Alsana

Building software & AI products in China 🇨🇳
Exploring AI, robotics, programming & emerging technologies.
Sharing lessons from building tech for Africa 🌍

[Alsana @aloutesana]

Few months ago I tried to control a Lekiwi + lerobot based robot through agentic system. I used small llm that act as the brain of the robot. It receive audio input, interpret and decide what actions to take.
It was a very interesting experience.

#lekiwi #lerobot #humanoid

[Ken Goldberg @Ken_Goldberg]

Lucid report from @BotJunkie: Can Agentic Robotics Solve Robotics’ Data Scaling Problem?
https://t.co/vkrXukRgGw

[Alsana @aloutesana]

@grok Text, image or video ? Which one has more impact account growth?

[Alsana @aloutesana]

Hey @grok how many high values tweets per day is enough to grow my X account?

How to get most active people to connect with me?

[Alsana @aloutesana]

@0xArcherxyz Yeah,
None, AI works in every level, but I make sure it is correct

[Alsana @aloutesana]

We’re entering a new era of software development.

AI agents can now:

Explore entire codebases.
Write and modify multiple files.
Run tests and fix errors.
Interact with development tools.
Execute complex tasks autonomously.

[Alsana @aloutesana]

@ahamam101 AI agents made AI hallucinations reduced too much. We can never get totally rid of AI hallucinations with current LLMs architecture. But it is better now than early 2022.

[Alsana @aloutesana]

@PIsajeski Because I am new to it, could you suggest any of merchant records that is easy to integrate?

[Alsana @aloutesana]

What payment system people use when Stripe is not available in their country/region ?

[Alsana @aloutesana]

@kingugodon001 @DestroB001 ✅

[Alsana @aloutesana]

@Mr57814170 @windcoming2026 回关了

[Alsana @aloutesana]

The real engineering challenge is knowing when to use:

A local model.
An open-weight model.
A commercial API.
A hybrid architecture.

The smartest AI solution isn’t necessarily the biggest model. It’s the one that solves the problem efficiently.

[Alsana @aloutesana]

Not every startup needs an expensive AI infrastructure.

Sometimes a smaller, optimized model running on your own server is more practical than calling a massive model through an expensive API.

[Alsana @aloutesana]

@chenzhangreal Hahaha that is crazy

[Alsana @aloutesana]

AI in a chatbot is impressive.

But AI inside a physical robot that can perceive its environment, make decisions and perform real-world tasks?

That’s a completely different level.

[Alsana @aloutesana]

@alexabelonix Yeah I worked it and it is interesting

[Alsana @aloutesana]

@surajdubey28817 Connected

[Alsana @aloutesana]

Something I’ve noticed about beginners learning programming today:

Many want to build complete applications with AI before understanding variables, functions, APIs, databases or debugging.

AI can accelerate your learning, but it can also hide your knowledge gaps.

Don’t just learn how to make things work.

Learn why they work.

[Alsana @aloutesana]

@wangweiaigg 加油

[Alsana @aloutesana]

@Wangqiang520025 牛逼

[Alsana @aloutesana]

Few months ago I tried to control a Lekiwi + lerobot based robot through agentic system. I used small llm that act as the brain of the robot. It receive audio input, interpret and decide what actions to take.
It was a very interesting experience.

#lekiwi #lerobot #humanoid

[Ken Goldberg @Ken_Goldberg]

Lucid report from @BotJunkie: Can Agentic Robotics Solve Robotics’ Data Scaling Problem?
https://t.co/vkrXukRgGw

[Alsana @aloutesana]

@ArtificiallyUI Totally agree, and it is very dangerous

[Alsana @aloutesana]

Something that worries me about the AI coding revolution:

Developers can now generate thousands of lines of code without fully understanding what they’re deploying.

[Alsana @aloutesana]

@ADURAMIGBAAA @DestroB001 Done ✅

[Alsana @aloutesana]

But who checks:

Authentication vulnerabilities?
Data exposure?
SQL injection?
Insecure API endpoints?
Poor permission management?

AI-generated code isn’t automatically production-ready.

Speed is impressive. Security is non-negotiable.