It is easy to miss all the many details DEVONthink has to offer, even with a comprehensive manual that we refer to - I know I do. One thing that’s helped me recently (besides the forum here) is to use the new AI capabilities of DT4 for help. I just tried your question as a test (used gpt-oss:20b but you could use others). With the web-search enabled, it provided this very forum thread as the answer. If web-search is disabled, it still seems to have access to DEVONthink help manual. It was actually faster than the web-search test:
Is there a way I can set the default view while working with Markdown notes to be side by side, so I can see the source and the preview at the same time?
Yes! DEVONthink 4 lets you keep the Markdown source and the rendered preview side‑by‑side by default.
Just enable the split‑view option in the Markdown preferences.
How to set it
Open Preferences
⌘ + , → Preferences dialog.
Go to the “Markdown” tab (in the sidebar).
Check “Show preview pane in split view (source + preview)”
Close Preferences.
Now every time you open a Markdown note the editor will automatically display the source on the left and the live preview on the right.
Quick toggle (if you want to flip it on a per‑note basis)
In the Markdown editor, press ⌘ ⇧ P (or click the Preview button in the toolbar) to switch between single‑pane and split‑pane mode for the current note.
The setting you just enabled will remember the split‑view for that note type the next time you open it.
Need help customizing further?
You can also set the default view for a group(right‑click the group → Default View → Split View).
If you prefer a different layout for specific notes, use the View → Split menu while the note is active.
Let me know if you’d like a quick demo or if you run into any hiccups!
That’s great! For some reason I didn’t even see the side panel when looking at help.
What’s interesting as well though, is that if ChatGPT is selected as the “AI”, then I can ask a question, but if Ollama is selected, I can’t type into the prompt field.
Only commercial providers (except Perplexity) are supported and a model which is both suitable and inexpensive like GPT 4.1 (Mini) is automatically chosen.