Eddy | VegliaDev

Practical AI tools & open-source alternatives that replace expensive SaaS | Dev workflows, agent experiments & real breakthroughs | Building in public

[Eddy | VegliaDev @VegliaDev]

I’ve been a big fan of The Diary Of A CEO podcast for a while. The guests, conversations, and depth of ideas were so valuable that I wanted a better way to study them.

So I vibe-coded a Chrome extension that let me download the video transcripts. I ended up collecting over 180 transcripts, then categorized them by theme and fed them into an LLM.

From there, I worked with the model to synthesize the recurring ideas, patterns, and principles that kept appearing across the conversations.

That process became the foundation for this article.

[Eddy | VegliaDev @VegliaDev]

https://t.co/M0eWzTCdt4

[Eddy | VegliaDev @VegliaDev]

PewDiePie dropped Odysseus — open-source AI workspace with 45k stars already.

Local models, agents, MCP tools, memory, notes, image editor.

Built for devs who want Claude-level productivity without the bill.

This is the energy.
Big creators shipping real OSS alternatives.

[Eddy | VegliaDev @VegliaDev]

Plausible Analytics is the cleanest privacy-first Google Analytics replacement.

Lightweight, no cookies, beautiful dashboard, self-hostable.

Ditched GA months ago and my site feels 10x better.

https://t.co/0VF3q82vLC

Indie hackers: stop feeding Big Tech your visitor data.

What analytics are you using?

[Eddy | VegliaDev @VegliaDev]

This is insane:
self-hosted AI that generates full slide decks from a prompt or PDF, exports real editable PPTX, and runs completely offline.

Presenton just dropped proper custom templates with HTML/Tailwind.

No more Gamma lock-in.

https://t.co/Pe1FO0gHka
https://t.co/8tLuHoeH1K
Bookmark this.

[Eddy | VegliaDev @VegliaDev]

@NFTCPS Looks interesting, haven't seen many people talk about it. I should try it this weekend

[Eddy | VegliaDev @VegliaDev]

This is one of the cleanest open-source automation tools I’ve tested lately:
https://t.co/iTY6gFfYoT

Think Zapier…
but self-hostable and developer-friendly.

Good fit for:

- AI workflows
- internal tooling
- content automations
- small SaaS ops

The market for “own your workflows” keeps growing.

What are you automating lately?

[Eddy | VegliaDev @VegliaDev]

OpenHands might be the most important open-source AI coding project right now.

https://t.co/puRW0EglEd

What makes it interesting:

- AI agents can inspect full repos
- run terminal commands
- edit files
- browse documentation
- debug issues
- open PRs

Not just autocomplete.
Closer to autonomous software operators.

The bigger shift:

AI coding is moving from:
“generate snippets”

to:
“manage development workflows.”

OpenHands also supports:

- CLI workflows
- local GUI
- cloud deployment
- multi-model setups

Feels much closer to the “open-source Devin” vision than most projects.

Worth watching closely if you care about:

- AI agents
- dev automation
- local-first AI
- self-hosted tooling

Project site:
https://t.co/fj0GA3VjaR

Bookmark this.

[Eddy | VegliaDev @VegliaDev]

The funniest trend on GitHub right now:

people are building entire ecosystems around AI coding agents.

Not just:

- prompts

But:

- memory systems
- skill packs
- orchestration layers
- model routers
- agent terminals

This repo exploded recently:
https://t.co/nEDUTszGkD

Developers are productizing workflows now.

Reply with the weirdest AI tooling you’ve seen lately.

[Eddy | VegliaDev @VegliaDev]

Local image upscaling that actually delivers.
Upscayl turns low-res assets into crisp 4x versions using open models.

Perfect for creators cleaning up old screenshots, mockups, or generating better thumbnails without paying for Midjourney.

https://t.co/WO1cgf5AG2

Free forever, runs on your GPU.

Worth trying?

[Eddy | VegliaDev @VegliaDev]

This paper is fascinating:
https://t.co/gb8qKywSdB

Researchers are exploring whether agent workflows should be “compiled into model weights” instead of orchestrated externally.

Translation:

Instead of massive agent frameworks…
the behavior becomes part of the model itself.

Could make agents:

- cheaper
- faster
- more private
- less context-heavy

The agent wars are just getting started.

[Eddy | VegliaDev @VegliaDev]

@TheEcomNomad That is true, but we are getting better and better models faster and faster. Soon the gap will disappear.

[Eddy | VegliaDev @VegliaDev]

This is one of the coolest open-source voice projects right now:
https://t.co/aQSqfWj57V

Local text-to-speech has become shockingly good.

Interesting use cases:

- AI narration
- automated videos
- language learning
- game dialogue
- accessibility tooling

Feels like another category moving away from expensive APIs.

[Eddy | VegliaDev @VegliaDev]

@ihtesham2005 It must have taken ages to collect and categorize everything and developers are giving it for free. Amazing stuff!

[Eddy | VegliaDev @VegliaDev]

Local coding assistants just got serious.

Tabby is a self-hosted AI coding tool that runs your models privately and integrates with VS Code.

No data leaving your machine. Full context from your private repos.

It’s the realistic Copilot replacement solo devs have been waiting for.

https://t.co/cjdNTI8YGL
https://t.co/9rGz8n3haN

Worth trying?

[Eddy | VegliaDev @VegliaDev]

I tested a surprisingly effective content pipeline:

1. Research with Perplexity
https://t.co/srl5DY6kfo

2. Store ideas in Obsidian
https://t.co/6g7aavNZOV

3. Generate visuals with Grok Imagine
https://t.co/1Y0hyW8Rln

4. Schedule through Typefully
https://t.co/YCZFupFtCr

The key insight:

Content quality improves dramatically when the workflow friction disappears.

Most creators quit because publishing feels heavy.

[Eddy | VegliaDev @VegliaDev]

Why is nobody talking about how useful Excalidraw became for AI workflows?

https://t.co/E0pz8j8AVb

You can start with:

- map agent flows
- design automations
- sketch backend systems
- plan products visually

Throw rough diagrams into Claude/OpenAI afterward and the implementation quality improves massively.

Text prompting alone is overrated.

Visual thinking matters.

Worth trying?

[Eddy | VegliaDev @VegliaDev]

GitHub trending right now is basically one giant signal:

developers want AI tools they fully control.

The fastest growing categories are:

- local inference
- agent memory
- AI coding infra
- self-hosted workflows
- MCP tooling

OSSInsight tracker:
https://t.co/LyN6InL3Ts

Feels like the “AI wrapper” era is slowly dying.

[Eddy | VegliaDev @VegliaDev]

This repo feels like cheating for AI coding workflows:
https://t.co/TIDdDGU6iW

Hugging Face built a lightweight agent framework that’s actually understandable.

No massive abstraction maze.
No enterprise nonsense.

Just:

- tools
- agents
- execution
- code

More AI frameworks should feel this simple.

Worth trying?

[Eddy | VegliaDev @VegliaDev]

DeepMind team just dropped something amazing! And we can try it for free!

[Google DeepMind @GoogleDeepMind]

We’re dropping Gemini Omni: our first step towards a model that can create anything from anything - starting with video.

It combines Gemini’s intelligence with our generative media systems - representing a leap forward in world understanding, multimodality, and editing 🧵

[Eddy | VegliaDev @VegliaDev]

@servasyy_ai It looks sick! I hope examples are not cherry picked

[Eddy | VegliaDev @VegliaDev]

@souravxbera Glad I could help. Good luck with your project 👊

[Eddy | VegliaDev @VegliaDev]

LangGraph isn’t hype. It’s the missing piece for reliable agents.

Stateful graphs, human-in-the-loop, persistence that actually works.

Turned my flaky prototype into production-grade overnight.

https://t.co/dDpRHl3wYl

If you’re serious about agents, start here.