Luka Rautar

Growth Advisor - @Astra__AI

[Astra AI @Astra__AI]

ASTRA LEAK: We know we won't be the most popular Astra benchmark today, but if you are studying and need the help of AI, we are here.

[Luka Rautar @finomehanik]

@razib_ul47671 MAP

[Klemen Selakovic @klemenselakovic]

Wild stat: @Astra__AI now processes 1/3 of the AI tokens that a $150B public software company does.

Based on $CRM's own earnings disclosures, Salesforce processes around 3T tokens/month. We're past 1T/month — and accelerating fast.

Every major technology platform shift reshuffles the deck. Startups win because we can move at full speed and take real risks — we have no legacy business to defend from disruption.

Incumbents aren't adopting AI fast enough. That's the opening.

[Klemen Selakovic @klemenselakovic]

At @Astra__AI we consume over 1 TRILLION tokens per month. In our experience, the key to serving AI at scale is to own your HARNESS.

While we don't compromise on the quality of our outputs, we continue to invest significant effort to maintain our ability to serve our customers at scale at an unprecedented growth rate.

Because we own the harness switching between model providers (open-source and closed-source), has become a task of hours for us. If we spent all the tokens on Fable, our cost would rise to over $18M per month.

Open-source, closed-source, and diversity in AI model providers are what keep the whole AI ecosystem healthy and functioning.

[Luka Rautar @finomehanik]

Similar philosophy and a team of doers is was what got us to $25M ARR over 10 months at @Astra__AI

[Alfred Lin @Alfred_Lin]

https://t.co/t5TW3Rbzlj

[Luka Rautar @finomehanik]

Can't agree more 👇

[Klemen Selakovic @klemenselakovic]

Bootstrapped from zero to +$25M in revenue in just 2,5 years at @Astra__AI . A few honest reflections:

Venture capital is like rocket fuel. But rocket fuel only helps if you already know where you're pointing.

Early on, we didn't. We were still learning what customers valued, what they'd pay for, what to kill. We really wanted to focus on our customers and nailing the product experience to perfection before we entered the scaling ladder. Being bootstrapped meant we could change our minds faster and without distraction; no next-round story to protect.

Many friends who raised funds early were trapped focusing too much on the fundraising side and meeting the targets and goals they set up years ago, even though they might not make that much sense anymore, as the business, especially in the early stages, is highly fluid and you need to change your mind often when facts change. Even subconsciously, some gripped onto narratives and past moves they knew deep down were wrong but didn't want to explain to their investors. It's human nature.

Raise when capital is the biggest constraint on your business. Until then, fully focus on understanding your customers and their problems.

[Klemen Selakovic @klemenselakovic]

Investors are focused on the AI model layer, while the AI application layer is EXPLODING.

This is our hockey stick growth in the AI application layer. At Astra AI, we now have +$25M in ARR, up 25x YoY. We did this bootstrapped organically without a single investment.

During this time, our gross margin has expanded even further. The value of AI is diffusing through the whole AI stack: chips, data center infra, AI model layer, and AI application layer.

[Klemen Selakovic @klemenselakovic]

We are serving more than 11M users across more than 194 countries with our AI product, and here is the reality of running an AI business at that scale.

I often see the debate of who will capture most of the economic value when it comes to AI models: will it be the SOTA frontier models or »second layer« models?

This year was the first year in which AI models have achieved quality thresholds for certain use cases, so that each future incremental improvement in model capabilities doesn't automatically justify switching to the SOTA. When people run pilots, SOTA always feels like the most attractive option, but in reality, once you move into production workloads, serving AI at scale becomes another issue.

Serving intelligence at scale has become key. It's what separates experimental prototypes from real-world AI solutions capable of serving millions of users.

[Klemen Selakovic @klemenselakovic]

The AI model is far from being the moat in the application layer. My company is in the AI application layer; we have grown from $1M ARR to +$25M ARR in a year.

Model progress has helped us a lot, but the AI harness, usage patterns, and distribution are even more critical and build the true moat in the AI application layer.

[Tjaž Silovsek @silovsek]

Spoke at @vercel Ship in Berlin last week.

Our story: a lean team from Slovenia, 10M+ students, bootstrapped all the way.

Everyone at Ship was there to build. Left recharged.

Back home with one goal: 100M students. Let's ship. 🇸🇮

[Andrej P. Skraba @SkrabaAndrej]

$25,000,000 in ARR, up roughly 25x in 12 months.

We did it by staying extremely focused:

1. Build a product students actually use
2. Move fast across product and growth
3. Localize deeply across markets
4. Keep the team small, ambitious, and accountable

Our @Astra__AI mission is simple:

Build the most accessible, personal AI tutor in the world.

We believe every student should have access to world-class learning support, regardless of country, school, background, or income.

And this is still the beginning.

We did this in 3 years, bootstrapped, profitable, with a small team in Slovenia.

No investors. No permissions needed.

Now we want to build the company that defines how hundreds of millions of students learn with AI.

Thanks to this network for all the support during our journey. ❤️

[Andrej P. Skraba @SkrabaAndrej]

This weekend, @Astra__AI surpassed 10 million registered users.

It all started with a simple message from @klemenselakovic: "Want to build Astra AI?"

Three years later... 10 million registered users. 194 countries. One of the fastest-growing startups in Europe.

0€ in VC funding.

Bootstrapped from Slovenia. 🇸🇮

What still blows my mind is how we got here.

No big launch, no investors, no safety net.

Just a small team shipping relentlessly, week after week, while everyone said bootstrapping at this scale was impossible.

We're building the best AI tutoring product, accessible to anyone, anywhere in the world.

Next stop: 100M.

[Andrej P. Skraba @SkrabaAndrej]

Včeraj zvečer smo v @Astra__AI dodali ustni izpit.

V samo parih urah so uporabniki z našim AI tutorjem govorili 8.796 minut. To je približno 147 ur pogovorov, skoraj en poln mesečni delovnik.

Prej si za to potreboval starša, sošolca, inštruktorja ali učitelja ...

Try it out 🤳

[Andrej P. Skraba @SkrabaAndrej]

Predstavljamo @Astra__AI Lab ⚗️

Ta nova sekcija bo všeč predvsem učencem, z Astra AI Lab pa želimo pomagati tudi učiteljem in profesorjem. Vse prikazano v threadu je na voljo brezplačno! 💙

[Andrej P. Skraba @SkrabaAndrej]

Učinkovito učenje zahteva načrt, strukturo in vsakodnevno spremljanje napredka.

S pripravo na test predstavljamo največjo posodobitev @Astra__AI — učencu ustvari personaliziran učni načrt z enim glavnim ciljem: doseganje odličnosti. Več v videu spodaj.

Hvala za vsak repost 💙

[Luka Rautar @finomehanik]

@profgalloway @Elan_Skis from Slovenia in the background 💪

[Luka Rautar @finomehanik]

@toplovodar Propad Rima v 4K definiciji :)

[Luka Rautar @finomehanik]

Help make it happen for KAISR Original: The Ultimate Inflatable Air Lounge https://t.co/wE56d30MTV #indiegogo via @indiegogo

[Zweitakter @cvajtaktar]

Begunci gor/dol... Urzah za meh raztegnt je pa bil #toplovod