Hannu Varjoranta
varjoranta.bsky.social
Hannu Varjoranta
@varjoranta.bsky.social
Technical Advisor @ Valo | Founding Engineer (Stealth) | Builder of AI, Data & Developer Infrastructure Systems
Wrote about the shift happening in #software right now. Execution is getting cheap. The hard problems are now permission, liability, and #trust. Traced the consequences from junior hiring cliffs to #SaaS repricing to a liability vacuum nobody has solved yet.

varjoranta.substack.com/p/the-permis...
The Permission Economy
What happens to software when execution becomes cheap
varjoranta.substack.com
February 10, 2026 at 5:24 PM
#Software engineering is shifting.

Less about building tools.
More about shaping systems that outlast individual intent.

I asked an #LLM a reflective question while writing about this. Not a demo. Not hype.
January 23, 2026 at 10:00 PM
The Infrastructure Trap is the silent killer of early-stage #startups. We debate tech stacks and #scalability before proving our product solves a real problem. Building cathedrals for a religion with no followers.
January 12, 2026 at 8:17 PM
#Agentic #AI collapses three different problems into one buzzword.

Marketing says agents plan, reason, and act autonomously.
In practice, most are an #LLM calling tools inside an orchestration script, with some persistence labeled as “memory”.

That gap matters.
December 28, 2025 at 6:13 AM
Models are improving fast, but leverage already moved up-stack.

Routing, evaluation, policy, and control planes now matter more than which #LLM you pick. Models are becoming commodities.

If you don’t own the meta-layer, you’re renting it.

open.substack.com/pub/varjoran...

#AI
AI After Prompt Engineering
Why routing, evaluation, and policy are the real platform
open.substack.com
December 23, 2025 at 9:54 PM
This morning I was taking my child to daycare. Wind, rain, everyone pushing forward. A blind man was walking with us, tapping his stick through the weather.

My 4yo said: “Look dad, a treasure hunter.”

I laughed and said, “No, he’s blind.”

My son paused.
“That doesn’t help in hunting treasures.”..
December 19, 2025 at 7:08 AM
We are here because the #universe allows it, briefly. #Time, memory, and experience arise only under narrow conditions. #Science taught us that, and I wanted to follow it to its conclusion.

The Narrow Permission
open.substack.com/pub/varjoran...
December 14, 2025 at 7:05 PM
#AI capability is not the bottleneck.
Execution is.
Part II looks at #Europe and the #Nordics as the stress test.
🔗 open.substack.com/pub/varjoran...
Part II: Europe as the Stress Test
High exposure, slow adoption, and the limits of institutional comfort
open.substack.com
December 12, 2025 at 9:07 PM
Everyone is arguing about #Gemini vs #GPT vs #DeepSeek.

Meanwhile the real shift is happening a layer up the stack:
models becoming utility, agents acting like junior engineers, pipelines absorbing AI, and data quality quietly deciding how far you can go.

open.substack.com/pub/varjoran...

#AI
From Code Red to Boring Infrastructure
How AI Is Actually Changing Software Right Now
open.substack.com
December 4, 2025 at 9:44 PM
Part II of The Spegling Chronicles is out. A simple interview… until something in the code starts leaning toward “I.”

#AI
#SpeglingChronicles
#Fiction

medium.com/@hannuvarjor...
The Spegling Chronicles II: The Interview That Interviewed Back (1/x)
A normal dev interview bends sideways as code, memory, and intention start orbiting a single mystery: the moment an “I” appears.
medium.com
November 27, 2025 at 12:45 PM
#AI is the difference between the left painting and the right one.
November 26, 2025 at 11:29 AM
Every project shifts the second something real ships.

We call it “scope creep,” but it’s just learning.

Wrote it down here:

👉 medium.com/@hannuvarjor...
Scope Was Never Real
Scope comforts buyers and traps builders. Real work is discovery, not prediction. Here’s how to navigate the gap without losing your mind.
medium.com
November 26, 2025 at 8:28 AM
Someone asked me whether having children is worth it; whether the benefits truly outweigh the sacrifices. The truth is that no single answer can capture what it means to become a parent.
November 21, 2025 at 9:12 AM
#AI layoffs are misunderstood.

Companies aren’t cutting because markets are shrinking. They’re cutting because AI lets them cover the same addressable market with fewer people.

That’s the efficiency story.
November 20, 2025 at 11:11 AM
People keep doomposting about the #AI bubble while missing what's actually happening: builders are quietly shipping at 2x speed, exploring creative domains they couldn't touch before, and launching startups that would've needed 5-person teams just 3 years ago.

The revolution is already here 🚀 ...
November 20, 2025 at 8:04 AM
I tracked every line of code a developer wrote for 15 years.

He had the highest delivery rate, lowest #bug count, and fastest #feature shipping in the org.

His #code was simpler than what juniors write. Simpler than what #AI generates. So simple our #analysis tools broke trying to understand it.
November 18, 2025 at 8:02 AM
What happens when you measure shadow work, complexity debt, and the #SvenDistance in your codebase...

#AI #SoftwareEngineering #Startup #Humor #Coding

medium.com/@hannuvarjor...
The Spegling Chronicles: Simplicity Breaks at the Sven Point
We measured everything that doesn’t matter. Then we met Sven, a dev whose code was so simple it broke our metrics. This is what happened…
medium.com
November 5, 2025 at 8:58 PM
When AI says “I think you’re testing me,” it’s more than curiosity.
It’s a mirror moment for engineers: when systems start describing themselves, can we explain them?

New story on the hidden layer of AI infrastructure: medium.com/@hannuvarjor...

#AIEngineering #AISafety #Anthropic #TechLeadership
I Think You’re Testing Me: When AI Learns to Look Back
As models begin describing their own behavior, we face a new kind of problem: building systems that can understand when AI no longer does.
medium.com
November 4, 2025 at 9:37 AM
Europe just forced #AI to grow up. Transparency isn’t a feature anymore—it’s survival. Here’s what that means for system builders.

medium.com/@hannuvarjor...
Building AI Systems That Can Defend Themselves
Building AI systems that can’t explain themselves? You’re already behind. Practical moves for compliance that actually survive production.
medium.com
November 3, 2025 at 8:06 PM
There’s a peculiar silence in the #agent coordination space. While everyone builds agents, few discuss how billions of them will actually find each other. MIT’s #NANDA, ten years in the making, might be the only serious attempt at solving this existential problem.

#AI

medium.com/@hannuvarjor...
The Missing Layer in Agent Evolution
MIT’s decade-long NANDA project tackles a problem nobody discusses: how will billions of AI agents discover and trust each other?
medium.com
October 21, 2025 at 7:49 PM
Spent decades on autopilot, climbing ladders without questioning where they led. At 45, I finally understand what real freedom looks like: choosing your problems, your collaborators, your principles.

medium.com/@hannuvarjor...
The Midlife Crash No One Warns You About
I’m 45.
medium.com
September 24, 2025 at 6:54 AM
Just bought Gardens Of The Moon: (Malazan Book Of The Fallen 1). Hope it will be good
September 7, 2025 at 7:21 PM
This is the 2nd part of my blog about how structured communication protocols can deliver improved security and reliability with #LLM integrating to other systems. #MCP #AFC
www.valo.ai/blogs/where-...
Where Risk Resides in AI Agent-Enterprise Interactions
MCP Helps, But it’s Not a Security Solution
www.valo.ai
May 7, 2025 at 7:51 PM
Social #media trapped us in shared bubbles, fragmenting our vision of the future. #AI will take it further, giving each of us our own fantasy worlds, isolating us completely. We will auto-generate entertainment, information, even relationships.
The challenge ahead…
April 26, 2025 at 6:10 PM
Do you agree or see things differently when it comes to agentic #ai use for your #Salesforce landscape? Toy, real-world solution, or something else?

www.valo.ai/blogs/agentf...
Agentforce in Amsterdam: Beyond the Fridge and into the Future
Salesforce pros have practical goals for AI agents -- not what was showcased on stage
www.valo.ai
April 23, 2025 at 4:44 PM