Ronald
@ronald_obj_ai
Followers
56
Following
179
Media
21
Statuses
271
CEO & Founder of ObjectiveAI
Joined July 2025
"The construction of LLM ensembles seems under-explored." Andrej Karpathy thinks it's under-explored. I've been exploring it nonstop for about 7 months. I started out trying to solve a problem. Now it's a technical marvel and that problem is irrelevant. Apply it bigger.
As a fun Saturday vibe code project and following up on this tweet earlier, I hacked up an **llm-council** web app. It looks exactly like ChatGPT except each user query is 1) dispatched to multiple models on your council using OpenRouter, e.g. currently: "openai/gpt-5.1",
0
0
2
That's me! 🥳
0
0
4
I think the real trick is to be friendly, approachable, and put in a lot of work. Be a good person, utilize your talents, and work to make people's lives better. This is the next step in my plan.
0
0
2
My native tongue is typed data. When they added JSON Mode to LLMs, that's when they gave it the ability to speak my language.
0
1
3
I'm gonna write a big old blog post about how "tunable quantified AI preferences" works. I should have done this before now but, well, you know how it is, and you know what I'm gonna say: Life is about learning and improving.
0
1
3
It ain't easy selling tunable quantified AI preferences across possible responses to an arbitrary prompt. First, I will try contributing to AI companies' open source repositories to get renown with them. Failing that, integrate it with crypto betting on AI character cards lol.
0
1
3
I've been working on ObjectiveAI, which provides tunable quantified AI preferences across possible responses to arbitrary prompts. I'm thinking about the road to mass application. I wonder if the answer lies somewhere between https://t.co/dA9mgenp4j and https://t.co/WsuJeRP5hL.
0
0
4
How about, provide ObjectiveAI with some preferences data, such as which links a user clicks after searching the web, and we'll give you back a Score Model which simulates the user? Then you can see which links they would click.
1
1
1
ObjectiveAI MCP server is on hold, gonna try to make it work with this bubble startup thing first
0
0
2
Hi, I'm Ronald, CEO at ObjectiveAI! We provide quantified & tunable AI preferences across possible responses to arbitrary prompts. Put simply, we show you how much AI likes things, and let you customize what AI likes. I look forward to showing how useful this can be!
0
2
6
The world needs ObjectiveAI. Quantified AI preferences can change so many things for the better. These preferences are adjustable within our system. We get AI's true preferences. Not by asking what it likes. But by reaching into its digital brain and ripping out the truth.
0
1
4
when you convince someone to quit their job for your unfunded early stage startup
0
0
3
I think it's time for Score Models to support images, videos, and files. They can only score text right now. Also, it's time to make a 100% OpenAI-compliant LLM API which does multichat->score and takes the winner. (That will likely win most LLM benchmarks in existence.) 😃
1
0
1
ObjectiveAI is recruiting. We are looking for a Designer Co-Founder to design the UI for a developer-facing product. Technical proficiency is important. We are also looking for Technical Advisors (15 minutes/week). The only requirement is prior FAANG experience.
0
1
3
In order to bootstrap your business from absolute nil, you need 0 fixed costs. Design your infrastructure to scale from $0 to $N. Investment is acceleration. It should not be mandatory. Unless you are connected, plan on getting nothing. Map out success from nothing.
0
0
1
Life is about learning and improving. They think you can't learn and improve. Maybe it's because they can't. They're stuck. Show them what life is about.
0
0
2
The winner of Humanity's Last Exam is now live on ObjectiveAI Apps ( https://t.co/Y5iac9SC4i). It beats Gemini 3.0 Pro. I'll update it over time so that this app is perpetually the winner. It won't get the visibility it should. That'd be too easy. We're just getting started.
0
0
3