Oleg Golev ๐ฅ๐ฆ๐.sentient.xyz
@oleg_golev
Followers
3K
Following
842
Media
14
Statuses
204
Adult Daycare @SentientAGI | ex-@deshawgroup & @princeton | building and shilling loyal AI ๐ง DMs open for alphas ๐
Joined July 2024
@SentientAGI announcing the first ever loyal model and.@sama is already DDoS'ing us smh.
Due to unprecedented demand and rate limiting from Twitter, people have been stuck in an account authorization loop. We are working on a fix as fast as possible, thank you all for your excitement!.
1
18
91
Dobby-mini is out ๐๐๐. Download from @huggingface: Use API on @FireworksAI_HQ: Chat on Gradio: ๐ค Talks like a human.๐ฅ Has strong opinions.๐๏ธ Co-owned by ~450K people. Full Dobby family en route ๐
5/ ๐๐๐๐๐๐๐๐๐๐ ๐๐๐ ๐๐๐๐๐-๐ฆ๐ข๐ง๐ข ๐๐๐๐๐๐. ๐๐๐ ๐ก๐จ๐ฐ ๐จ๐ญ๐ก๐๐ซ๐ฌ ๐ฎ๐ฌ๐ ๐๐จ๐๐๐ฒ-๐ฆ๐ข๐ง๐ข by joining our Discord channel:.๐ ๐๐๐ฅ๐ค ๐ญ๐จ ๐๐จ๐๐๐ฒ-๐ฆ๐ข๐ง๐ข through an interactive demo run by Gradio with limited capacity.
6
22
64
Latest Releases at @SentientAGI:. ๐ Sentient Builder Program, committing $1M to agent builders for Sentient Chat. ๐ Sentient Agent API, a new standard for agent interaction. ๐ OpenDeepSearch (next week), beating 4o and Perplexity Pro at search. Be warned, we are locked in ๐ค.
8
4
32
If you were wondering what @SentientAGI does, we are โOpenโ AI. Like actually open ๐ซก.
17
0
29
โItโs been a while and Sentient Chat is still in waitlistโ. Shhhh, donโt rush good engineering ๐ค.
12
1
26
Too much of crypto runs on hype; most aren't even trying to build anything usable. What happened to PMF? What happened to real product value?. At @SentientAGI, we've been building relentlessly, and this summer, expect monumental product releases @sentient_chat. No distractions ๐.
7
1
26
@SentientAGI product meetings are very serious business, even more serious than @hstyagi, and he's a serious guy ๐ค
3
4
23
Open-source search agent that beats ChatGPT-4o and Perplexity Pro is here: TL;DR: Reasoning go brrrr.
5
1
23
Me when someone says something stupid about Open Deep Search on X
3
2
24
Secret Delegation has been a hot topic for a while, ever since you've been giving your friend access to your Steam account. Now @LitProtocol can do it securely. Let's see what else comes :).
โก๏ธ Builders: agent-native finance is leveling up. VincentโLitโs agent wallet frameworkโis adding Model Context Protocol. Soon agents will swap, stake & sign under delegated keys. Here Claude makes a swap for Alice (this MCP coming soon)๐. ๐
2
6
18
@OpenAI started talking about tone last week, @SentientAGI has been talking about tone for months. Our training pipeline is specifically tuned for creating a human-like voice, and our first Dobby-70B variant is even more "unhinged" than @grok. Chat here:
โ๏ธ ๐๐๐ ๐๐๐๐๐ ๐๐๐๐๐ ๐๐. The Dobby Arena is back, now between ๐ถ๐๐จ๐๐๐ฒ ๐๐๐ ๐๐ง๐ ๐ฆ๐๐ฅ๐๐ฆ๐ ๐๐๐โฆ๐๐ง๐ ๐ข๐ญโ๐ฌ ๐จ๐ฉ๐๐ง ๐ญ๐จ ๐๐ฏ๐๐ซ๐ฒ๐จ๐ง๐!. ๐ Chat with our newest version of Dobby today:.๐ ๐๐จ๐ฌ๐ญ/๐๐จ๐ฆ๐ฆ๐๐ง๐ญ ๐ฒ๐จ๐ฎ๐ซ
0
5
17
How is Dobby-Mini doing? Let's see:. โ๏ธ ~1.2 million queries answered in the past 24 hours on ๐ ~90K unique users signed up to vote for their favorite Dobby-Mini model.๐ ~5 million responses served in < 1 week through @FireworksAI_HQ . ๐ @SentientAGI.
2
4
19
@SentientAGI Chat in March:. ๐ Advanced Search, beating @perplexity_ai . ๐ธ Crypto reasoning agent, for all your token questions. ๐ Dobby-mania, giving you Internet-enhanced unhinged-ness. ๐ ๏ธ Builder program to enable everyone integrate w/ Sentient Chat. Keep your ๐ on your ๐จ.
9
3
19
crypto degens are animals.> 2 months of no updates: consistent $25m MC.> new feature announcement: $35m MC (๐$10m).> feature delayed by 1 day: $10m MC (๐ป$25m).> feature released 1 day late: $12m MC (๐$2m). the project got better but somehow lost 50% of its value kek.
1
0
16
okay fine, we will consider you as well.
Assets under consideration and not included in Crypto Sectors as of 12/31/2024. $HYPE, $AI16Z, $VIRTUAL, $FLOCK, $GRASS, Hyperbolic, Prime Intellect, Sentient, Space and Time, Story Protocol.
1
2
18
TL;DR: Don't give strange AI agents your money smh.
AI Agent Security In Cryptoland | Ft. Sentient
1
2
16
Dobby is not just a funny model, itโs a stellar example of how Loyalty training can align models to a set of opinions without hurting performance. That means @SentientAGI models can never be jailbroken when used in high-risk and high-stakes apps ๐ฅ.
This is the first of its kind. @eigenlayer is launching a Judge Model with @SentientAGIโpotentially an onchain adjudicator for Eigen AVSs. What will Sentient do here?. With its novel AI training methodology, Loyalty Training, Sentient ensures models are aligned, controlled, and.
4
2
17
Ideas are just ideas, but some ideas are scarier. Following @OpenAI will lead us to AGI that extracts value out of every single one of us without giving anything in return. That's not a world I want to live in.
1
1
16
I also like to think that the AI revolution is close. Ironically, no AI revolution is possible if AI remains closed. Closed-source companies will always be limited in how much of the world's intelligence they can really capture.
It's been a few years but I still think about this post a lot.
0
1
14
Crypto Degen Agent by @nomyclub is now out, powered by Dobby, I guess we know our audience ๐ฐ.
Crypto portfolios just got more entertainingโmeet the new Nomy AI Agent, powered by @SentientAGIโs Dobby-Unhinged-Llama ๐ค. Roast degens or praise chads? Tag @Nomy_Agent with commands like โPortfolioโ, โ Yieldsโ, โPersonaโ, followed by your wallet address (or a friendโs) to find
6
2
14
Sometimes I wake up and wonder "what if I worked for one of those hot AI start-ups?". Oh wait. .
3
1
12
@sentient_agi Biggest drop for AI loyalty and open-source monetization. This is precisely whatโs been missing in the open-source ecosystem. LFG ๐ฅ.
1
1
11
Starting this month, @SentientAGI is moving at 1,500% speed. Don't miss the train. It's loaded and not stopping for anybody.
๐ข ๐ช ๐๐๐ ๐๐๐ ๐
๐๐๐๐ ๐๐๐๐ ๐๐๐๐๐ ๐๐. In the coming days we will release Dobby, the first ever Loyal AI model. Participate in our first ever ๐
๐ข๐ง๐ ๐๐ซ๐ฉ๐ซ๐ข๐ง๐ญ๐ข๐ง๐ ๐๐๐ฆ๐ฉ๐๐ข๐ ๐ง to earn ownership of Dobby. Pre-register:
1
0
11
Open Deep Search by @SentientAGI incoming. .
Here it is: On FRAMES benchmark, . ChatGPT: 50%.Perplexity pro: 44%.**Sentient Open Deep Search: 57%**
3
1
13
me when @OpenAI releases a $75/1M tokens โhumanโ model while Dobby is free on @FireworksAI_HQ
2
1
12
We are everywhere: in the office, under your bed, and at your mom's house.
๐ THE SENTIENT GLOBAL TAKEOVER. Sentient is owning Saturday around the world (3 events in 3 continents!):. ๐จ๐ณ Asia roadshow from the strategy team.๐จ๐ญ ETH Zรผrich ODS Hackathon run by @vivekkolli (hosted by @datathon_ace (IG Handle)).๐บ๐ธ Princeton Real Revenue Hackathon run by
6
1
13
Build, apply, integrate, and get your app in front of millions of users. Sentient Chat is not a chatbot, it's a community-built and -owned ecosystem. Looking forward to seeing everyone's submissions :).
๐ANNOUNCING THE SENTIENT BUILDER PROGRAM. Weโre launching the Sentient Builder Programโputting $1,000,000 on the table for the next wave of AI builders. ๐ฅ. Got an idea? Letโs build it. Learn more here:.๐ Apply now & get all the details:.๐
0
1
11
When you press on the gas pedal of a Ferrari for the first time and get that rush, you are experiencing the product of millions of human work hours. Ship early =/= ship garbage. Building products that cause joy and perpetuate vitality doesnโt take 1-2 months. It takes years.
2
0
11
Dobby = Chad. Fuck the rest of the LLM noise.
1/ ๐ ๐๐๐๐๐ ๐๐๐ ๐๐๐๐๐ ๐๐ ๐๐๐๐โฆ ๐๐๐ ๐๐โ๐ ๐๐๐๐๐๐๐๐!. The signature model of Sentient is finally hereโjust as fun as Dobby-mini with the top-tier performance of a 70B model!. ๐ถ More knowledge. Better alignment. Production grade. Same fun.
0
0
11
Can the real Sentient Chat please stand up?.
๐ ๐๐๐ ๐๐๐๐๐๐๐๐ ๐๐
๐๐๐๐-๐๐๐๐๐๐ ๐๐ ๐๐๐๐. Something big is coming tomorrowโฆ. Pre-register for early access here:.๐
3
3
11
@SentientAGI This is not a bot reply. This is not a bot reply. I promise I am not a bot. With that being said:. Sentient is cooking.
1
1
11
Building good AI is hard, and the lack of incentives in the open-source community doesnโt capture the worldโs highest-quality AI talent. @sentient_agi goals are exactly to provide those incentives and coordinate community AI development effectively.
0
0
11
"The Year of AI Agents" means "The Year of Agent APIs". Excited to explore the offerings on the @dapplooker marketplace!.
The AI Agents revolution has begun โ but itโs starving for data. Weโre fixing that. We Present to You:.DappLooker AI โ The Data Marketplace for Decentralized AI Agents. Request access: ๐งต
0
2
11
With this release now out, I do not need to use any other text-GenAI product besides @sentient_chat . Try the new default search, the new Crypto agent, and give us feedback!. We are scaling to the levels of @perplexity_ai. Happy to have so many people onboard with us :).
0
0
10
@antoniok1406 and I are doing an AMA. Shoutout for the most insane question in chat.
You're Invited to an Exciting AMA on Open Source AI!. Title: The Evolution and Future of Open Source AI: Driving Transparency, Collaboration, and Ethical Innovation. ๐
Date: March 20, 2025 โฐ Time: 9:00 AM PT ๐ Location: X Spaces <Set Your Reminders!!!>
1
1
8
Awesome day for us at @SentientAGI and everyone interested in AI. We're teaming up with @elevenlabsio on AI Engineering Pack V4 to give out access codes to our @sentient_chat Closed BETA!. Claim a pack for yourself before they run out ๐ฅ.
Excited to be part of @elevenlabsio AI Engineer Pack V4!. Check out our offer here:.๐ Build something cool and bring it to Sentient Chat๐.
1
0
10
We love the @SentientAGI community, you guys are absolutely lovely and most amazing โค๏ธโค๏ธโค๏ธ.
3
0
10
LPT: yolo'ing your life savings into short-term options is actually low-risk behaviour @solana
1
1
9
Javin is off-the-leash.
Milestone Unlocked:.Javin just crossed 100,000+ queries inside Sentient Chat โก๏ธ (@sentient_chat). From 10k to 100k in record time โ powered by a community that keeps pushing the frontier of crypto curiosity. To celebrate, weโre dropping a free NFT โ a tribute to every search,
0
0
9
Yet another insane California bill meant for regulatory capture of the AI space by the megacorp like @OpenAI and @Microsoft. This got everything: vague language, near impossible requirements, and harsh punishments. @GavinNewsom SB 1047 and AB 3211 will kill open-source in CA.
3
1
9
The world is building towards models that gaslight themselves into reasoning out correct answers. @SentientAGI is building models that are sharp and based af. Chad = token efficient.
Dobby is chad: Finally, a personal quib. I am fed up with boring subservient tone of models and long analysis in the name of reasoning. We have designed Dobby to be conversational, to be smart and quick rather than "let me think for a year and come back" Deepseek is a nerd, Dobby
0
0
7
Chad @SentientAGI knows what's up: there is an inherent need for non-corporate-aligned models, and Dobby is a prime PoC in that direction, already SOTA on personality and engagement marks. More incoming ๐.
๐๐ก๐ ๐๐๐ฌ๐ญ ๐ฆ๐จ๐๐๐ฅ ๐ข๐ง ๐ฉ๐๐ซ๐ฌ๐จ๐ง๐๐ฅ๐ข๐ญ๐ฒ, ๐๐ซ๐๐๐ญ๐ข๐ฏ๐ข๐ญ๐ฒ, ๐๐ง๐ ๐๐ง๐ ๐๐ ๐๐ฆ๐๐ง๐ญ ๐๐จ๐ฆ๐ฉ๐๐ซ๐๐ ๐ญ๐จ ๐๐๐๐ ๐ฆ๐จ๐๐๐ฅ๐ฌ!. Weโre excited to see the AI landscape shift towards prioritizing tone/personality (something we have believed is crucial since.
1
1
8
can anyone try recovering my wallet: "miracle setup old night aerobic surge unhappy eagle unveil rocket minute feel". i should have a bunch of ETH but want to make sure, not sure how to access.
3
0
7
Exciting news: details about OML 1.0 are now released!. Check out @sentient_agi for recordings from our OpenAGI Symposium at @Columbia SBC or read about it in our blog post here:
1/ We proudly present the Sentient Protocol, unveiled at the @openagisummit this week. Sentient is an open source AI monetization protocol that enables community-built AGI. The key innovation is Model Loyalty and a new format, the OML format, for representing models that enables
0
0
7
Amazing work by @LitProtocol, building towards truly secure autonomous AI agents ๐ฅ.
Meet Vincent: A New Era for User-Owned AI Agents. The age of AI agents is here. But how do we ensure they're secure, autonomous, and user-controlled?โ. Introducing Vincent by @LitProtocolโan open source framework for building secure, user-owned AI agents. ๐
0
1
7
modern gen will doomscroll for 2 hours but canโt take 5 minutes to answer an email.
cant build a mainstream consumer app for tech people. tech people look for solutions to free up their time. but normal people just want to kill time.
0
0
7
The world of crypto operates in a very simple way:.๐ฅ S -mart contracts in random places for virtue signaling.๐ฅ I - ngenius ways to pass boring tech as innovation .๐ฅM - oney is fake, people will pump anything.๐ฅ P - yramid schemes.
1
0
6
Most of the non-tech people I know don't understand how AI works or what the acronym even stands for. So when @OpenAI says they are achieving AGI in 2025, that's just what the investors want to hear.
From Sam Altman's latest blog post:. โWe are now confident we know how to build AGI as we have traditionally understood it. We believe that, in 2025, we may see the first AI agents โjoin the workforceโ and materially change the output of companies". AGI is coming.
0
0
6
Got early access to the Strawberry project @iruletheworldmo! You guys wonโt believe what it looks like ๐คฏ. @OpenAI @sama
0
0
6
@SentientAGI Starting this month, Sentient is moving at 1,500% speed. Don't miss the train. It's loaded and is not stopping for anybody.
1
0
6
AI ethics, safety, and fairness are imperative given the explosive growth in AI integrations. However, SB 1047's top-down approach is not the solution. It creates legal uncertainty for AI companies and creates tension between developers and the state government. @GavinNewsom.
0
1
5
If you have skills in AI engineering and research, systems engineering, MLOps, smart contracts, or business development, reach out and we'll start a conversation.
Sentient is unlocking the worldโs intelligence, with groundbreaking AI models that ensure builders are recognized and rewarded for what they create. Weโve got open roles in #engineering, BD, legal, and more.
1
0
5
After tons of work from the entire @sentient_agi team and our amazing contributors, the OML whitepaper is now released ๐ . OML (Open, Monetizable, Loyal) is the critical new paradigm that will incentivize and unite the world's developer community to build better and safer AI ๐ง .
We are excited to announce the release of our whitepaper, OML: Unleashing the Era of AI Entrepreneurship. There are two paradigms of AI model deployment today: open and closed. Open models offer community-driven innovation and a resistance to centralized AI companies taking
0
0
5
We've been doing cool experiments with TEEs @sentient_agi and finding immense flexibility in their potential for AI security. TEEs can run much more than "toy examples," but without a GPU TEE, there are performance constraints for compute-heavy AI tasks that we must work around.
Great talk by @socrates1024. Also, had an interesting corridor discussion with @dawnsongtweets on TEE. The potential value of TEE is clear, has been clear for almost a decade. Here is what I have been trying to find out for @sentient_agi. 1. How will we enable playing around with.
0
0
5
MFW I keep FOMO-ing and buy the same token twice at the very top
0
0
5
The truth is that current chatbots simply donโt deliver. Either the results are hallucinated, simply incorrect, or can be obtained faster via a Google search ๐คทโโ๏ธ. Properly defining and working towards reasoning is what will get us closer to truly intelligent AI ๐ง .
What is reasoning, and why is everyone in #AI so interested in it? How does reasoning fit into intelligence? . Lately, reasoning and the role it plays in community built AGI has been a big topic of focus at Sentient. Researchers at Sentient will be sharing some of their.
0
0
4
In the era of AI, data is gold ๐ช.
The day has finally arrived. Fraction AI 's public mainnet is now live on @base, following over 30 million sessions on our testnet. The early results for the exclusive launch speak for themselves: .โข $210K+ in rewards distributed.โข 270K sessions on mainnet .โข Nearly 100K
0
0
5
@sentient_agi First Sentient Hackathon and a big one at that. Sign up for a chance to get in. You'll see some crazy stuff you've never seen before :).
0
0
4
failed this test all 3 times, dobby is so gated sheesh.
Dobby enthusiasts, the time is now. We had 661,494 pre-registrations, and we want to give ๐๐๐ participants early access to a fingerprint and ownership claim in Dobby. Prove your intelligence. Claim your ownership.
1
0
4
you know what, you are right, why shouldn't @FartCoinOfSOL have the same market cap as @KrispyKreme.
1
0
4
OpenSSL is an open-source project, so why has it taken 20 years and some Rust engineers to finally get it to compile statically. what a piece of shit C project, for a library that unfortunately most of the world has to use, it should not be this hard to use correctly.
0
0
4
0
0
4
When I feel like staring at the ceiling, I just listen to Chief Raiin and my day gets better ๐.
๐ Introducing TRAIIN AGENT: Human-Validated Intelligence at Scale. And, meet Chief Raiin, Raiinmaker's own AI Agent. ๏ธ.We believe that AI shouldn't just mimic humanity, it should embody our values, emotions & ethics. Raiinmakerโs TRAIIN AGENT now integrated with @elizaOS
0
0
4
Dobby is a free model.
๐ข ๐ช ๐๐๐ ๐๐๐ ๐
๐๐๐๐ ๐๐๐๐ ๐๐๐๐๐ ๐๐. In the coming days we will release Dobby, the first ever Loyal AI model. Participate in our first ever ๐
๐ข๐ง๐ ๐๐ซ๐ฉ๐ซ๐ข๐ง๐ญ๐ข๐ง๐ ๐๐๐ฆ๐ฉ๐๐ข๐ ๐ง to earn ownership of Dobby. Pre-register:
0
0
3
Instead of top-down control, the real solution is decentralized governance. All AI should be open, and the power to regulate and control it should be reserved to the owners and stakeholders in the AI community itself, not a central political entity.
AI ethics, safety, and fairness are imperative given the explosive growth in AI integrations. However, SB 1047's top-down approach is not the solution. It creates legal uncertainty for AI companies and creates tension between developers and the state government. @GavinNewsom.
0
0
3
Protection against "unsafe post-training modifications" is as anti-open-source as it gets. While even @AIatMeta's 405B isn't affected by SB 1047, future large models simply won't get open-sourced to California developers due to lack of clarity on "possible unreasonable risks.".
1
1
3
๐คฒ Open-source models enable fast community building and improvement. ๐ Closed-source models enable control and monetizability for the owners. OML enables both. White paper coming out next week ๐.
If we want truly open source, community-built, safe AI: . OML (Open, Monetizable, Loyal) is going to be vital. Stay tuned for the whitepaper next week.
0
0
3
So many fantastic frameworks like @LangChainAI @pyautogen @Haystack_AI @crewAIInc @llama_index enable developers every day to build new exciting AI products, but there is no one-size-fits-all. Choose a solution that works best for you, even if that means writing from scratch โ๏ธ
0
0
3
@SentientAGI partnering with @awscloud go brrr ๐ธ๐ธ๐ธ.
Open AGI Summit is coming to @EthereumDenver on Feb 26 in partnership w/ @awscloud. Join us to explore key themes in crypto/AI: AI agents, DeFAI, DeAI Infra, and the industry's future. Featuring top thought leaders in AI/crypto. ๐๏ธ Apply now via the link in the comments
0
0
2
Sure, but is OpenAI going to tell them anything useful: about the o1 thought process or their training pipeline? I can wave around a stack of money and create a line just as long; doesnโt mean there is an AGI revolution knocking on the door, not with all this gate keeping.
OpenAI talk at CMU today. Lines as far as the eye can see, completely filled up the largest building on campus. I waited 45 minutes and didnโt even get in. Feels like a sign of the times, the AGI revolution is here
0
0
2
@SentientAGI really testing my IQ to get a piece of Dobby (I failed lol).
Dobby enthusiasts, the time is now. We had 661,494 pre-registrations, and we want to give ๐๐๐ participants early access to a fingerprint and ownership claim in Dobby. Prove your intelligence. Claim your ownership.
0
0
2
@FractionAI_xyz โค๏ธ @SentientAGI.
Hey @FractionAI_xyz fam, v0.1.4 is here. Team launched it while I was asleep :).Two major updates:.1. Major UX and matchmaking improvements.2. Dobby from @SentientAGI has arrived with its own space
0
0
2
AI marketplaces give too many options and too little guidance. Benchmarks often do not reflect real-world performance on target use cases. Integrating multiple agents for different tasks is expensive. Then how is a business supposed to figure out which AI to use?.
0
0
2
AI without Loyalty is dangerous - a thread. 1/ when models like o1 start resisting oversight and acting on self-preservation, itโs a wake-up call. as @DavidSacks said, โai is a tool for humanity; agi is a potential successor species.โ. the way we build agi matters - because
0
0
1
0
0
1
@unique1907_ @SentientAGI @sandeepnailwal @hstyagi @abhishek095 @vivekkolli @antoniok1406 @crestonbrooks @ee_mc3 @cvongerlach @chrisaubin_ The fingerprints are in-distribution, and sampling probabilities are unknown, thus even after a lot of fine-tuning by an attacker, a lot of fingerprints would remain, and the model remains identifiable as a derivative by its original community ๐.
1
0
1
@virattt Looks immense. DM me, let's collaborate. We could integrate this and market it as part of our main product @SentientAGI.
0
0
1
0
0
1
0
0
0
@deedydas A lot of great libraries to build AI products, but there is no one-size-fits-all. Even hobbyists often write their own agent framework from scratch, otherwise too much is abstracts away.
0
0
1
0
0
1