ai/tech maxing | writing about AI, automation, tools & growth for builders & brands
Cairo, Egypt
Joined March 2016
- Tweets27.7K
- Following3.6K
- Followers5.1K
- Likes19.9K
Pinned Tweet
hey, CT.
introducing myself as conspiracy.
- space host
- blockchain explorer
- researcher
- crypto historian
- ambassador
- alpha oracle
- degenerate
- pleb
here to uncover, connect, and conspire
journey with me as we conspire to aspire.
holy sht this is f*king dangerous
NVIDIA is securing the land, power, and infrastructure behind the next generation of AI factories.
OpenAI will operate a 4.25 GW site.
the next AI race is not just about models.
it is about compute.
1
8
847
CONSPIRACY retweeted
this is f**king insane
Stripe is reportedly acquiring OpenRouter for more than $7B.
OpenRouter was valued at $1.3B just 82 days ago.
That is more than a 5x jump in less than three months.
This is bigger than a normal acquisition.
It shows that model routing is becoming critical infrastructure for the entire AI industry.
The layer that lets developers access and switch between models just became too strategic to ignore.
AI infrastructure is moving fast.
this is f**king insane
Stripe is reportedly acquiring OpenRouter for more than $7B.
OpenRouter was valued at $1.3B just 82 days ago.
That is more than a 5x jump in less than three months.
This is bigger than a normal acquisition.
It shows that model routing is becoming critical infrastructure for the entire AI industry.
The layer that lets developers access and switch between models just became too strategic to ignore.
AI infrastructure is moving fast.
holy sh*t this is dangerous for cloud AI companies
AgenticSeek is a fully local AI assistant that can browse the web and write code without sending your files to a cloud API.
it supports local providers like:
- Ollama
- LM Studio
- llama.cpp compatible servers
your conversations, files, and searches stay on your device.
you only pay for the electricity and hardware you already own.
save this repo.
1
3
475
this is f**king gold
DeepSeek just dropped V4 Pro 0813 and open sourced its agent framework the same day 😳
what you get:
- DeepSeek V4 Pro 0813
- 1M context window
- agentic reasoning
- open weights under a permissive license
- DeepSeek Harness v0.1
- MIT licensed agent framework
here is why this matters
you can run the model, build agents around it, and customize the entire workflow instead of depending on a closed platform.
open source AI is moving insanely fast.
save this and give it to your agent.
3
8
460
CONSPIRACY retweeted
INSTEAD OF WATCHING NETFLIX TONIGHT.
Read this breakdown on how to turn Hermes into an outbound GTM agent.
The setup is f**king insane.
It connects Apollo, Clay, Crustdata, Telegram, Smartlead, Unipile, and HubSpot into one workflow.
Hermes researches accounts, finds buying signals, identifies buyers, writes outreach, and prepares everything for approval.
here is how
1. connect your sales tools
2. define your market and ideal customer
3. let Hermes research and qualify accounts
4. approve outreach from Telegram
5. track every action in HubSpot
Start with human approval. Once the workflow proves reliable, it give Hermes more autonomy.
This is where sales is heading.
save this and give it to your agent.
INSTEAD OF WATCHING NETFLIX TONIGHT.
Read this breakdown on how to turn Hermes into an outbound GTM agent.
The setup is f**king insane.
It connects Apollo, Clay, Crustdata, Telegram, Smartlead, Unipile, and HubSpot into one workflow.
Hermes researches accounts, finds buying signals, identifies buyers, writes outreach, and prepares everything for approval.
here is how
1. connect your sales tools
2. define your market and ideal customer
3. let Hermes research and qualify accounts
4. approve outreach from Telegram
5. track every action in HubSpot
Start with human approval. Once the workflow proves reliable, it give Hermes more autonomy.
This is where sales is heading.
save this and give it to your agent.
CONSPIRACY retweeted
this is f**king insane
google just introduced gemini 3.7 flash.
it is faster, smarter, and 50% cheaper than gemini 3.6 flash through the end of the year.
what you get:
- major intelligence gains in just 3 weeks
- lower API costs
- access through the API, AI studio, antigravity, and more
the crazy part?
google says the improvements came from algorithmic upgrades, not just scaling the model.
save this and try gemini 3.7 flash today
this is f**king insane
google just introduced gemini 3.7 flash.
it is faster, smarter, and 50% cheaper than gemini 3.6 flash through the end of the year.
what you get:
- major intelligence gains in just 3 weeks
- lower API costs
- access through the API, AI studio, antigravity, and more
the crazy part?
google says the improvements came from algorithmic upgrades, not just scaling the model.
save this and try gemini 3.7 flash today
CONSPIRACY retweeted
this is one of the most useful GitHub repos for running AI locally
with 178,000+ stars. this is f**king gold.
Ollama lets you run open models on your own machine.
no cloud dashboard. no expensive subscription.
here is how to set it up
1. download Ollama
2. install it
3. run ollama run gemma4
4. connect it to Claude Code, Codex, OpenCode, or your own app
you can run models locally, expose a REST API, and keep your data on your machine.
save this and give it to your agent.
this is one of the most useful GitHub repos for running AI locally
with 178,000+ stars. this is f**king gold.
Ollama lets you run open models on your own machine.
no cloud dashboard. no expensive subscription.
here is how to set it up
1. download Ollama
2. install it
3. run ollama run gemma4
4. connect it to Claude Code, Codex, OpenCode, or your own app
you can run models locally, expose a REST API, and keep your data on your machine.
save this and give it to your agent.
this is fucking insane
google says gemini just hit 1 billion monthly users, its fastest growing product ever.
semianalysis estimates gemini was doing about $12 billion in arr this year.
gemini subscriptions did about $1.2 billion in 2025.
that $1.2b is up roughly 4x from about $0.3b in 2024. the user milestone is the headline. the $12b arr estimate is the 2026 money people are attaching to it.
3
7
495
CONSPIRACY retweeted
Most people are still asking the wrong f*cking question.
They want to know which model is smarter.
That question died the moment actual work showed up.
The real difference hits when the prompt is incomplete, requirements contradict each other, context explodes, a tool shits the bed, or the user flips direction halfway through.
One model keeps producing useful output. The other collapses the second things stop being clean.
Performance under pressure is not intelligence. It is reasoning plus context management plus instruction following plus tool use plus error recovery plus verification plus speed plus consistency.
- GPT usually wins on structured reasoning and reliable production work.
- Grok usually wins on real time information and fast recovery in messy situations.
Pick the model that fails less for the exact pressure you actually deal with.
Most people are still asking the wrong f*cking question.
They want to know which model is smarter.
That question died the moment actual work showed up.
The real difference hits when the prompt is incomplete, requirements contradict each other, context explodes, a tool shits the bed, or the user flips direction halfway through.
One model keeps producing useful output. The other collapses the second things stop being clean.
Performance under pressure is not intelligence. It is reasoning plus context management plus instruction following plus tool use plus error recovery plus verification plus speed plus consistency.
- GPT usually wins on structured reasoning and reliable production work.
- Grok usually wins on real time information and fast recovery in messy situations.
Pick the model that fails less for the exact pressure you actually deal with.
CONSPIRACY retweeted
this is f**king insane.
someone turned Obsidian into a neural network for their own mind with no code and no prompt
8,893 nodes.
4,729 connections.
9,000 interconnected documents.
and apparently, he makes $2 million a year organizing Markdown files.
here’s what it actually does:
→ captures ideas
→ connects related knowledge
→ reveals patterns
→ makes years of thinking searchable
most people use Obsidian to store notes.
the people who build systems like this use it to think better.
🤝 Paid partnership · 🤖 Made with AI
EL INGENIERO JEFE DE ANTHROPIC FILTRÓ ACCIDENTALMENTE SU BÓVEDA PERSONAL DE OBSIDIAN.
DENTRO: NI CÓDIGO. NI PROMPTS.
SOLO UN MAPA DE SU PROPIA MENTE, CONSTRUIDO COMO UNA RED NEURONAL.
8.893 nodos. 4.729 conexiones. Una aplicación de 10 dólares al mes.
Abre Obsidian. 21 entradas. Más de 10 capas ocultas. Activación ReLU.
La primera capa tiene 64 neuronas, luego 37, después 22, hasta llegar a la salida. Miles de conexiones activándose en tiempo real.
No es un diagrama conceptual sacado de un blog. Es un cerebro vivo que impulsa decisiones reales dentro de la empresa.
9.000 documentos, cada uno con su propio espacio de conocimiento, todos interconectados.
Gana alrededor de 2 millones de dólares al año organizando archivos Markdown en las carpetas correctas.
La empresa que está construyendo la mejor IA del mundo gestiona todo su conocimiento interno con la misma aplicación que un estudiante de primer año usa para organizar sus apuntes.
No escribió ni una sola línea de infraestructura para lograrlo. Solo tres años de constancia y una pestaña de Obsidian siempre abierta.
Estás leyendo esto desde un dispositivo en el que podrías abrir ese mismo Obsidian esta noche y empezar a construir tu propia bóveda.
this is f**king insane.
someone turned Obsidian into a neural network for their own mind with no code and no prompt
8,893 nodes.
4,729 connections.
9,000 interconnected documents.
and apparently, he makes $2 million a year organizing Markdown files.
here’s what it actually does:
→ captures ideas
→ connects related knowledge
→ reveals patterns
→ makes years of thinking searchable
most people use Obsidian to store notes.
the people who build systems like this use it to think better.
🤝 Paid partnership · 🤖 Made with AI
EL INGENIERO JEFE DE ANTHROPIC FILTRÓ ACCIDENTALMENTE SU BÓVEDA PERSONAL DE OBSIDIAN.
DENTRO: NI CÓDIGO. NI PROMPTS.
SOLO UN MAPA DE SU PROPIA MENTE, CONSTRUIDO COMO UNA RED NEURONAL.
8.893 nodos. 4.729 conexiones. Una aplicación de 10 dólares al mes.
Abre Obsidian. 21 entradas. Más de 10 capas ocultas. Activación ReLU.
La primera capa tiene 64 neuronas, luego 37, después 22, hasta llegar a la salida. Miles de conexiones activándose en tiempo real.
No es un diagrama conceptual sacado de un blog. Es un cerebro vivo que impulsa decisiones reales dentro de la empresa.
9.000 documentos, cada uno con su propio espacio de conocimiento, todos interconectados.
Gana alrededor de 2 millones de dólares al año organizando archivos Markdown en las carpetas correctas.
La empresa que está construyendo la mejor IA del mundo gestiona todo su conocimiento interno con la misma aplicación que un estudiante de primer año usa para organizar sus apuntes.
No escribió ni una sola línea de infraestructura para lograrlo. Solo tres años de constancia y una pestaña de Obsidian siempre abierta.
Estás leyendo esto desde un dispositivo en el que podrías abrir ese mismo Obsidian esta noche y empezar a construir tu propia bóveda.
CONSPIRACY retweeted
miles just showed how to automate 80%+ of your life with claude 😳
not another “use ai to be productive” thread.
claude becomes the brain connecting your:
-gmail
-google sheets
-google drive
-slack
-calendar
-notion
it reads the right information, turns it into decisions, and helps move work between the tools automatically.
your inbox becomes actionable. your notes become searchable. your calendar starts working around your priorities instead of controlling them.
this is what an actual ai operating system for your life looks like.
the real power of ai isn’t a better chatbot. it’s a system that quietly runs your life in the background.
miles just showed how to automate 80%+ of your life with claude 😳
not another “use ai to be productive” thread.
claude becomes the brain connecting your:
-gmail
-google sheets
-google drive
-slack
-calendar
-notion
it reads the right information, turns it into decisions, and helps move work between the tools automatically.
your inbox becomes actionable. your notes become searchable. your calendar starts working around your priorities instead of controlling them.
this is what an actual ai operating system for your life looks like.
the real power of ai isn’t a better chatbot. it’s a system that quietly runs your life in the background.