AI not responding or Ollama issues

You clicked an AI button. Nothing happened. Or you got an error. Or the spinner spun forever and gave up.

Here's the diagnostic flow.

Step 1: Is Ollama Running?

This is the cause 80% of the time.

Mac: Look for the llama icon in your menu bar (top-right of your screen). If it's not there, Ollama isn't running. Open Ollama from Applications.

Windows: Open Task Manager → Processes tab. Look for ollama.exe. If absent, Ollama isn't running. Open it from the Start menu.

Linux: Run pgrep ollama in your terminal. If nothing comes back, start it: ollama serve (or restart your ollama.service systemd unit).

Step 2: Can Anything Reach Ollama?

In your terminal, run:

curl http://localhost:11434

You should see Ollama is running. If you don't, Ollama is either:

• Not running (back to Step 1)
• Running on a different port (rare, but check OLLAMA_HOST environment variable)
• Blocked by a firewall (most often on Windows — temporarily disable to test)

Step 3: Is The Model You Selected Actually Installed?

In your terminal:

ollama list

You'll see all models you've pulled. Compare that list to whatever Time Garden is configured to use (Choosing Your AI Models).

If a model Time Garden expects isn't in ollama list:

ollama pull <model-name>

Step 4: Try The Model Directly

To verify the model itself works (not Time Garden):

ollama run <model-name>

Type "Hello." Watch it respond. If it does → the model works, and the issue is between Time Garden and Ollama. If it doesn't → the model is broken; try ollama pull <model-name> again.

Step 5: Check Time Garden's Settings

Open Obsidian Settings → Time Garden Plugin and scan for:

• License status (if Eternal): is your license valid? See your orders page if not
• Model dropdown for the failing operation: does it show the right model?
• Any error messages near the top of the settings panel

Step 6: Check The Console

Obsidian has a hidden developer console with helpful errors:

• Mac: Cmd + Option + I
• Windows / Linux: Ctrl + Shift + I

Click the Console tab. Click an AI button in your note. Watch for red error messages. Common ones:

• Failed to fetch → Ollama not reachable. Back to Step 1.
• Model not found → Model name mismatch. Step 3.
• License invalid → Eternal license issue.
• Out of memory → Model is too big for your machine. Pick a smaller one.

Step 7: Restart Everything

The classic, sometimes-needed step:

1. Quit Obsidian
2. Quit Ollama (Mac: menu bar → Quit Ollama; Windows: Task Manager → end task)
3. Re-open Ollama, wait 5 seconds
4. Re-open Obsidian → Time Garden vault
5. Try the AI button again

Step 8: If The AI Runs But Output Is Bad

That's a different issue — see Dealing with useless AI answers.

Step 9: When To Reset The Plugin

If nothing has worked, the nuclear option:

1. Open Obsidian Settings → Time Garden Plugin
2. Find the Reset Welcome Experience command (in the Command Palette: Cmd/Ctrl + P → type "Reset Welcome Experience")
3. Or run Reset License Data — useful if your license appears stuck

This resets local plugin state without deleting your journal data. You'll need to re-enter your license key after.

Step 10: Reach Out

If you've tried everything, write to the easter-egg form — Karl reads every message. Include:

• Your OS and version
• Your Ollama version (ollama --version)
• The model name that's failing
• Any error from the developer console (Step 6)

That'll get you a real human reply quickly.

Common Quick Fixes

Up Next

• Installing Ollama — full Ollama setup if you need to start over
• Dealing with useless AI answers — when AI runs but output is bad
• Performance & Slowness — broader performance issues