My outlook and prediction for AI Automation and AI enablement going into 2026:
I keep saying "automation" but I'm not sure that's what enterprises are actually buying.
They're buying reliability. Predictability. Something that doesn't randomly fail on Tuesday because the LLM had a bad day. By 2026 everyone will be able to string together API calls, so that's not a barrier anymore. The barrier is being able to promise it'll work the same way EVERY TIME.
The agencies selling magic will churn. Autonomy compounds failure. 95% accuracy per step sounds good until you're 15 steps deep and your success rate is 46%. Enterprises don't want systems that works "most" of the time.
They want systems that work or they don't buy.
I thought generalised automation was the play at first. Build tools that work across industries, sell to everyone. Wrong constraint. The bottleneck isn't technical capability. It's that LLMs are brilliant when you narrow them down to specific problems with specific context. Useless when you point them at general workflows. CFOs don't approve "AI tools" anyway. They approve "cut claims processing costs by 40%". And you can't promise that without actually understanding claims processing.
Domain knowledge is the barrier. Not prompts.
Which means you need to pick an industry and go deep instead of trying to sell horizontal tools. You don't like that answer but it is true.
Compliance is boring but it might be where the actual money is. The EU AI Act shifted liability from model providers to deployers. Companies have shadow AI usage everywhere and no visibility into it. They need control planes. Systems that monitor inputs, scrub PII, enforce policy before anything touches the model. This is a compliance purchase. Bypasses the ROI conversation entirely because the alternative is legal risk. Recession-proof. Unsexy infrastructure work that most agencies won't touch.
On prem matters more than I thought six months ago. Enterprises are realising that sending proprietary data to OpenAI creates an IP leak. Open weights models will be good enough for most internal tasks by 2026. Maybe not bleeding edge but good enough. The friction is infrastructure complexity. DevOps and ML engineering talent. Not just calling APIs. High barrier to entry if you can't actually build it.
Autonomous agents are beautiful in demos and terrible in production. Everyone's selling "employees in a box" but error rates compound. A system that works 80% of the time isn't an asset when it's handling critical business processes. Screen scraping SAP is still brittle. Legacy enterprise software won't expose clean hooks fast enough. Standards like MCP will help but not by 2026.
The technology will exist. The reliability won't.
The intelligence cost is trending to zero.
The value moved to context injection and output verification. Agencies building knowledge graphs that let AI understand business data will scale.
Agencies building Zapier/Make/n8n style connectors will get commoditised. I think. The Magpie Phase is over. Enterprises spent two years buying shiny things. Now they want guarantees.
f*ck it, i'm leaking my complete app building system...
for only 24hrs, you get access to:
- my full vibe coding setup with Anything AI & Max
- 10 validated app ideas i was gatekeeping
- how to go from 0 to $10k/month with the setup
basically the entire workflow for building apps with an autonomous AI engineer
reply "MAX" + RT & like and i'll send it over (must be following so i can dm)
i tried to keep this private but Vas convinced me to release it...
we built the Whop Apps Printer - the complete system for making money with mini-apps, including:
- why Whop Apps are printing money right now (the golden opportunity)
- vas's step-by-step vibe-coding guide to build your first app
- my exact blueprint for selling to community owners (leaking the 3 best niches)
free for 24 hours only, then it's gone forever
reply "PRINTER" + RT and follow @vasuman to get your access (follow me too so i can dm)
My conversation with @naval. Enjoy!
(0:00) - The Theory of Everything
(4:48) - How do you know what’s true?
(7:51) - Groups search for consensus, individuals search for truth
(13:07) - We have never run out of a single resource
(15:25) - Are we destroying the Earth?
(17:48) - Marxism denies wealth creation
(21:28) - Regulation kills innovation
(27:05) - Degrowth and the fall of Western universities
(33:31) - Why the West is best
(35:47) - Federalism
(38:10) - Everyone wants to live forever
(41:44) - Humans are universal explainers
(43:25) - Collectivism vs. individualism
(50:44) - You cannot explain the universe without explaining humans
(55:02) - How @DavidDeutschOxf’s ideas have changed Naval’s life
(1:02:31) - The scientific method isn’t possible
(1:05:07) - The low-hanging fruit theory is a bad explanation
(1:08:19) - The biggest threats to Western civilization
The opening ceremony of the Olympics is not even hiding the satanic rituals anymore.
There is no longer anything hidden in plain sight.
These are the demonic dark forces of Satan who want to eclipse the light.
You can already see:
- Skulls
- Motifs of death
- Satanic red lighting
- Children trapped in tunnels
Did anyone notice that the single rider on a pale horse is straight out of the Book of Revelation?
"And I looked, and behold a pale horse: and his name that sat on him was Death, and Hell followed with him. And power was given unto them over the fourth part of the earth, to kill with sword, and with hunger, and with death, and with the beasts of the earth."
- Revelation 6:8
“Cancer cannot get a hold on the body if the Liver is working at optimum performance”
How many remedies involve the juice from lemons? So so many.
Fun fact - The Chinese believe your ‘soul’ resides in the Liver.