled Eno, its first general-purpose robot, pairing proprietary hardware with its GENE robotics foundation model. The robot uses a mobile base, an adjustable tower frame, and dexterous human-sized hands to handle multi-step tasks across manufacturing, logistics, labs, healthcare, and hospitality. Production and first customer deployments are targeted by the end of 2026.

Our Take: This isn’t about the IBM i market. It’s about the broader enterprise market. While many are still figuring out how to use AI digitally, the market is going physical.

🪙 Your AI budget runs on a unit nobody can actually define

Time to dig into the token problem facing IBM i shops using AI for development work.

Tokens aren’t arbitrary, but they’re vendor-defined, proprietary, and opaque, which means project costs become unpredictable in a way old-school hourly billing never was. IBM’s Bobcoin model for Bob layers on more complexity rather than less.

Read more on what tokens actually are and why this matters for project budgeting, down in the Learning section below.

Career

🔄 You are not just being interviewed. You are interviewing them.

Whether it is your first IBM i interview or your first in years, this is the move most candidates miss.

First-timers spend so much energy preparing answers they never think about their own questions. Veterans slip into evaluation mode, answer what is asked, and wait. Both leave without the information they needed.

Asking smart questions is not just polite. It signals maturity. It shows you have thought seriously about the role. And it gives you real information about whether this is actually the right place for you.

A few that work at any level:

“What problems would I be expected to solve in the first 90 days?”
“How is success measured in this role?”
“Where does AI help your team today, and where does human judgment still matter?”

The takeaway. Walk in with two or three real questions. Not about hours or benefits. About the work, the team, and what winning looks like. Book a 20-minute call.

Development

🟦 New VS Code extension tracks source history

Ricky Thompson built an AI-assisted VS Code extension that brings the local folder Timeline view to IBM i source members. Open a member and it snapshots the source, then captures another snapshot on every subsequent save (20 by default, adjustable in settings), so developers can compare versions without leaving the editor. It’s live in the extension marketplace now.

Our Take: Small tools like this matter. RPG developers have asked for better version visibility inside Code for i for years, and this fills a real gap without requiring a separate source control setup.

Check out the extension on GitHub

🔧 Another sign RDi’s days as the default IBM i IDE are numbered

PTC is rolling out full VS Code support in Implementer 12.7, putting the company’s change management tool on the platform that recently passed RDi in usage among IBM i developers. PTC’s software development manager said the team is focused on the next generation of IBM i developers, who simply aren’t going to choose RDi or green screen tools. The release also adds a new RESTful API designed to support integrations, including AI co-pilots.

🛠️ No vendor tools needed for real-time CPU alerts

Here is a native RPG program that monitors job CPU usage in real time, filling a gap left by Navigator for i and Performance Data Investigator, which only show historical data. Using Db2 for i’s ACTIVE_JOB_INFO table function in a continuous polling loop, the program flags jobs exceeding a defined CPU threshold and alerts the system operator instantly.

Learning

🪙 What is an AI token, and why should your project budget care?

IBM i shops have always quoted projects in hours. Internal or outsourced, the unit was stable and verifiable.

AI breaks that math. Most AI tools, including ChatGPT and IBM Bob, bill by the token instead of the hour. A token is roughly three-quarters of a word, but technically it’s the output of a tokenizer applied to your input text, mapped against a provider’s vocabulary and model. Every part of that process is proprietary to the AI company you’re using.

That’s the problem: the company selling you tokens also defines what a token is. There’s no independent standard, the way there’s an agreed-upon gallon or hour. Dan Darnell and Eric Whitcomb, who write extensively on IBM i and AI, put it plainly: “the issuer controls definition, conversion, accounting, and audibility.”

Why this changes project budgeting

Hours times rate used to mean a knowable cost upfront. With token billing, the rate can shift mid-project. Costs double, then double again, with no clear link to better output. IBM’s Bobcoin system for Bob adds another accounting layer on top, making costs even harder to predict.

This isn’t just an IBM i problem

Investing.com reports GitHub Copilot moved to usage-based token billing on June 1, 2026, as agentic AI workflows, where AI handles entire tasks rather than suggestions, burn tokens fast: “agentic workflows consume far more tokens per session than single-turn completions.”

It’s serious enough that Troutman Pepper Locke warns public companies may need to disclose material AI token spending in SEC filings, since “the shift to usage-based AI pricing leads to simultaneously larger and more unpredictable material expenses.”

The takeaway

Before scoping AI-assisted work, internal or outsourced, ask how token usage is measured, billed, and capped. It doesn’t behave like hourly billing, and the opacity is by design. This is no longer a developer complaint. It’s a board-level issue.

Modernization

⏳ The Cornish language died with one woman in 1777. Government IBM i shops are watching their own version play out

This article compares the retirement of late-career COBOL, assembler, and RPG specialists in government to the death of the last native Cornish speaker. State and local agencies running tax, licensing, justice, and benefits systems on IBM i and IBM Z face a shrinking pool of people who actually understand the code keeping those systems running.

The piece lays out a familiar modernization playbook: training pipelines, hybrid cloud, DevOps, API integration, and a structured roadmap rather than a rip-and-replace panic move.

Our take: We’ve been saying this for a while and this only reinforces it:

there’s no slowdown happening in the IBM i space.

If anything, it’s the opposite. The organizations actually running on this platform, where it’s the backbone of daily operations, are moving forward with real momentum. APIs, DevOps, agile, and now AI aren’t threats to IBM i. They’ve become the catalyst that’s rewriting the narrative around it.

🔓 Don’t confuse legacy with broken

A Forbes Technology Council piece pushes back on the “rip and replace” instinct, arguing IBM i’s legacy label confuses interface limitations with platform failure.

Our take: When people call IBM i legacy, they’re usually reacting to an interface that hasn’t kept pace with modern UX expectations, not a flaw in the platform itself. The system remains one of the most stable, high-volume transaction processors ever built, with a track record for backward compatibility that few platforms can match.

Next steps: Before defaulting to a full exit, the smarter move is treating IBM i as a participant in AI initiatives rather than an obstacle to them. That starts with auditing embedded business logic, evaluating AI discovery tools and API integration layers, and building the extension layer as the actual delivery mechanism for AI capability.

Security

🕵️ Your team is already pasting company data into ChatGPT. You just can’t see it.

A new guide breaks down shadow AI: employees using unapproved AI tools on accounts IT doesn’t control. 81% of employees admit to using AI tools their employer hasn’t approved. Blocking these tools outright tends to backfire, pushing usage to personal devices where there’s zero visibility. The better path: find out what’s actually being used, set rules by data sensitivity, and give people a sanctioned alternative before they go find their own.

Our take: We’ve been down this road before. Shadow IT was unapproved software showing up on company networks. Then it was shadow mobile, personal phones and apps quietly accessing business data. Now it’s shadow AI, and the pattern is identical: employees find a faster way to get work done, and security finds out after the fact.

Worth being precise here: traditional IBM i pen testing, exit points, object authorities, exposed APIs, isn’t designed to catch this. Shadow AI isn’t someone breaking into your system. It’s an authorized employee pasting sensitive data into a tool that lives entirely outside your perimeter. No exploit required.

That said, if any AI tools are integrated into IBM i through APIs, service accounts, or browser extensions, those integration points absolutely belong in your next pen test scope. The bigger gap is governance: do you actually know what’s touching your data right now?

That’s the question worth answering before someone else answers it for you.

Vendors

📈 More proof vendors are doubling down on IBM i and Power, not exiting it

FalconStor’s CEO detailed how a 2021 bet on IBM Power data protection turned into 73% year-over-year revenue growth, with recurring revenue now making up 87% of total billings. The company protects over 900 IBM Power customers and more than 3,700 petabytes of IBM i data. Whether vendors are new to this market or have been here for decades, the signal is the same: this ecosystem is worth committing to.

Join

Contact us

If you are an RPG programmer looking to explore opportunities or a client who is looking for a talented IBM i professional, please contact us. We look forward to assisting you.

Do you know of someone who could benefit from Talsco Weekly? If so, please use the social media buttons to spread the word. Thank you!