Kunso

At Kunso, we help enterprises and high-growth scale-ups accelerate their growth by setting up the right technology strategy and driving digital innovation.

We work closely with founders and project leaders to build engineering teams that deliver real results — fast, reliable, and aligned with business goals.

Services:
• Digital Venture Development
• Custom Software Development
• Remote Hiring and Team Augmentation
• Fractional CTO / CTO Hiring

Chief Executive Officer

HeroSnatch

Founder / CTO

DoubleDutch

Lead Software Engineer

CEO at Kanso Code | Digital Innovation for B2B Enterprises and Startups. Unlock digital growth.

CEO and Founder of a boutique agency to build B2B software solutions for scaling startups and large enterprises. 

We focus on fast-to-market, low-code, modern technologies to eliminate product stagnation and speed up product development. 

"Software Development should not be Slow"

With a team of elite 35 Engineers, we have developed Frameworks and Methods to build and iterate faster and unlock innovation for our clients.

We are very passionate about: Low-code, ReactJS, NodeJS, Airtable, Firestore/Firebase and other technologies.

Our primary industry is B2B SaaS.

Mikki Kobvel

Some non-business milestones. 

Some time ago, we did a race in Paris with 
Maxime Barbier and Othmane Sghir 

That time in Paris (1:11:55), I felt unprepared 
and a bit lost between stations. 

After 8 months of structured work: 
• Interval Runs 2 times a week
• Zone2 runs 2 times a week
• Hyrox-specific Weight training 

.... I was able to shave 6 minutes off the previous score.

𝗕𝗶𝗴 𝗿𝗲𝗺𝗶𝗻𝗱𝗲𝗿: Process + Consistency > Spike Efforts

Not my favorite sport, but definitely more social! 

Next Race - Hyrox Bangkok 2026 💪

PS: 
If you're interested in running/training in Singapore 🇸🇬, 
- always happy to train together

CEO @ Kunso - Strategic Engineering for Fast-Scaling Founders

Day 1 — Singapore Fintech Festival

Finance finally sounds realistic about AI.

Focus on workflows, not slogans.
Investments judged by failure rates and accuracy, not optimism.

But software delivery is still lagging.

Developers without AI.
Private models without security.

Plenty of ambition.
Still too little execution.

CEO @ Kunso — Strategic Engineering for Fast-Scaling Founders

No alternative text description for this image

Heading to Singapore FinTech Festival (Nov 12–13, 2025)

Exploring how AI is changing fintech infrastructure and secure financial systems.

Topics of interest: secure AI usage, operations portals, lean AI development, debt collection technology, and voice agents.

Who is coming? 

#SingaporeFinTechFestival #SFF2025

CEO @ Kunso — Engineering Pods & Tech Strategy for Ambitious Founders Scaling Fast

Most founders think their main problem is “tech.”

It’s not.  
The real issue starts way earlier: unclear requirements and zero alignment on what the product is supposed to do.

This is the usual pattern:

- “I saw a cool app, we should build something similar.”  
- No defined user.  
- No defined problem.  
- No defined outcome.

Half of these products don’t need to exist.  
A lot of them should have been simple automation instead of hiring a full engineering team.

Technical debt doesn’t come from bad tools.  
It comes from building without a clear target.

As software gets more commoditized (thanks to AI tools), nobody cares about your stack.  
They care if the thing actually solves a real problem.

If you can’t answer:  
• What are we building?  
• For whom?  
• What are all ways to solve it?

…then the tech will fail no matter how much you spend on developers.

On the photo: 
Adrien De oliveira Co-Founder of Timeleft 
playing around with his Replit dashboards. Being this hands-on and willing to try new things probably saved the company 50–100K that would’ve gone into unnecessary builds.


#EngineeringStrateg  #FoundersWhoBuild #TechForBusiness #Replit #ProductEngineering

Adrien De Oliveira from Timeleft working on a Replit dashboard on his laptop during a team work session.

I enjoy working with people who’ve had a dark past.

Through my career, I’ve met and hired many talented, hard-working people.
But one group stands out:

The ones who’ve been to the bottom.

When someone’s hit very low in life – they move differently.
They have a different kind of grit.
They appreciate opportunity in a way others just don’t.

But more than that –
For them, money or reputation isn’t the only thing at stake.
It’s their self-trust.

They’ve lost it once.
They know what it’s like to doubt themselves.
So now, they fight to never go back.

That’s why they show up.
That’s why they care.
That’s why integrity isn’t optional — it’s survival.

No matter how low you are –
if you care, and you work hard –
you will be recognized.

– Or are we still pretending clean CVs matter most?

Five months of work, no weekends.
Everything looked perfect.

Until the client said, “We need it on an iPad.”

A short story about one of the first lessons that pushed me to build Kunso.

The Setup There was a point in my career when I decided that the tech ecosystem is broken from the inside. I was working for a large company who was s

Open article: The $500K Lesson About Manager Bias by Mikki Kobvel • 3 min read

Asked a developer to share his screen.

Task was taking too long.

He was using IntelliJ with basic autocomplete.

No Cursor. No MCP docs. Just 2010 workflows.


If you don't lead your team to Research and Use AI, they just won't. 

Tech Leadership must incentivize behaviours otherwise things won't happen.



2025 Year you should really allocate 30% of time for Dev Productivity, 
otherwise you will pay with efficiency.

Every single day. 

#TechLeadership  #EngineeringManagement #DeveloperEfficiency

At Singapore Fintech Festival, every AI product had the same pitch.

“Automated compliance.”
“AI-powered decisioning.”
“End-to-end fraud detection.”

The demos were slick.
The branding was polished.

But most of these products don’t survive production.
We’ve tested them.

And this isn’t just a fintech problem.
It shows up in healthcare, logistics, insurance—anywhere accuracy matters.

There are only two real models of AI integration:

All-in-one automation
Built for speed and optics.
Everything is bundled—model, rules, logic, UI.
Looks great in a demo.
But it’s rigid.
When something breaks, you can’t trace or tune it.

Service-based integration
Built for real ops.
You keep your data, workflows, thresholds.
The model supports decision-making—not replaces it.
You start with one use case.
Test it. Tune it. Then scale.

We’ve seen both in action.

The all-in-one version sells faster.
The service-based one actually works.

Especially when the cost of a false positive is measured in lawsuits, churn, or blocked revenue.

The hype still favors flash.
But in the field, boring wins.

The teams that scale AI don’t chase demos.
They build feedback loops.
Human + model + control.

That’s what makes AI stick.

Not on stage.
In operations.

Last couple of weeks: talking to founders about their AI game.

Same mindset every time. There's a magical AI tool that can fix an entire department.

Here's the problem: AI integration is like making people use laptops instead of paper. You need to enable them with the correct workflows first.

If you have no good operational workflow - AI only makes it worse.

My advice to founders:
1. Find most productive employees
2. Incentivize them to use their workflows with secure AI models
3. Once you see productivity boost - automate it

Don't expect 3 → 2 → 1.

It never works backwards.

We tested Anthropic Claude Opus 4.5 and Finance is next a next stop for co-pilot work.


𝗜𝗻𝘀𝗶𝗴𝗵𝘁 𝟭: 𝗖𝗼𝗱𝗶𝗻𝗴 𝘃𝘀 𝗙𝗶𝗻𝗮𝗻𝗰𝗲
The programming principles of working with LLMs are the same. Atomic steps into compiling Excel documents. No — do this report for me.

 
𝗜𝗻𝘀𝗶𝗴𝗵𝘁 𝟮: 𝗦𝘄𝗶𝘁𝗰𝗵𝗶𝗻𝗴 𝗨𝗜
The ability to jump between the Claude interface and the Excel assistant boosts productivity by 50%. If you are stuck in a single interface, you can get stuck at certain steps.

 
𝗜𝗻𝘀𝗶𝗴𝗵𝘁 𝟯: 𝗧𝗼𝗼𝗹 𝗖𝗮𝗹𝗹𝗶𝗻𝗴
The ability of the system to consistently call pandas and Python libraries for verification makes it far less draining for the mind. You don’t need to switch hats.
 
 
𝗜𝗻𝘀𝗶𝗴𝗵𝘁 𝟰: 𝗣𝗿𝗶𝗰𝗶𝗻𝗴
You run out of tokens fast. Really fast.
 Here you need to switch the mindset from “expensive” to “How much does it actually help to release brainpower?”
 
 
This is an important paradigm shift for the world of finance.
 
Have you already battle-tested Opus 4.5 for finance?

Yesterday I was chatting with a friend who works in a Creative Tech startup.

Creative teams often have a bad feeling about AI, mostly because of the familiar “AI will replace all of us” phobia.

We walked through a dead-simple AI adoption strategy that requires almost zero effort from the Founder.

Step 1
Create a custom agent (literally just a chat with clear instructions) and use it to validate the output of a specific department.

Nothing fancy. Any time you feel an itch that AI could help, cross-check the work with the agent and see what it suggests.

Step 2
Save those instructions into a shared document and make each department continuously improve and expand them.

Over time you’ll build a growing, living set of guidelines that AI agents can follow to produce and validate work.

This keeps AI adoption lightweight, low-friction, and fully aligned with how the team already works.

#NonTechnicalFounders #FounderOperations #StartupExecution #CreativeTeams #AIForFounders

Diagram explaining how to configure different validation steps for multiple departments

It used to be all about code. Not anymore.
 
There is a new paradigm of building software.
 
Spec-driven development boosts agent-assisted coding precision by 30-40%.
 
If you first define “what you want to build”
and store it as files for your coding agent,
 

then when a developer is doing assisted development, the model hallucinates less and provides better-quality code.
 

 
The trade-off is more up-front investment in storing specifications.

But that’s anyway what good teams have been doing before agentic coding.
 
Start with the plan.
 
Is your team already familiar with 𝘀𝗽𝗲𝗰-𝗱𝗿𝗶𝘃𝗲𝗻 development?

All engineers should move into a Product Engineering mindset.

Challenge requirements, advocate for user needs, and stop falling in love with technologies.

This mindset improves ROI for the business, 
removes unnecessary over-engineering, and 

improves the satisfaction of products built by engineers.

All the middle-layer roles like:
• Product Managers
• Business Analysts
• Engineering Managers

will gradually be merged with engineers.

There is no point in having 20 meetings to align on things that will later be implemented incorrectly because of a context gap.

The closer engineering is to the end user, the higher the chance to hit the goal.

Action Items: 
• Teach Engineers critical thinking 
• Allow to challenge Requirements 
• Incentivize product initiatives

           
#𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 #𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 #𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴𝗟𝗲𝗮𝗱𝗲𝗿𝘀𝗵𝗶𝗽

We did 5 Tech Audits, 

Shocking, the biggest issue was not a lack of AI, but 
Security and Production safety. 

Surprisingly, many Founders request to validate their 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲,
so we can see where AI might fit to optimize the process.

But first, we check the fundamental security.

And that's where it gets crazy, with an AI boom, 
We see 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗞𝗲𝘆𝘀 flying around like spam emails.

The 2nd Biggest issue is poor architecture and 8,000-line code files. 

Those things are likely caused by AI development,
But the Security risk entirely outweighs other issues. 

• Big Files - Slow Delivery

• Data Leak - Lawsuit / Shut down of business. 

________________________
          
If you want Kunso to validate your Technology Strategy,
DM me - 𝗧𝗲𝗰𝗵 𝗔𝘂𝗱𝗶𝘁. 

And I will be sending you the qualification Form to make your business bullet proof  ✓

The two biggest problems in Engineering: 
1. AI Productivity
2. Technical Debt 

And, funny enough, there's a direct correlation.
The more you use AI without a proper process, the more Tech Debt you create.

On the other hand, if you do only 2010-style programming, 
- not using AI Tools,

You're just unbelievably slow compared to AI-enabled engineering. 

And the truth is somewhere in between. 

Our list of rules for AI Engineering: 

1. Explore new Models/Tools on side projects
2. Integrated tested experiments into engineering teams fast
3. Keep old practices of QA / Security 
4. Always have HITL (Human/Architect in the loop) reviewing all the code.

It is very easy to go maximalist on one or the other end,
But with the mindful integration of AI into engineering, you can get reasonable productivity.

Google quietly shipping a new Developers AI-Native IDE

While everyone is fighting about what is the way to build AI products,
Google ships.

Pretty phenomenal catch-up so far from the outlier who once was the leading horse in AI.

Google already shipped:

- NotebookLM – a personalized AI document assistant
- Google Flow – experimental AI filmmaking product
- Google Firebase Studio – AI-native development environment
- Google Stitch – AI design wireframing
- Gemini CLI – interactive agentic CLI tool

NEW:
🔵 Google Antigravity – Agentic Development Platform

Already testing it with Kunso to see how useful it is on the actual battlefield of development.

Link to the tool in the comments.

What Google products do you actively use?

101 Technical Debt Explained 

When you build from 0 to 1, everything appears from the void. 

Nothing is impossible, and things just pop up every week. 
Everyone is excited, and the train is moving.

Then you have Millions of Users, and you want to
"Just add this simple screen." 

And it takes 1 month! 
How come? We've built an entire app in 3 months

Adding a new screen is another 4 weeks? 
Something feels off....

The quality of the 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝘆𝗼𝘂𝗿 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 will determine
The speed of execution of future features.

And there is no right or wrong. 

Building scrappy and throwing away is also a strategy
It should just be explicitly communicated.

When you build a house, you need to know: 
• How many people will live in it? - So you don't have to build attached buildings
• Do you need windows? - so you don't have to break walls later
• Are there hurricanes in the area? - So it is not getting destroyed after 1 season

All the questions apply to software.

Ask things early - NOT because you won't have technical debt,

Because you will have expected a measured Effort / Technical Debt Ratio.

Your engineers are falling behind if they don’t do these 4 things on a weekly basis.
 
As much as everyone is too optimistic about AI products / Agents, coding agents actually work perfectly well under Senior Engineering guidance.

If you keep sending your team random SaaS for 19$/month thinking it will improve productivity, you’re just distracting them.
 
Here are the real 4 things any tech-enabled company should do not to fall behind in AI Tech:
 
𝟭. 𝗜𝗻𝗰𝗲𝗻𝘁𝗶𝘃𝗶𝘇𝗲 𝗘𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝘀
Give your team dedicated weekly time to rebuild a part of the code with a new model and expand the documentation layer around it. 

Reward the effort even when the experiment fails. Engineers need structured time to try new things, not late-night weekend hacks.
 
𝟮. 𝗢𝗿𝗴𝗮𝗻𝗶𝘇𝗲 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽𝘀
Have a dedicated weekly slot with an agenda where engineers share new tools, libraries, and things they’ve already adopted.
 
𝟯. 𝗖𝗿𝗼𝘀𝘀-𝗰𝗵𝗲𝗰𝗸 𝗔𝗜 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀
Mix teams and let engineers see how others are using different models, workflows, and agents across the system.
  
𝟰. 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘇𝗲 𝘄𝗼𝗿𝗸𝗶𝗻𝗴 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀
Most teams stop at the “cool demo”. If one team is already using something that works, turn it into a repeatable asset: documented, versioned, and built into your stack. 

Ship it so the whole org benefits, not just the people who experimented with it.
_________
           
What’s the one thing your engineers actually do every week from this list?
 
#EngineeringLeadership #AIEngineering #AIAgents #TechExecution #StartupLeadership #EngineeringCulture #ProductVelocity

Internal Software for CS, Ops, and Sales is changing forever. 

In 2025, the cost of generating a brand-new UI for your internal process
goes to Zero.

If top LLM models can generate UI in real-time
And often more thoughtful compared to the team of:

• Designer 
• Business Analyst 
• Product Manager

What's the point of putting the upfront cost of building internal software? 

The point is in the Data Modeling and designing software 
from an AI-native way. 

I think it is a 3rd wave B2B Software Leap.

1st Wave - Traditional UI B2B SaaS (Salesforce, Xero, etc.) 
2nd Wave - Fluid Low-code (Airtable, Coda, etc.)
𝟯𝗿𝗱 𝗪𝗮𝘃𝗲 - 𝗭𝗲𝗿𝗼 𝗨𝗜 (𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲𝗱 𝗼𝗻 𝗱𝗲𝗺𝗮𝗻𝗱) 

Think of the strongly structured data stored in the system and 
instead of learning how the system works, 

asking chat to show the latest charts -> allow to make decisions on how to show the data.


We are not there yet, but as a team building B2B SaaS,
We can see the shift is coming.

How to start AI strategy for CEOs?

1. Start with opinion in every department.
Each department is enabled by a model.

Every day, people ask questions and use AI to verify their thinking with a co-pilot.

2. Co-pilot model becomes very good in judgment → Automate.

Make a single working workflow with clear input and output.

Run it on autopilot.
	
3. Chain multiple automations into an agentic workflow.

Have a human validate every step and approve.
	
4. Once approval becomes redundant button-clicking — voila, 
you have an autonomous agent.

#AIstrategy #EnterpriseAI #Fintech #WorkflowAutomation #AutonomousAgents #TechLeadership

taplio