How to stay employable through AI era-shifts: the Foxpro lesson
FoxPro engineers got let go in 2003. So did Flash developers in 2014. Dreamweaver shops in 2016. Tools die on a schedule. Skills transfer. Here are the four that survive AI.
Summarize this article with:
- Eight tool extinctions in 30 years, with dates. The pattern is identical every time: the new tool eats the old tool, the practitioners who learn the new tool keep working, the practitioners who refuse get let go.
- What the 1990s Visual Basic shops did when they had to learn Macromedia Flash. What the 2010s Flash shops did when they had to learn React. What the 2020s React shops are doing now they have to learn AI.
- The four skills that transfer across every tool era: written communication, systems thinking, client trust, judgment.
- The three skills that don't transfer: tool-specific procedural knowledge, technique-specific manual labor, gatekeeping the obscurity of your tool.
- The practical move for any era shift: 30 minutes a day, 90 days, on the new tool. Same lesson the FoxPro people learned in 2003. The math hasn't changed.
Every AI-doom post is the 2026 version of the 2003 FoxPro panic. The math is the same. The fear is the same. The right move is the same. This piece is the receipt: every prior era-shift, what extinguished, what survived, and how to use that pattern to ride this one out instead of being run over by it.
A friend of mine spent 12 years writing FoxPro applications for medium-sized businesses. He was good at it. He had 14 long-term clients. In 2007, Microsoft announced Visual FoxPro 9.0 would be its final commercial release (Microsoft VFP lifecycle). Within six months all 14 of his clients were asking when he was migrating them to Visual Basic .NET.
He had two options. Learn .NET in 6 months and migrate the clients. Or refuse and watch the clients hire someone who would.
He learned .NET. The migration took 18 months instead of 6. He kept 11 of the 14 clients. He is still consulting in 2026, now on Postgres + React stacks. The FoxPro extinction did not end his career. It forced one painful upgrade.
The AI doom posts in 2026 are the FoxPro panic in 2003. Same shape. Same fear. Same right answer.
Eight tool extinctions in 30 years
With dates. So we can see the pattern:
- 1995. Lotus 1-2-3 loses share to Microsoft Excel. Years of Lotus-specific consultants either learned Excel or stopped consulting.
- 2007. Microsoft announces Visual FoxPro EOL. VFP 9 is the last commercial release. Mainstream support ended January 2010. Shops migrated to .NET through 2015.
- 2007. Apple ships the iPhone without Flash. Jobs publishes the famous "Thoughts on Flash" letter in April 2010 explaining why.
- 2017. Adobe announces Flash end-of-life with a December 31, 2020 terminal date. HTML5 + CSS3 took over.
- 2020. Adobe puts Dreamweaver in maintenance mode: still sold, no new features. Hand-coded React + modern build tools replaced the WYSIWYG era.
- 2018-2026. jQuery share declines slowly (peak ~78% in 2021 → ~73% in 2025 per W3Techs). React / Vue / Svelte dominate new builds; jQuery persists in legacy + WordPress.
- 2022. Photoshop-only retouchers start competing with Midjourney + generative fill. The fastest retouchers learned the new tools. The slowest got priced out.
- 2024+. Manual workflows compete with AI-assisted workflows across writing, design, code, data analysis, marketing. We are mid-extinction now.
Each extinction took 4 to 8 years from "new tool announced" to "old tool unhireable." In every case, the practitioners who learned the new tool kept working. The ones who refused did not.
The four skills that always transfer
Across every tool era, four skills move with the practitioner. Worth investing in these because they survive whatever comes next.
1. Written communication
The Lotus consultant who could write a clear status email beat the FoxPro consultant who couldn't. The React developer who can write a tight README beats the React developer who can't. The marketer who can write a clear brief beats the marketer who can't.
AI threatens nothing here. AI threatens the people whose only skill was clean typing. People who can write well work alongside AI to write more, faster.
2. Systems thinking
The ability to look at a problem and see the system around it. Where does the data come from? Where does it go? What breaks if we change this? What second-order effects show up in 3 weeks?
AI is mediocre at this in 2026 and will probably remain mediocre. Systems thinking requires holding 20 things in mind at once and knowing which 3 to focus on. AI gives you a confident generic systems answer. Real systems thinking is specific.
3. Client trust
The plumber's customers don't hire him because his pipe-banging is faster than YouTube tutorials. They hire him because they trust him to show up, do the job, and not break something. Client trust transfers across every tool era because it was never about the tool.
AI doesn't replace the trust relationship. Clients still want to know there is a human on the other end who is responsible if it breaks. AI tools are the tools the human uses; the human is what they're paying for.
4. Judgment
What to ship, what to kill, what to charge, who to hire, when to push back on the client. Every era has needed it. AI has not made it less necessary; AI has made the rest of the work faster, which leaves more time for the judgment calls that are the actual valuable part.
The three skills that don't transfer
These were valuable for one tool era and then became liabilities.
1. Tool-specific procedural knowledge
Knowing every Lotus 1-2-3 keyboard shortcut. Knowing the Flash timeline. Knowing the Dreamweaver shortcut for inserting an image. None of these survived. They all expired with the tool.
2. Technique-specific manual labor
Hand-coding HTML by typing every tag was a skill in 1998. WYSIWYG editors made it slower than the alternative by 2005. Manual Photoshop masking was a skill in 2015. Generative fill made it slower than the alternative by 2024.
The skill the person had wasn't the labor. The skill was understanding what to produce. Once they switched tools, the production was faster, but the underlying capability was the same.
3. Gatekeeping the obscurity of your tool
Every era has practitioners who try to make their tool sound harder than it is. "Only I can do this because the tool is so complex." This works until the tool gets easier. Then the gatekeeping value drops to zero.
AI is making every tool easier. Practitioners whose value is locked in tool-gatekeeping are losing it. Practitioners whose value is in the four transferable skills are not.
The practical move for any era shift
Same playbook every time. 30 minutes a day. 90 days. On the new tool.
The FoxPro consultant who learned .NET did it this way. The Flash animator who learned HTML5 + CSS3 did it this way. The jQuery developer who learned React did it this way. The 2026 marketer learning Claude is doing it this way.
90 days is enough to be productive. 12 months is enough to be competitive. 24 months is enough to be senior. The math hasn't changed in 30 years.
What this looks like for 2026 specifically
The tools to learn this era are short-listed:
- Claude or ChatGPT for general LLM workflows.
- Midjourney + Ideogram for image generation.
- Cursor or GitHub Copilot for code-assisted work.
- Gemini for huge-CSV and web-research workflows.
- Notion AI or Anthropic API for workflow automation.
Pick 2 to 3 based on your work. Spend 30 minutes a day for 90 days. The transition is real but it is not different from any of the prior eight. You have done this before, or your colleagues have. The recipe is known.
Frequently asked questions
Is AI really like Flash or FoxPro? It feels more disruptive than that.
What if I'm 50+ and don't want to learn another tool?
Don't AI eras come faster than tool eras? Won't we have another shift in 2 years?
What if AI gets so good it eliminates judgment too?
What's the one piece of advice for someone facing the AI transition?
How do I tell when an era shift is over?
Related reads
- How to learn AI the right way. The 90-day curriculum, with capstone.
- Which parts of your job to hand over to AI. The plumber rule for the split.
- How to use AI for ad creatives. One worked example of riding the shift in marketing.
Maddy
Maddy runs every WeActive8 engagement personally. Nine years working on growth across SMB and funded-startup stacks. Builds the 8CRM, Team8s, 8Host, and 8Automations products.
