YeeroAI

Hey IH,

Working on YeeroAI (an AI collaboration tool) and shipped a bunch of stuff recently.

**Multi-select messages → "pick out to branch"** — when you have useful replies scattered across a long conversation, you can now select multiple messages and move them to a new or existing branch in one click. Came from my own frustration of losing good responses in long branching conversations.

**CLI v0.2.0 with yeero do** — describe what you want in natural language, no need to memorize commands. The cool part: unlike agents, yeero do only calls the LLM once to generate a Python script, then runs it locally every time after. Zero token cost for repeated tasks. If you have tool-like tasks you run often, this saves a lot compared to using an agent every time.

**Desktop v0.1.0** — lightweight Tauri app for Windows, macOS, and Linux. Still has some rough edges honestly: auto-start hasn't been thoroughly tested, auto-update has signing issues, and macOS doesn't have a developer cert yet so you need to run xattr -cr after install. Working on fixing these, but it's usable now.

**Patch-mode app updates** — the biggest frustration with AI-generated apps was that every update rewrote everything. Now there's a patch mode: describe what you want changed, and only that part gets modified. The rest stays intact.

**Drag-selection batch operations** — hold down the mouse button and draw a rectangle, everything inside gets selected. Works across 9 list scenes: conversations, apps, files, knowledge base, folder references, app discussions, attachment dialogs, prompts, memories. Same interaction as a file manager.

Also shipped: pinned conversation tabs, Python script app type, global search "All" category, logout all devices.

---

Other stuff I've built:

- 300+ models answer side by side
- Git-style conversation branching
- Branch tree visualization panel
- Message outline navigation
- Knowledge base + memory injection
- Background streaming for chats & AI apps
- Global semantic search + ⌘K quick actions
- AI generates web apps through conversation

yeero.ai if you want to check it out. Happy to answer questions or hear feedback.

Hey IH,

Working on YeeroAI (an AI collaboration tool) and shipped two small things in the last two weeks, both around how you pick a model:

**Default sort is now "most popular"** — the model list used to default to upload order, but I always ended up scrolling past the bulk to reach the ones I actually use. Switched the default to popularity by usage. The old default is kept as a "Latest" option. Six sort modes total: most popular / latest / earliest / longest context / cheapest / most expensive.

**Modality filter** — when there are hundreds of models, searching by name doesn't help when what you really want is "anything that takes image input" or "anything that outputs video". Added a modality filter row to two places: the global search (⌘K, on the "models" tab) and the "select model" dialog in chat. You pick text / image / audio / video / file for input and output independently. With an empty keyword + any modality picked, it lists matching models right away. Same component on both sides so the interaction stays consistent.

Both are small quality-of-life things, but they really cut down the friction of picking the right model.

---

Other stuff I've built:

- 300+ models answer side by side
- Git-style conversation branching
- Branch tree visualization panel
- Message outline navigation
- Knowledge base + memory injection
- Background streaming for chats & AI apps
- Global semantic search + ⌘K quick actions
- AI generates web apps through conversation

yeero.ai if you want to check it out. Happy to answer questions or hear feedback.

Hey IH,

Working on YeeroAI (an AI collaboration tool) and shipped two features this week that I'm happy with:

**Global search quick actions** — The search box (⌘K) used to only search history. Now it also shows shortcuts for common tasks: new chat, upload file, switch theme, jump to settings. It tracks what you use locally and puts frequent items first.

**App background generation** — If you close the page while an app is generating, it keeps running in the backend. Reopen and it catches up automatically. There's a "background streaming" toggle in settings. The apps list shows a badge on cards that are still running.

Both are small quality-of-life things, but they've saved me from losing work and clicking through menus.

---

Other stuff I've built:

- 300+ models answer side by side
- Git-style conversation branching
- Branch tree visualization panel
- Message outline navigation
- Knowledge base + memory injection
- AI generates web apps through conversation

yeero.ai — feel free to try it. Happy to answer questions or hear feedback.

Hey 👋 back again. Shipped a couple things I'm pretty happy with:

**Branch tree**

We got conversation forking (branch off any message). After like 5-6 branches though you just lose track. So threw in a side panel showing the whole tree — kinda like git log --graph. Click any node to jump. Parent/child relationships visible at a glance. Can't believe I didn't build this sooner, I use it every day now.

**Message outline**

Long chats = pain to scroll through. Now there's a little sidebar that auto-generates a one-liner per message, click to jump. Expand/collapse supported. That's literally it.

---

Previously shipped if you missed the first post:
- 300+ models answering in parallel, side-by-side compare
- Git-style conversation branching, fork from any node
- AI knowledge extraction + memory injection
- Background streaming — close the tab, AI keeps running
- Global semantic search
- AI-generated web apps, iterative via conversation

---

yeero.ai if anyone's curious — always down to chat.

Close the tab and AI keeps generating. Come back and pick up exactly where you left off. Built on Redis for multi-instance deployment.

Also added reasoning visibility — expand to see step-by-step thinking from o1/o3 models. Plus multi-modal input/output and a model marketplace with 300+ models.

Still bootstrapped, still shipping.