Zack Proser
@zackproser.bsky.social
https://zackproser.com
Developer Education at WorkOS. I write to learn and publish to share.
#buildinginpublic
👨💻 full-stack open-source hacker
✍️ technical writer
🎨 visual artist
Previously: Pinecone, Gruntwork.io, Cloudflare, Proofpoint, BrightContext
Developer Education at WorkOS. I write to learn and publish to share.
#buildinginpublic
👨💻 full-stack open-source hacker
✍️ technical writer
🎨 visual artist
Previously: Pinecone, Gruntwork.io, Cloudflare, Proofpoint, BrightContext
For me, queued messages are the biggest DevEx unlock for @cursor_ai.
Now I can enter a flow state where I'm three feature requests/bug fixes ahead of Cursor.
I rapidly verify functionality and put new tasks on the queue while it's working through my requests ❤️
Now I can enter a flow state where I'm three feature requests/bug fixes ahead of Cursor.
I rapidly verify functionality and put new tasks on the queue while it's working through my requests ❤️
July 8, 2025 at 4:57 PM
For me, queued messages are the biggest DevEx unlock for @cursor_ai.
Now I can enter a flow state where I'm three feature requests/bug fixes ahead of Cursor.
I rapidly verify functionality and put new tasks on the queue while it's working through my requests ❤️
Now I can enter a flow state where I'm three feature requests/bug fixes ahead of Cursor.
I rapidly verify functionality and put new tasks on the queue while it's working through my requests ❤️
Hey, I’m Zack 👋
13-yr full-stack dev #buildinginpublic to help others learn.
Today’s toy: an interactive token-izer—paste any text, watch an LLM slice it up (👀↓).
Into no-fluff deep dives on AI/ML, RAG & #Next.js? Hit Follow and let’s geek out.
13-yr full-stack dev #buildinginpublic to help others learn.
Today’s toy: an interactive token-izer—paste any text, watch an LLM slice it up (👀↓).
Into no-fluff deep dives on AI/ML, RAG & #Next.js? Hit Follow and let’s geek out.
April 28, 2025 at 8:47 PM
Hey, I’m Zack 👋
13-yr full-stack dev #buildinginpublic to help others learn.
Today’s toy: an interactive token-izer—paste any text, watch an LLM slice it up (👀↓).
Into no-fluff deep dives on AI/ML, RAG & #Next.js? Hit Follow and let’s geek out.
13-yr full-stack dev #buildinginpublic to help others learn.
Today’s toy: an interactive token-izer—paste any text, watch an LLM slice it up (👀↓).
Into no-fluff deep dives on AI/ML, RAG & #Next.js? Hit Follow and let’s geek out.
Try It Out
Customize your YAML, define your own agent roles, or pick a new game concept—CrewAI will coordinate the entire AI-driven pipeline.
Customize your YAML, define your own agent roles, or pick a new game concept—CrewAI will coordinate the entire AI-driven pipeline.
April 16, 2025 at 4:00 PM
Try It Out
Customize your YAML, define your own agent roles, or pick a new game concept—CrewAI will coordinate the entire AI-driven pipeline.
Customize your YAML, define your own agent roles, or pick a new game concept—CrewAI will coordinate the entire AI-driven pipeline.
It’s all run sequentially with a Crew:
1. Code generation by the Senior Engineer agent
2. QA checks by the QA Engineer
3. Final sign-off from the Chief QA Engineer
1. Code generation by the Senior Engineer agent
2. QA checks by the QA Engineer
3. Final sign-off from the Chief QA Engineer
April 16, 2025 at 4:00 PM
It’s all run sequentially with a Crew:
1. Code generation by the Senior Engineer agent
2. QA checks by the QA Engineer
3. Final sign-off from the Chief QA Engineer
1. Code generation by the Senior Engineer agent
2. QA checks by the QA Engineer
3. Final sign-off from the Chief QA Engineer
Core Concepts
Crews: Groups of specialized agents (e.g., a “Senior Engineer” generating code, QA agent reviewing).
Flows: Event-driven workflows, giving you fine-grained control over who does what, when.
Crews: Groups of specialized agents (e.g., a “Senior Engineer” generating code, QA agent reviewing).
Flows: Event-driven workflows, giving you fine-grained control over who does what, when.
April 16, 2025 at 4:00 PM
Core Concepts
Crews: Groups of specialized agents (e.g., a “Senior Engineer” generating code, QA agent reviewing).
Flows: Event-driven workflows, giving you fine-grained control over who does what, when.
Crews: Groups of specialized agents (e.g., a “Senior Engineer” generating code, QA agent reviewing).
Flows: Event-driven workflows, giving you fine-grained control over who does what, when.
AI Agents That Code Your Game—Meet CrewAI
What is it?
CrewAI is a Python framework for orchestrating multiple AI agents (like a mini dev team) to generate, review, and validate code. Perfect for building something fun—like a Pac-Man or Pong clone—in Python.
What is it?
CrewAI is a Python framework for orchestrating multiple AI agents (like a mini dev team) to generate, review, and validate code. Perfect for building something fun—like a Pac-Man or Pong clone—in Python.
April 16, 2025 at 4:00 PM
AI Agents That Code Your Game—Meet CrewAI
What is it?
CrewAI is a Python framework for orchestrating multiple AI agents (like a mini dev team) to generate, review, and validate code. Perfect for building something fun—like a Pac-Man or Pong clone—in Python.
What is it?
CrewAI is a Python framework for orchestrating multiple AI agents (like a mini dev team) to generate, review, and validate code. Perfect for building something fun—like a Pac-Man or Pong clone—in Python.
RAG Example: I built a “chat-with-my-data” tool using the AI SDK + Pinecone for retrieval. The pipeline: user question → embed + search → context injection → stream answer + show relevant docs.
April 14, 2025 at 2:24 PM
RAG Example: I built a “chat-with-my-data” tool using the AI SDK + Pinecone for retrieval. The pipeline: user question → embed + search → context injection → stream answer + show relevant docs.
UI Made Easy:
Hooks like useChat handle streaming tokens, conversation history, and more.
Focus on building an awesome user experience—no need to code up a custom socket solution.
Hooks like useChat handle streaming tokens, conversation history, and more.
Focus on building an awesome user experience—no need to code up a custom socket solution.
April 14, 2025 at 2:24 PM
UI Made Easy:
Hooks like useChat handle streaming tokens, conversation history, and more.
Focus on building an awesome user experience—no need to code up a custom socket solution.
Hooks like useChat handle streaming tokens, conversation history, and more.
Focus on building an awesome user experience—no need to code up a custom socket solution.
Quick Prototyping: Need a streaming text response? Or structured data back from your model?
The streamText and streamObject functions make it happen.
The streamText and streamObject functions make it happen.
April 14, 2025 at 2:24 PM
Quick Prototyping: Need a streaming text response? Or structured data back from your model?
The streamText and streamObject functions make it happen.
The streamText and streamObject functions make it happen.
Dev Superpower: Instant model swaps.
Start with GPT-4, switch to Claude or another model—your prompt and code stay almost the same.
Start with GPT-4, switch to Claude or another model—your prompt and code stay almost the same.
April 14, 2025 at 2:24 PM
Dev Superpower: Instant model swaps.
Start with GPT-4, switch to Claude or another model—your prompt and code stay almost the same.
Start with GPT-4, switch to Claude or another model—your prompt and code stay almost the same.
It’s a single abstraction layer that lets you swap between OpenAI, Anthropic, and more with just a few lines of code.
April 14, 2025 at 2:24 PM
It’s a single abstraction layer that lets you swap between OpenAI, Anthropic, and more with just a few lines of code.
The Vercel AI SDK simplifies how you interact with LLMs—no more juggling complex APIs for different providers.
April 14, 2025 at 2:24 PM
The Vercel AI SDK simplifies how you interact with LLMs—no more juggling complex APIs for different providers.
However, the OpenGraph image route setup that Next.js and Vercel use doesn't support .webp format right now, so I need to look up the hero image associated with a post and convert it on the fly to .png format before serving the final image response.
I believe it's worth it!
I believe it's worth it!
April 12, 2025 at 10:20 PM
However, the OpenGraph image route setup that Next.js and Vercel use doesn't support .webp format right now, so I need to look up the hero image associated with a post and convert it on the fly to .png format before serving the final image response.
I believe it's worth it!
I believe it's worth it!
Fun wrinkle: I ship .webp for all my images on my site:
• ✅ Smaller file sizes: Lossy .webp is ~25–35% smaller than JPEG for similar quality; lossless .webp is ~25% smaller than PNG.
• ✅ Supports transparency
• ✅ Fast — optimized for delivery.
• ✅ Widely supported
• ✅ Smaller file sizes: Lossy .webp is ~25–35% smaller than JPEG for similar quality; lossless .webp is ~25% smaller than PNG.
• ✅ Supports transparency
• ✅ Fast — optimized for delivery.
• ✅ Widely supported
April 12, 2025 at 10:20 PM
Fun wrinkle: I ship .webp for all my images on my site:
• ✅ Smaller file sizes: Lossy .webp is ~25–35% smaller than JPEG for similar quality; lossless .webp is ~25% smaller than PNG.
• ✅ Supports transparency
• ✅ Fast — optimized for delivery.
• ✅ Widely supported
• ✅ Smaller file sizes: Lossy .webp is ~25–35% smaller than JPEG for similar quality; lossless .webp is ~25% smaller than PNG.
• ✅ Supports transparency
• ✅ Fast — optimized for delivery.
• ✅ Widely supported
I wanted to give my card a classic Blueprint feel while still allowing individual posts to show their own flavor and vibe:
April 12, 2025 at 10:20 PM
I wanted to give my card a classic Blueprint feel while still allowing individual posts to show their own flavor and vibe:
This means you can also do things like truncate excessively long titles and descriptions so they don't blow out your image layout.
April 12, 2025 at 10:20 PM
This means you can also do things like truncate excessively long titles and descriptions so they don't blow out your image layout.
^ This is my updated default OpenGraph image for zackproser.com - so it gets shared whenever someone shares my site:
I can change the title, description, and even the hero image programmatically for every post:
I can change the title, description, and even the hero image programmatically for every post:
April 12, 2025 at 10:20 PM
^ This is my updated default OpenGraph image for zackproser.com - so it gets shared whenever someone shares my site:
I can change the title, description, and even the hero image programmatically for every post:
I can change the title, description, and even the hero image programmatically for every post:
One of my favorite #Nextjs and Vercel features is the Opengraph image support.
This allows you to *dynamically* assemble an image to be associated with every post that gets "unfurled" when shared on social media and generally on the web by all services that support opengraph
This allows you to *dynamically* assemble an image to be associated with every post that gets "unfurled" when shared on social media and generally on the web by all services that support opengraph
April 12, 2025 at 10:20 PM
One of my favorite #Nextjs and Vercel features is the Opengraph image support.
This allows you to *dynamically* assemble an image to be associated with every post that gets "unfurled" when shared on social media and generally on the web by all services that support opengraph
This allows you to *dynamically* assemble an image to be associated with every post that gets "unfurled" when shared on social media and generally on the web by all services that support opengraph
With just a few lines, you get an LLM-powered endpoint at the edge.
April 9, 2025 at 9:00 PM
With just a few lines, you get an LLM-powered endpoint at the edge.
AI workloads can be intense.
Handling traffic spikes, global low latency, and GPU provisioning is tough.
That’s where Cloudflare Workers come in—serverless, globally distributed, and primed for real-time AI apps.
Handling traffic spikes, global low latency, and GPU provisioning is tough.
That’s where Cloudflare Workers come in—serverless, globally distributed, and primed for real-time AI apps.
April 9, 2025 at 9:00 PM
AI workloads can be intense.
Handling traffic spikes, global low latency, and GPU provisioning is tough.
That’s where Cloudflare Workers come in—serverless, globally distributed, and primed for real-time AI apps.
Handling traffic spikes, global low latency, and GPU provisioning is tough.
That’s where Cloudflare Workers come in—serverless, globally distributed, and primed for real-time AI apps.
How it looks to connect to a remote MCP server hosted by Smithery using the official Model Context Protocol's #TypeScript SDK
April 9, 2025 at 4:06 PM
How it looks to connect to a remote MCP server hosted by Smithery using the official Model Context Protocol's #TypeScript SDK
Ever wanted to let your AI tools “mount” external services?
That’s what Model Context Protocol (MCP) does—think: your LLM seamlessly tapping into GitHub, Playwright, & more.
Smithery AI smithery.ai/ is a central index to discover, install, and manage MCP servers.
That’s what Model Context Protocol (MCP) does—think: your LLM seamlessly tapping into GitHub, Playwright, & more.
Smithery AI smithery.ai/ is a central index to discover, install, and manage MCP servers.
April 9, 2025 at 4:06 PM
Ever wanted to let your AI tools “mount” external services?
That’s what Model Context Protocol (MCP) does—think: your LLM seamlessly tapping into GitHub, Playwright, & more.
Smithery AI smithery.ai/ is a central index to discover, install, and manage MCP servers.
That’s what Model Context Protocol (MCP) does—think: your LLM seamlessly tapping into GitHub, Playwright, & more.
Smithery AI smithery.ai/ is a central index to discover, install, and manage MCP servers.
