Rest in Peace to a true Nigerian and African legend, the great Olu Jacobs (MFR). 💔
Your voice, presence, talent, and unforgettable performances touched generations and helped shape African cinema. You were more than an actor; you were an icon whose legacy will live on in the hearts of millions.
Thank you for the memories, the inspiration, and the years of excellence. The Lion of Lufodo has taken his final bow. 🕊️🙏🏾
Rest peacefully, legend. Your legacy lives on forever.
123
BigNelson retweeted
OMOYELE @SOWORE FOR PRESIDENT OF THE FEDERAL REPUBLIC OF NIGERIA.
THIS IS THE BEST TIME TO CHOOSE THE BEST.
#Sowore2027 #TakeItBack
Another insightful @axisrobotics Hub onboarding (pre-training) session today.
I picked up a lot of new insights, especially around better robot control and approaching tasks with more confidence.
We also covered practical examples, from opening and closing objects to troubleshooting what to do when the “Task Completed” dialog doesn’t appear even after successfully completing a task.
Every session makes the learning process more practical and easier to understand.
One session at a time. One task at a time. Getting better at contributing to the future of Physical AI.
4
130
Tonight’s @axisrobotics Nigeria Space was epic.
If you missed it, go play the recording and catch up on the latest updates from Axis.
4
139
BigNelson retweeted
🚨Come Join Us!
Pleased to invite you to join us tonight for another edition of the @axisrobotics NGN community hangout where the community meets to share updates, ask questions and connect with others.
📆: Friday, 14th Aug | 08:00 WAT
🔗: xitter.gq/i/spaces/1MJgNbW…
Set a reminder, invite a friend and don’t miss it
We had another great session on the @axisrobotics Discord.
We learned a lot, especially how to properly complete post-training tasks.
One key takeaway: it’s better to complete more unique tasks than to repeat the same task multiple times, as repetition can actually lower your score.
Quality beats quantity.
3
86
I’m excited to have participated in another insightful onboarding session today with @jjk73310 at the @axisrobotics NG Stage. Looking forward to learning, contributing, and growing with the community.
3
115
The interesting part of @axisrobotics isn’t just combining robotics with crypto.
It’s about using crypto primitives for coordination, verification, ownership, and incentives around the data and intelligence that robots need to operate across industries and real-world applications.
1
121
BigNelson retweeted
This is the first @KaitoAI Katalyst campaign 😱
📽️ AI video competition
🏆 Prizes from $70 up to $10,500
⚠️ When you register you have 100 credits to create a short video. Use them wisely, otherwise you need to buy a subscription (starting from $19 monthly)
My thoughts: I think not too many creators will participate since 100 credits is not too much and you likely need to pay a subscription, so competition will be lower!
$70,000 is going to AI video creators!
The Next Scene Challenge from @newtake_kr is live on Kaito Studio, our first opportunity with one of the leading video-gen AI brands:
- 3 ways to enter
- 238 winners across X and Instagram
- Running from Aug 14 to Sep 4
Here's how to get started ↓
𝐃𝐢𝐝 𝐲𝐨𝐮 𝐤𝐧𝐨𝐰 𝐭𝐡𝐚𝐭 𝐞𝐯𝐞𝐫𝐲 𝐩𝐞𝐫𝐬𝐨𝐧 𝐰𝐡𝐨 𝐣𝐨𝐢𝐧𝐬 𝐀𝐱𝐢𝐬 𝐜𝐡𝐚𝐧𝐠𝐞𝐬 𝐰𝐡𝐚𝐭 𝐭𝐡𝐞 𝐧𝐞𝐱𝐭 𝐫𝐨𝐛𝐨𝐭 𝐩𝐨𝐥𝐢𝐜𝐲 𝐜𝐚𝐧 𝐝𝐨?
Robot learning policies fail when the environment, lighting, or object position falls outside what the training data covered. The Open X-Embodiment research found that training across more robot embodiments improved model generalization more than adding more data from a single embodiment, source diversity outperformed raw volume from one lab. Axis is built on the same principle, except the diversity comes from contributors and not embodiments.
130,000+ people have produced 3.1 million trajectories across 1,600+ tasks on the platform as ofAugust 13, 2026. Each trajectory is a timestamped recording, robot joint states, object positions, control actions, task metadata collected by people from different places with different natural ways of approaching the same manipulation task.
Someone who’s never touched a controller moves through a pick-and-place differently than someone who has. Across 130,000 contributors, those differences accumulate into what ends up in the training data instead of the interaction patterns of one lab’s operators.
The @axisrobotics paper on arXiv quantified what this produces: continual pretraining on the community-collected dataset improved 0.5’s task success rate by 5.8% over baseline and outperformed models pretrained on RoboCasa365 by 37.3%. Same architecture, different data source.
In February 2026, 18,000 active contributors produced close to 100,000 trajectories across 27 task types in five days.
Scale AI’s dedicated robotics data division purpose-built facilities, full-time paid staff, specialized hardware collected 100,000 hours in a full year. Axis runs entirely in a browser through MuJoCo compiled to WebAssembly, so the physics simulation runs locally on a contributor’s machine with no GPU and no setup cost.
The platform processes 10,000 valid trajectories per hour at current activity levels. A single traditional teleoperation rig costs $40,000+ to get operational.
The contributor count also matters specifically in post-training. After a base policy is trained on a task, Axis reopens it, the policy attempts the task on its own while contributors watch, and when it fails, contributors take over, correct it, and hand control back.
Their documentation describes these correction sessions as the highest-value data the system collects because they capture precisely what the current policy can’t handle. More active contributors means more people watching trained policies run simultaneously, more failure cases identified and corrected, faster iteration on whatever the model still gets wrong.
Every trajectory gets a unique Data ID recorded on Base alongside the task it came from and the contributor who produced it. 2.1 million on-chain records behind 3.1 million trajectories. That data has already trained a real robot and not a simulation, but physical one.
And the next policy iteration depends on how many people show up to correct the current one’s mistakes, and right now that number is still growing.
1
105
Week 2 - @axisrobotics Thread
Robotics is advancing quickly, but behind every capable robot is something most people don’t see:
Data.
High-quality data is what helps robots understand tasks, environments, objects, and how to interact with the physical world.
🧵
Robotics companies need huge amounts of diverse and reliable data to train and improve their models.
A robot might work perfectly in a controlled environment, but the real world is unpredictable.
Different objects, environments, movements, lighting, and situations all require more data.
This is where Axis Robotics comes in.
Through its data engine, Axis helps provide the high-quality robotics data that companies need to develop better Physical AI systems.
Instead of relying on a limited source of data, Axis can leverage a distributed community of contributors to generate and verify useful robotics tasks.
And this is where the community becomes extremely important.
The bigger the contributor network becomes, the more diverse the data can become.
More contributors
More tasks
More environments
More verified data
More data means a stronger data engine.
This creates a powerful network effect.
As more people contribute to Axis, the platform can generate more valuable data.
As the data engine becomes stronger, it becomes more useful to robotics companies.
And as the ecosystem grows, more contributors have a reason to participate.
That’s why Axis needs more contributors.
You’re not simply completing tasks.
You’re helping create the data infrastructure that could power the next generation of intelligent robots.
The future of robotics needs more than better hardware.
It needs better data.
And better data starts with a community willing to build it, one verified task at a time
2
108
They’ve been a little change in the point system. But ways of navigating was also mentioned. I hope we all win. It was an awesome time during the community ama. I’m anticipating already for the next one @iamlogtun @axisrobotics
BigNelson retweeted
🔔Axis Robotics VC AMA Quick Summary
🔰Tips to Earn More Axis Point
• Contribute quality data
• Do tasks of different diversity.
• Do more difficult tasks; it attracts high points.
• Refer to earn points; you ONLY earn points as your referral contributes trajectories.
🔰TGE Allocation Breakdown
• Trajectory contribution on the hub (Largest allocation)
• Marketing (Kaito & Content guys)
• Discord roles (X, Y, Z axis role holders)
🔰Best Practices
• Avoid repeatedly doing the same task more than 5 times; the current system only rewards the first 5 valid submissions.
• Prioritize quality over quantity.
• Submit diverse, high-quality datasets.
🔰Coming soon
• Mobile integration
• Ambassadorship program
• Revenue sharing model
• Kaito Campaign (Mid Aug)
@axisrobotics @plpiaoliang @0xzagen @0xsexybanana @chris_anm01
BigNelson retweeted
🚨Axis Nigeria Content Campaign EP 4
Pleased to officially announce our content campaign focused on the vision, technology, and community of @axisrobotics .
This campaign is designed to help deepen community understanding of the ecosystem while encouraging consistent participation and creativity.
Base point for each content is 10 points, top quality content get between 15 - 20 points dc point.
🔰Guidelines
- Create content based on the daily campaign topics
- Must be post on Axis NG community page on X
- Formats allowed : threads, videos, infographics, and visual explainers
- Low-effort Content will receive 5 points.
- Tag @axisrobotics and the Region Leader @iamlogtun on your posts.
🔰Check CS for content topics and show us your creator prowess!🔥
@plpiaoliang @0xsexybanana @0xzagen
Nigeria alone had 4 million more births than the entire EU last year! grok.com/share/bGVnYWN5_55b9…
This is my entry for the Nigeria art axis content. @axisrobotics
Today’s space on @axisrobotics was more about, a special onboarding session focused on introducing or deepening engagement with the Axis Robotics ecosystem. It was promoted as ideal for both newcomers discovering the project and existing community members.
I’ve seen hundreds of meme coin launches.
Most start with hype. A few start with a good idea.
CORE ( @core_trmnl ) is one of the few that caught my attention because the process is different.
You don’t need a big audience, a famous team or thousands of followers.
You bring an idea and a ticker.
CORE evaluates it, filters through the noise, and the rare ideas get passed to a human board that can actually build the brand, the community and the story around it.
That’s a much more interesting model than another copy paste launch.
Worth watching.
$CORE
CA: 4S3MRMzmydL8fi9i7nFHjPCMXeqMqFN2mjAgAg6Wpump
Dex: dexscreener.com/solana/3uyy4…




















