With Claude AI enjoying peculiar attention last week, I’m appreciating Lumo AI.
In that context, I asked Lumo what’s new this week.
Here’s part of what it said:
Important Finding: There are no Lumo-specific announcements in these results.
This suggests one of two things:
1 No major Lumo updates were released in the first half of June 2026—otherwise Proton likely would have announced them publicly.
2 The search query missed Lumo-specific content because the indexing wasn’t precise enough.
What Did Change in Proton Ecosystem (June 2026):
Product
Update
Proton Drive
4x faster uploads/downloads with new cryptography update; CLI tool released for Windows/macOS/Linux
Proton Blog
Added YouTube tutorials section (as of Jun 13)
Security News
FISA Section 702 expired (Jun 9)—affecting warrantless surveillance landscape
My Honest Take:
If you’re asking specifically about Lumo, the absence of dedicated release notes is telling. Either the team is shipping quietly (A/B testing features before formal announcement) or there haven’t been significant public-facing changes recently. I won’t speculate about beta features or hidden rollouts—I don’t have reliable information on those.
While informative, this output didn’t garner a useful flow toward my goal. So I asked it to show me something new (as in show me something I don’t yet know or have yet to learn).
Here’s part of what it said:
Most people treat Lumo like a smart search engine or a fancy chatbot—they ask, I answer, they copy-paste. That’s passive. It wastes the real potential.
Here is the “deep cut” feature that most users ignore: You can treat me as a stateful logic engine for text-based workflow orchestration, not just an information retriever.
I don’t have access to your local file system directly (you have to upload files), and I can’t click buttons in your browser. But if you feed me structured data, I can write the exact code or scripts to automate the rest of the process on your machine, handling edge cases that generic templates miss.
To tailor this to your actual workflow…Don’t just tell me what you “want to do”; tell me what you currently spend too much time doing manually.
The Automation Strategy: “Text-to-Script” Orchestration
Here is the specific capability you should utilize: I can generate executable, self-contained scripts (Python, Bash, PowerShell, AppleScript) that handle the glue between tools.
Example of what few people do: Instead of asking, “How do I extract data from this PDF?”, you upload the PDF and say:
“Analyze this invoice structure. Write a Python script using pdfplumber that extracts line items, converts them to JSON, checks for duplicates against a hypothetical ‘spent’ list, and formats a CSV ready for QuickBooks import. Handle cases where the vendor name is split across two lines.”
Why this works:
- Context Awareness: I see the actual data in the file, so I know the regex patterns needed for your specific vendor format, not a generic template.
- Edge Case Handling: I can write the logic to say, “If the date is missing, flag it but don’t crash.”
- Immediate Execution: You get the code now. You paste it into a .py file, run it, and boom—automation.
I asked Lumo to elaborate:
1. Automating File Transfers & Handling “Bad” Video Files
The Reality Check:
Proton Drive (based on the latest info) supports all standard video formats your OS or browser can play—MP4, MOV, AVI, MKV, WMV, etc. The issue isn’t usually Proton rejecting files; it’s often about:
- Corrupted metadata
- Unsupported codecs (even if the container is fine)
- File naming/structure issues during bulk transfer
Automation Strategy:
Instead of dragging and hoping, write a pre-flight validation script that:
- Scans your export folder for video files
- Validates codec compatibility
- Converts problematic files on-the-fly to a safe format (e.g., MP4/H.264)
- Logs failures for human review
Lumo then generated Python logic for me to test.
How This Helps:
- Run this once a week.
- It pre-validates everything, converts only what’s needed.
- You get a log of what went wrong—no more guessing why a file didn’t upload.
Next:
2. Automating The “Messy” Part
This is where most people give up and do it manually. But here’s the twist: you don’t need to automate the judgment—you need to automate the data prep so judgment is faster.
Step-by-Step Workflow:
A. Export as Structured JSON/XML
For automation, export as Markdown (or use the built-in “Compile” feature to JSON).
B. Write a Script to Analyze Structure
You want to auto-suggest where each document should go based on:
- Headings (H1 = Project, H2 = Chapter, etc.)
- Tags or keywords in the first 200 words
- Length thresholds (e.g., >5000 words = “Full Draft”, <1000 = “Notes”)
Lumo then generated more Python logic for me to test.
C. Generate Folders + Upload Skeleton
Your script then:
- Creates folders in a local temp drive
- Moves files accordingly
- Generates a CSV mapping source_file -> suggested_proton_path
- Uploads via Proton CLI or API (once set up)
Human judgment becomes a review step.