Everyday AI Podcast – Ep 712

AI Agent Crash, Software Collapses, and Non-Human Economies

Host: Jordan Wilson
Date: February 12, 2026
Episode Theme:
A forward-looking, practical exploration of top AI trends, risks, and predictions for 2026, focused on autonomous agents, software industry disruption, and the formation of new, non-human economic ecosystems.

Episode Overview

Jordan Wilson opens the first volume of his annual “2026 AI Predictions and Roadmap” with a set of 13 high-impact, often provocative predictions about where AI, the software industry, and enterprise realities are headed in 2026. The episode covers agent-based economies, collapse of software valuations, emerging agent risk surfaces, the regulatory landscape, and how work itself is being impacted by AI agents that interact with one another more than humans.

Jordan’s distinctive, unscripted style brings a candid, real-world take, enhanced by his insights from 700+ episodes and frequent conversations with leading AI developers, enterprise users, and strategists. He shares internal notes, insider conversations, and “first wave” observations to help listeners prepare for a year likely to be defined by rapid change, disruption, and unexpected crises.

Key Discussion Points & Insights

Prediction Summaries & Deep Dives

Jordan quickly lists his “top 13” for the year, then discusses each for 2–4 minutes. Below are the core insights and memorable explanations for each prediction.

1. AI Labs Will Pay Universities for Student Data

• Insight: Elite AI labs (e.g., Microsoft, OpenAI, Anthropic, Google) will form structured partnerships with universities to access high-quality, human-originated reasoning and conversation data.
  • “I’m not going to say acquires a university because that sounds crazy even for me. …but I think in 2026, at least one frontier AI lab will announce a structured partnership with a university centered explicitly on access to student-generated data—human data. Yeah, we need it.” (19:13)
• Why: AI has “so much slop” in its datasets—enterprise and web data aren’t enough. Human-generated, original classroom thinking is scarce but essential for next-gen models.
• Notable Quote: “Students that participate are going to get compensated, right? Whether it’s free tuition, reduced tuition, whatever. AI labs have money to burn… Colleges are going to go broke. Perfect storm.” (23:28)

2. Venture Capital Firms Will Morph Into Venture Studios

• Insight: VC firms, especially in software and SaaS, will shift to an in-house “venture studio” model, drastically reducing the need for external engineering talent as models write most new code.
  • “Engineering cost is going down to pennies on the dollar. Why wouldn’t a venture firm turn into a venture studio? ...They have all the processes down, cuts your engineering cost by 97%.” (26:03)

3. ‘Grounded Only’ AI Models Become an Enterprise Staple

• Insight: Major AI platforms will offer “grounded only” modes—where the AI draws only from user’s internal/company data, not the wider web or generic model training data.
  • “There just needs to be a ‘grounded’ button, right? Use my data only, nothing else.” (31:30)
• NotebookLM (Google’s product) is cited as the early leader; expect competitors to ramp up.

4. Google Wins the Short Game While Building the Long Game

• Short Game: Google continues outperforming in AI benchmarks with Gemini, dominating every release cycle (“at any point, Google… can snap their fingers and it’s going to be done.” (35:16))
• Long Game: Google’s unique asset is YouTube (“a gold mine as models get smarter”)—critical to embodied AI, robotics, and video understanding (38:00).
• Quote: “In the end, it is about embodied AI, and Gemini is consistently ranking at the top.” (39:00)

5. Internal Agent-to-Agent Economies Emerge

• Insight: Enterprises will deploy internal marketplaces where AI agents dynamically contract and delegate tasks to other agents.
  • “Agents are going to hire other agents. …You are literally going to have enterprise agents out there, they’re going to have budgets and they’re going to go to big Software A… and you’re just going to have agent-to-agent commerce.” (41:25)
• New protocols (“MCP”, “2A agent”) enable autonomous negotiation and commerce between agents—mostly invisible to humans.

6. Software Stocks and ETFs Face a Drawdown

• Market Impact: Expect a sustained fall in the value of traditional software company stocks, with entire ETFs (software baskets) already down over 20%.
  • “They have tech debt, right? They have to start over, or… spend months or quarters to implement features that startups can implement in hours.” (45:30)
• Agents bypassing traditional UIs mean many familiar applications will be rendered obsolete.

7. Messaging Becomes the Universal Agent Control Interface

• Insight: “How we talk as humans—day to day—is going to become the interface. …Slack, phone calls. I think that’s how we’re ultimately going to communicate with agents.” (49:02)
• Evolution: Texting, voice messaging, and Slack become the main way to summon, instruct, or monitor agents, removing the need for traditional GUI interfaces for most workflows.

8. ‘Agent Crash’–A National Headline Event

• Prediction: In 2026, there will be a high-profile, catastrophic autonomous agent failure (not just prompt hacking)—think “AI gone wrong” at scale, triggering real regulatory conversations.
  • “It’s going to be big. …The media coverage is just going to focus on autonomy gone wrong. …We are going to see a big incident.” (52:28)
• Quote: “It’s not like a fomo, it’s a rush to not get lapped… which is going to lead to, I believe, a huge and noteworthy agent crash in 2026.” (53:40)

9. Shadow AI Triggers Fortune 500 Data Breach

• Insight: Unapproved, “shadow AI” (i.e., employees using unauthorized, often sketchy tools) will result in a major, public data breach at a Fortune 500 company.
• Reasons: Lack of corporate AI training, locking down sanctioned tools, proliferation of rogue “wrappers.”
  • “People don’t know any better… oh, this is like ChatGPT but for construction—oh great, I’m in construction—no, you know, don’t.” (56:00)
• Anticipated Reaction: Some firms may lock down even copy-paste from employees.

10. Agent Audit Logs Become Mandatory in Regulated Sectors

• Insight: Owing to the previous two points, finance, legal, health sectors will require detailed, transparent logs of all agent actions (“no more black boxes, no more opaqueness in your AI processes” (59:11))
  • “Agent audit logs become lifelines for your company… otherwise, you… might get shut down.” (1:00:15)

11. Skill Marketplaces Become Largest Agent Risk Surface

• Insight: Open skill marketplaces (“like an app store for agents”) will become the top vulnerability, as malicious actors slip malware into productivity tools.
  • “The number one skill on the open-claw hub was malware. …People are going to rush to use these, and quietly just ruin lives, departments, companies.” (1:03:00)
• Warning: "You have to be extremely careful" about granting agent access to local files through unvetted skills.

12. 12-Hour Autonomous Task Horizon Becomes the New Brag Metric

• Insight: Out-the-box agents will be able to autonomously complete complex, multi-step projects lasting 12+ hours (at a >50% success rate).
  • “AI is going to move from a tool to a functional employee, maybe a junior level employee in 2026.” (1:06:34)
• Impact: Agents tackle not just tasks, but long-form, integrated, team-scale project work.

13. AgentOps Becomes Formal Enterprise Function

• Insight: Enterprises will establish dedicated “Agent Operations” teams to monitor, version control, and optimize fleets of agents.
  • “Agent uptime is going to be one of the most important KPIs. …AgentOps is going to become as normal as DevOps.” (1:08:55)
• Long term: By 2030, many knowledge workers will be in some flavor of AgentOps.

Notable Quotes & Memorable Moments

Jordan's candid, often humorous style brings out vivid language and analogies:

• “I have this running series of notes… Sometimes I can spot what’s coming before others because this is all I do.” (05:24)
• “AI labs are literally sitting on a stack of money so high they have to duck to fit into… to, to, to think, right? That’s how much money they have.” (23:50)
• On legacy enterprise software:
  “Even if it’s a great house… a house that’s 150 years old—good luck with the pipes… You’d rather have the new house, always.” (46:08)
• On the coming AgentOps era:
  “We are becoming the buns for the agent sandwich. We give it directions and context on the front end, and feedback and iterations on the back end. But the agent does the majority of the work—it is the meat, cheese, and the condiments in the middle.” (1:09:10)

Key Timestamps

| Segment | Timestamp | |--------------------------------------------------------------|-------------| | Introduction to Episode & Series Purpose | 00:16–05:20 | | Jordan’s process, background, grading past predictions | 05:20–07:40 | | Preview: 13 top 2026 predictions (list only) | 09:40–13:00 | | 1. AI Labs Pay Universities for Data | 13:00–24:10 | | 2. VCs Become Venture Studios | 24:10–27:56 | | 3. Grounded-Only Models Go Mainstream | 27:56–31:50 | | 4. Google Wins Short Game, Builds Long Game | 31:50–38:55 | | 5. Internal Agent-to-Agent Economies | 38:55–41:55 | | 6. Software/ETF Collapse | 41:55–45:40 | | 7. Messaging as Universal Agent Interface | 45:40–49:20 | | 8. Agent Crash Becomes National Headline | 49:20–53:45 | | 9. Shadow AI Triggers Fortune 500 Breach | 53:45–56:54 | | 10. Agent Audit Logs Mandatory | 56:54–59:30 | | 11. Skill Marketplaces as Top Risk Surface | 59:30–1:04:20| | 12. 12-Hour Autonomous Task Horizons | 1:04:20–1:06:56| | 13. AgentOps as Formal Function | 1:06:56–1:10:58| | Rapid Recap of All 13 Predictions | 1:10:58–1:12:40|

Takeaways & Flow

Jordan’s episode is fast-paced, candid, and packed with usable insight. Major themes include:

• The AI/data economy’s shift from “human-in-the-loop” to multi-agent, non-human markets.
• Imminent collapse of “legacy” software business models, as workflow automations and agents replace traditional SaaS interfaces.
• Emergence of new workplace roles (“AgentOps” and beyond) focused on overseeing, orchestrating, and auditing fleets of semi-autonomous AI helpers.
• Elevated stakes for AI safety, compliance, and security—especially as agent capabilities and risks multiply.
• A call to action for leaders: “You have to be ahead… and you have to be safe… Don’t rush.”

End of Volume One—preview for Volume Two tomorrow covering the rest of the 26 predictions.

For a free, more in-depth bonus guide and daily AI newsletter, Jordan directs listeners to repost the LinkedIn stream and subscribe on the show's website.

If you want a comprehensive edge on AI’s near future, this is required listening.