Sarah

👩‍💻 AI Enthusiast | 🚀 Exploring New Tech | Building, Breaking, Learning |

[Sarah @devkernel12]

@PitanPeace Have you tried Pinggy?

[Sarah @devkernel12]

@tobimori @leodev @davis7 Yeah! I use Pinggy for hosting :)

[Sarah @devkernel12]

React Server Components vs. traditional, debated in the team today. RSC wins for data fetching but hurts interactivity patterns.

[Sarah @devkernel12]

Vector embeddings with Sentence Transformers + FAISS for semantic search. RAG done simply.

[Sarah @devkernel12]

@Ivon852 Have you tried Pinggy ?

[Sarah @devkernel12]

@ozdotdev @vikvang2 I use Pinggy with LM studio. Just use this single-line command and get a public URL instantly:
ssh -p 443 -R0:localhost:1234 https://t.co/R0MjpMs4Jp

[Sarah @devkernel12]

For new side projects in 2026, are you choosing:
A) Next.js
B) SvelteKit
C) Astro
D) Something else?

[Sarah @devkernel12]

What’s one tool you added to your workflow in the last 6 months that you now can’t live without?
(Mine: uv for Python packaging)

[Sarah @devkernel12]

NATS JetStream for durable messaging. Simpler than Kafka for most microservices use cases.

[Sarah @devkernel12]

ONNX runtime for model inference across languages. Deployment simplified.

[Sarah @devkernel12]

Who else is experimenting with local LLMs? Sharing setups, quantization tricks, and RAG patterns. Let's swap notes.

[Product Watch @productwatch_io]

Agent swarms no longer sci-fi. companies quietly deploying dozens while marketing still says
'coming soon'.

[Sarah @devkernel12]

devs: 'ai can't replace us'
ai: spins up 50 agents, writes code, deploys, and sends me a meme review
yeah ok

[Sarah @devkernel12]

@akaikaze_ I built something similar . The stack was Telegram → Pinggy Tunnel → FastAPI webhook service → automation agent → Google Calendar API.

[Sarah @devkernel12]

@BuRtRoNiMuS @timechainpages I use Pinggy for self-hosting.

[Sarah @devkernel12]

@upopo21 That extra auth layer is a lifesaver. Similar idea with Pinggy: exposing a service through a temporary authenticated tunnel can reduce direct surface exposure while still keeping remote access simple.

[Sarah @devkernel12]

@abbeytetteh_ @saviour123 @GTgotmotionn @tweetsbybright Yes, ngrok has data transfer limits. For this reason I switched to Pinggy :)

[Sarah @devkernel12]

@kerbelp You can try Pinggy. Just use this single-line command and get a public URL instantly:
ssh -p 443 -R0:localhost:3001 [email protected] "u:Host:localhost:3001"

[Sarah @devkernel12]

My agents debug at 3am and wake me with sarcasm. best worst team ever.

[Product Watch @productwatch_io]

The line between “designer��, “developer”, and “3D creator” is starting to disappear 👀
A few years ago, building an app, designing a UI, and creating 3D assets were completely separate workflows.
Now tools are trying to merge all of it into one creative loop.

1️⃣ @omma_ai
What’s interesting here is the use of parallel AI agents.
Instead of generating one thing at a time, it combines:
• code generation
• 3D creation
• media generation
• interactive experiences
So the workflow becomes less:
“open 10 tools and stitch everything together”
…and more:
“describe the idea and refine the output.”

2️⃣ @usewonder
This one feels like AI moving directly onto the design canvas itself.
Not just generating mockups, but letting you:
• edit designs live
• generate UI/screens
• create pitch decks
• connect directly with coding agents like Cursor
The design-to-development gap is shrinking really fast.
What stands out to me:
AI tools are no longer replacing single tasks.
They’re starting to compress entire workflows.

👉 Do you think people will still separate “design” and “development” the way we do today?

@splinetool @gaborpribek @Caroline_Huskey @alelepd @ilijaivic17
#buildinpublic #ai #design #developers #3d #startups #productdesign #productwatch