Talking our way out of reading and writing
Now that computers can listen and speak, do we still need reading and writing? A thought on why text might be a local optimum we could climb out of, over generations.
#ai
Posts tagged with ai
2025 in review, and plans for the coming year of DevRel
Reflecting on my first year as DevRel at Parallel—2.5M X impressions, cookbooks, MCP integrations, and LLMTEXT—and outlining my 2026 strategy for building SaaS on the agentic internet.
Introducing LLMTEXT, an open source toolkit for the llms.txt standard
Introduces LLMTEXT, an open-source toolkit for creating, validating, and using llms.txt files to make websites accessible to AI agents through standardized documentation and MCP servers.
hardware-screenless
Argues against the need for new hardware for AI assistants, proposing that existing simple phones paired with good headphones can deliver bidirectional voice interfaces effectively.
Installing MacOS without an internet connection
Investigates creating a completely internet-free development environment with large LLMs, exploring privacy, security, and focus benefits while documenting installation challenges and workarounds.
Outcomputed
Predicts the decline of human utility as AI surpasses knowledge workers in capability, proposing becoming "compute allocators" to stay relevant in an AI-dominated future.
scalable-ideas
Demonstrates how LLMs enable scalable knowledge work by breaking down complex tasks into API-driven subtasks, using building websites for GitHub repos as a practical example.
The post-smartphone era is coming.
Explores the shift from smartphone apps to natural language voice interfaces powered by AI agents, advocating for open source approaches to eliminate traditional business boundaries.
Why are systems always so hard to integrate with other systems?
Analyzes the barriers preventing seamless system integration and proposes solutions like API search engines and GPT normalization to overcome vendor lock-in and integration complexity.
Turning old texts into audiobooks with talking heads
A quick proof of concept that turns pages of old public-domain books into narrated videos with a talking-head version of the author.
How Tolkie used ActionSchema to generate beautiful graphics for low-literate website understanding
Case study of a pilot with Tolkie, using ActionSchema to compare prompts and models for generating simplified definitions and word images for low-literate readers.
"Human-to-Human Content Generation: Embracing the Social Computing Age"
Explores the concept of transforming everyday conversations into various forms of digital content through AI-powered "Chatcasts," ushering in the age of social computing.
A Blueprint for AGI: Bridging the Gap Between Open and Closed Systems
Proposes decentralized oversight as a third way between open and closed AGI, where users and their nearest communities keep AGI use transparent and can intervene against misuse.
Leveraging ChatGPT, DuxSoup and ActionSchema to send message all recruiters in my personal LinkedIn network
How I combined ChatGPT, DuxSoup and ActionSchema to identify and message every tech recruiter in my personal LinkedIn network.
Cold emailing suitable investors
Shows how I filtered a dataset of 2000+ GenAI investors and used ActionSchema and GPT to write each suitable investor a personalised cold email.
Writing 285 amazing cover letters while hunting for remote jobs
Walks through using ActionSchema and GPT to find remote TypeScript jobs and write 285 tailored cover letters in bulk.
Theory of Intelligence
Frames intelligence as an algorithm that creates meaning out of chaos, argues that a business is a form of intelligence, and sketches what that means for building AGI with LLMs.
Classification tasks with ChatGPT
Presents a simple yet powerful method for using ChatGPT to perform fast parallel classification tasks, such as determining gender from names with 250ms response times.
**Integrating AI into Business Sectors**
Shares practical insights on integrating AI into businesses, covering GPT applications, privacy considerations, and advice for startups and enthusiasts embracing the AI revolution.
E2E Testing ChatGPT AI plugins with using MRKL
Demonstrates how to create end-to-end tests for ChatGPT AI plugins using the MRKL technology, with practical examples testing a ScrapingBee plugin.
The AI Boom is Near...
A reflection on how AI and machine learning will soon understand us better than we know ourselves, leading to personal AI advisors that could transform decision-making and human behavior.
How to create a sustainable world?
A philosophical exploration of humanity's environmental impact and potential solutions, discussing climate-control, evolution-control, and population management as strategies for creating a sustainable future.
Capitalism drives cyborgs to the top, if we like it or not
A speculative warning about the potential future where capitalism drives the creation of cyborgs who may become superior to humans in business, exploring both optimistic and pessimistic scenarios for human-machine integration.
Is Skynet already there? My prediction of the collapse of internet as we know it
A prediction about the potential emergence of a distributed deep learning virus that could achieve human-level intelligence by utilizing multiple computers, exploring the implications and preventive measures needed to avoid an AI threat scenario.