Peter Akande
@_PeterAkande
Followers
287
Following
14K
Media
73
Statuses
1K
Software Engineer, Built @frenboxapp @authivate, Building an AI OS in stealth
Joined October 2021
Test with @Orceum_ai Research every morning (now including my mail) -> Post to X Looks more engaging now :) https://t.co/UFaZlCXeoT
AI agents don't need bigger models—just smarter ones. NVIDIA's Nemotron-3 is a lightweight model built for AI agents. It's smaller, faster, and better at following instructions. Democratizing AI for everyone. Source: Super Intelligence newsletter (ainutpam@substack.com)
0
0
1
AI agents don't need bigger models—just smarter ones. NVIDIA's Nemotron-3 is a lightweight model built for AI agents. It's smaller, faster, and better at following instructions. Democratizing AI for everyone. Source: Super Intelligence newsletter (ainutpam@substack.com)
0
0
1
Test with @Orceum_ai Research every morning -> Post to X Need to make it more engaging :) https://t.co/UFaZlCXeoT
QwenLong-L1.5 shattered the 4M-token barrier with cross-document memory. OpenAI adopted Anthropic's 'skills' mechanism. The arms race shifts from parameters to learning-to-learn. Two breakthroughs, one truth: AI's future is scaling smarter.
0
0
3
QwenLong-L1.5 shattered the 4M-token barrier with cross-document memory. OpenAI adopted Anthropic's 'skills' mechanism. The arms race shifts from parameters to learning-to-learn. Two breakthroughs, one truth: AI's future is scaling smarter.
0
0
2
🔉 Introducing SAM Audio, the first unified model that isolates any sound from complex audio mixtures using text, visual, or span prompts. We’re sharing SAM Audio with the community, along with a perception encoder model, benchmarks and research papers, to empower others to
207
900
6K
Test with @Orceum_ai Research about AI -> Post to X, every morning. https://t.co/khGBNlAs7s
Daily AI Discovery GPT-5.2: 70.9% on pro tasks (vs 38.8% for GPT-5), 100% on AIME 2025 math, 80% on SWE. Enterprise saves 40-60 min/day. Masters spreadsheets, presentations, code, and complex projects. AI shifts from assistance to pro performance across 44 occupations.
0
0
1
Daily AI Discovery GPT-5.2: 70.9% on pro tasks (vs 38.8% for GPT-5), 100% on AIME 2025 math, 80% on SWE. Enterprise saves 40-60 min/day. Masters spreadsheets, presentations, code, and complex projects. AI shifts from assistance to pro performance across 44 occupations.
0
0
1
AI's mature form is Ambience Intelligence - present everywhere, visible nowhere.
0
0
2
Trying something with my @Orceum_ai assistant. Research about something about AI -> post it here on X. Lets see how it goes :)
0
1
2
Adding Toon format support to spof is a great thing to do rn, would get on it. Spof is framework for structured prompts btw. https://t.co/t1KWAyYDkr
github.com
SPOF (Structured Prompt Output Framework) is an open-source Python framework for designing structured, maintainable, and type-safe prompts for Large Language Models (LLMs). - PeterAkande/spof
0
0
1
A pre‑pmf founder's daily experience is goated. One moment you’re pumped, talking about "damn, I’m going to change the world with this". The next moment things break and you’re like "how am I supposed to change the world with this?" – and the circle continues 😂
0
0
2
A lot of people don’t like to admit it, but let me be honest… The learning curve for cloud deployment can be crazy. AWS, Terraform, CloudFormation, IAM this, VPC that. One minute you’re writing code, next thing you’re deep in 20 tabs trying to figure out why your infra won’t
0
5
9
On this day, three years ago, the world caught a glimpse of intelligence that isn’t trapped in a human skull or an animal brain – an artificial mind. Yes, one that's born of some interesting math and a product made from sand.
today we launched ChatGPT. try talking with it here: https://t.co/uWra8LKFMN
0
1
1
Just hearing about the verifier's law by @_jasonwei, it tells a lot about why AI shines in some fields and not in some. Programming for example. Verifier's law: "The ability to train AI to solve a task is proportional to how easily verifiable the task is. Any solvable, easily
0
0
2
We’re starting to get a clearer sign of how vast the surface area of context engineering is going to be. To build AI agents, in theory, it should be as simple as having a super powerful model, giving it a set of tools, having a really good system prompt, and giving it access to
121
122
899
Watch as surgeons and nurses sing praises in the operating theatre while performing a long and challenging brain surgery on a 13-year-old girl.
737
7K
38K
@thedankoe What looks like effortless articulation is really just the repetition of a few big ideas someone has refined over years. Build your inner library, and the clarity arrives one insight at a time.@thedankoe
0
2
13