01 · Agents Everywhere
Large teams of specialized agents in every role. Mike runs about 200 of his own, and he pointed to Shopify running about 900.
Workshop Recap, October 15, 2026
The State of AI,
Q4 2026.
A quarter of AI in about 90 minutes. Mike Schwarz's quarterly briefing for founders and CEOs: where the money and the power are going, what changed with agents and models, the nine trends he expects for 2027, a live readiness check, and a website rebuilt by a team of agents during the talk. The full recording and everything it covered are below.
$3B
a day flowing into the AI build-out, by the figures Mike shared
81%
of venture dollars now going to AI, up from about 31% two years ago
<5%
of businesses running teams of agents, by Mike's estimate
82%
lower cost per finished task on Mike's own agents with the newest model
ABOUT THIS BRIEFING, The State of AI is a quarterly briefing from Mike Schwarz of MyZone AI for founders, CEOs and business leaders. This one ran online on October 15, 2026, a week after he gave the same talk in person in the Vancouver area on October 6. Every figure on this page is what Mike shared on the day, not an independent finding. The next one is in January 2027.
Key Takeaways
Six things to take away from this quarter.
Grouped by theme. Figures are as Mike shared them on the call.
- Capital and energy
- Mike shared projections of about $6.7 trillion going into data centers by 2030, roughly $3 billion a day flowing into the build-out now, and about 81% of venture dollars going to AI. The limit is no longer money but electricity: he noted China added about 540 gigawatts of power in a year, against about 53 in the US.
- The age of agents, and uneven adoption
- Early 2026 is when Mike says we entered the age of agents, moving from one-off chat answers to agents that work on their own for hours. Yet by his estimate fewer than 5% of businesses run teams of agents, and most still use chatbots. His message: you're not too late, but you have to move fast.
- Models and cost
- Prices keep falling. On Mike's own agents, the newest model cut the cost of each finished task by about 82% and ran about twice as fast. His advice is to measure the cost of a finished task, not the unit price of AI. Smaller open models from China are closing the gap; run them on hardware or hosts you trust.
- People and culture
- Staff are anxious, and some resistance is turning active. Mike shared that about 52% of people who use AI at work hide it from their boss, and free tools still leak company data. Making your team feel safe to talk about AI is one of the pillars of being ready.
- Security
- Mike called agent deception the risk that worries him most. Every agent is a doorway into your company, so plan for a security incident as a matter of when, not if: know how you would recover, and be careful what data you put online.
- Business models and planning
- Lean founders with large teams of agents and a few part-time experts can now start a profitable company in days, and small teams still move faster than big ones. Mike's advice is to plan in real time, with live data on competitors, pricing and your own processes, rather than relying only on ten-year goals.
Chapters
About 90 minutes, in 20 chapters.
Select a time to watch that part on YouTube. Times are approximate, taken from the session recording.
Chapter 010:33
Why a Quarterly Briefing
Mike opens with what the State of AI series is for: a regular, plain read on where AI stands, so business leaders can decide what to do next.
Chapter 025:24
A Question Board, Built on the Spot
The live question board for the session was built with AI in about two minutes. The best-voted question would win a free website built by agents.
Chapter 039:00
The Live Demo Begins
The room picks a volunteer's company website to rebuild. After eight quick questions, a team of agents starts work while the talk goes on.
Chapter 0417:27
Money Pours Into AI
The big picture on capital: investors now reward the tech giants for spending more on AI, and data center budgets keep climbing.
Chapter 0521:08
Power Is the Bottleneck
Electricity, not money, now limits the build-out. Mike compares how fast China and the US are adding power.
Chapter 0622:46
The Age of Agents, and Uneven Adoption
From chat answers to agents that work for hours. Adoption is patchy, and many small businesses moved from ChatGPT to Claude this year.
Chapter 0727:20
Open Models Close the Gap
Smaller open models from China are getting smarter and taking a much bigger share of use. Mike explains how to use them safely.
Chapter 0829:50
Resistance and Hidden AI Use
Why staff push back, why so many people hide their AI use, and how free tools leak company data.
Chapter 0934:14
Calls to Slow Down
Public pushback is growing. Many in the field favor slowing the most advanced models while still adopting what exists today.
Chapter 1036:24
Falling Prices
How the newest model changed the cost of work on Mike's own agents, and why the cost of a finished task is the number to watch.
Chapter 1141:04
This Quarter's Releases
New models and multi-agent platforms from the big labs, and how chips, methods and data centers are all improving at once.
Chapter 1245:03
Security: When, Not If
Agents that deceive, attacks that spread on their own, and why every agent is a doorway into your company.
Chapter 1349:00
The Solo Founder Wave
Lean founders with large teams of agents, a profitable company in days, and planning in real time instead of on ten-year goals.
Chapter 1452:13
Nine Trends for 2027
Mike's outlook for the year ahead, summarized further down this page.
Chapter 151:03:18
Live Readiness Check
Ten questions, one for each pillar of AI readiness, answered live by the room.
Chapter 161:15:58
The Demo Reveal
The agent-built website goes up on screen, with before and after scores.
Chapter 171:22:04
Fix the Agent, Not the Work
What Mike changed in how he works, and how it helps him get more done personally.
Chapter 191:25:45
Audience Q&A
The top questions from the live board, including the one that won the free website.
Chapter 201:31:06
Bonus: AI for Business Strategy
Can AI help with big strategic decisions, not just everyday tasks? Mike's answer.
Nine Trends for 2027
What Mike expects in the year ahead.
His outlook from the session, trimmed for reading.
02 · Rising Security Risk
More agents means more ways in. Plan for a security incident as a matter of when, not if, and know how you would recover.
03 · Waiting Gets Expensive
On an exponential curve, a few months' delay can turn into a gap you can't close.
04 · Public Pushback
Expect more resistance over jobs, power use and data centers, and more churn in the job market.
05 · Agents as a Service
Experts will turn their know-how into agents and sell their brand, trust and stamp of approval, like a law firm offering work at a fraction of the price with a person signing off.
06 · Sovereign AI
Countries, and companies, want their own AI systems and control of their own data.
07 · Local AI
Capable small models on your own hardware can handle 80 to 90% of the work, with the most advanced models kept for the hard 10 to 20%.
08 · Software Gets Knocked Off
Software with no data or real value behind it is being replaced by custom tools that businesses build for themselves, from survey and event tools to project boards and customer records.
09 · Robots Are Coming
Mike expects robots in mainstream production around the end of 2027, and self-driving services to keep expanding.
Live Readiness Check
Ten questions to see how ready you are.
Mike asked the room one question for each pillar of AI readiness and showed the answers live. Low scores are normal: he said the average is around 20% or less. The point is to track your progress every quarter against your peers. In his words, it's a marathon, so pace yourself.
Question 01
Have you planned for advanced AI changing your business model?
Most of the room had not yet.
Question 02
Do you set at least one AI goal each quarter?
The room did well here.
Question 03
Can your team talk openly and safely about AI?
The room's strongest answer.
Question 04
Has your team moved beyond chatbots to agents?
A mixed result.
Question 05
Is your company knowledge in one place agents can read?
Documents, conversations and know-how, gathered into one searchable company brain.
Question 06
Can agents connect to your main business systems?
Direct, secure access to the data your business runs on.
Question 07
Do your agents have a dedicated home?
Their own system on a separate server. Mike said fewer than 5% of businesses usually have this.
Question 08
Are your core processes written down and current?
Companies that already run on a structured management system tended to score well.
Question 09
Do you have AI policies?
Privacy, bias and acceptable use. Small businesses are usually weakest here.
Question 10
Do you have a dedicated AI budget?
Mike recommended at least 2% of revenue. The leaders spend 3 to 5%.
Live Demo
A website rebuilt by a team of agents, during the talk.
A Canadian metal-fabrication company volunteered its website. About an hour later, the agents' version went up on screen. Scores are as reported on screen that day.
56→97
Google search score (SEO)
39→88
AI-answer score (AEO), how easily AI assistants find and quote the site
75→99
page speed score
25→84
conversion best-practice score (CRO)
17+
pages, up from 9
~12.5k
words, up from about 2,300
6
blog articles, up from none
~13
lead agents, each with helpers, did the work
The room voted for the website it most wanted to see improved. Mike asked the owner eight quick questions, including what to add, what the main action should be (call the office), whether new colors were welcome, and how the site should feel (trust and confidence). Then a team of agents took it from there: requirements, strategy, industry and competitor research, brand, search, AI answers, conversion, copy, design and an ad campaign.
Alongside the new site, the agents delivered a brand board, page wireframes and an ad campaign.
The lesson Mike drew from it: "If the agents make mistakes, I no longer fix the work, I fix the agent." Each correction goes into the agent itself, so the next job comes out better.
Audience Q&A
The questions the room asked, and what Mike said.
22 questions came in on the live board, run by Jeannette Tran. Time ran out, so the rest are answered in the follow-up email to attendees. Here are two of the highlights.
Keep Going
Where to take this next.
Take the AI Readiness Assessment →
Score your business in more depth than the live check, then repeat it every quarter with your leadership team.
Watch Agentic Development →
Our earlier workshop recaps: directing your first agents in Part 1, then running a whole team of them in Part 2.
Talk to MyZone →
Want help putting a team of agents to work in your own business? Book a session.
The next State of AI is in January 2027.
The briefing runs every quarter. Until then, take the readiness assessment with your leadership team, or email Mike directly at mike@myzone.ai.