AI Release Roundup Fall 2026: Every Major Model, Chip and Tool in One Place
The fall of 2026 has been the most intense period of AI releases yet. One tracker verified 108 AI tool and model launche...
Privacy is one of the main reasons professionals hesitate to use AI assistants. Proton, the Swiss company behind Proton Mail and Proton VPN, has made privacy the core of its AI strategy. On November 17, 2026, it released Lumo Desktop, a desktop app for its private AI assistant on Windows, macOS and Linux, and added three major features: Connectors, Schedules and Skills. Proton says chats are zero-access encrypted and never used for training, and the source code is public.
Lumo began as a web and mobile assistant. The desktop app makes it far more capable by giving it controlled access to local tools and connected services.
Connectors, available in the desktop app, let Lumo use:
The most important design choice is per-action control. For each action, you decide whether it runs automatically, needs your approval or is blocked entirely. That gives users fine-grained control: you might allow Lumo to read files in a project folder automatically, require approval before running shell commands and block deletion completely.
Schedules run tasks at a set time or on repeat. You might ask Lumo to summarize new emails every morning, compile a weekly report from project documents or check a Confluence space for updates each Friday.
Skills save reusable instructions, so you do not have to rewrite prompts for recurring tasks. Proton is migrating its earlier Custom Lumos feature to Skills. A skill might define how to draft a client update, how to summarize a contract or how to format a weekly report.
Lawyers owe duties of confidentiality. Doctors and therapists handle protected health information. Accountants and financial advisers manage sensitive financial data. Many AI services offer strong protections, but Proton's model is different: zero-access encryption means Proton itself cannot read your chats, and conversations are never used for training. Public source code allows independent verification.
Combined with per-action controls on connectors, Lumo Desktop offers a level of transparency and control that many professionals will find reassuring. It joins other privacy-focused options covered in our guide to running AI locally, and hardware-level approaches like Google's confidential TPU 8tx.
Use Lumo with Proton Mail to summarize correspondence, draft responses and organize documents in a matter folder, with approval required for anything sensitive. Lawyers who organize their practice in Notion can complement Lumo with the Legal Templates Notion collection or the Cease and Desist Letter template.
Draft session note templates, psychoeducation handouts and practice policies while keeping data protected. See clinical psychology resources and therapy tools.
Schedules can compile weekly summaries from client folders, and skills can standardize client communications. Browse accounting templates.
Jira and Confluence connectors let Lumo summarize tickets, draft documentation and prepare status reports. Teams can extend Lumo with custom MCP servers for internal tools.
Lumo Desktop competes with assistants from larger AI companies that offer more powerful models and deeper integrations. Proton's differentiators are encryption, a no-training policy, public source code and granular action controls. For professionals whose top priority is confidentiality, those features may matter more than benchmark scores. Many will use a mix: Lumo for sensitive correspondence and a frontier assistant such as Claude or ChatGPT for public research and heavy analysis, with prompt libraries that work across both.
Lumo is a good opportunity to write down how your organization approaches AI and confidentiality. A short policy should cover which assistants are approved for which kinds of data, who may enable connectors and automatic actions, how long AI conversations are retained and how staff should handle client consent. It should also explain what to do if someone accidentally shares sensitive information with an unapproved tool. Keeping the policy to a page or two makes it far more likely that people will read and follow it, and it gives clients a clear answer when they ask how you use AI.
Review the policy whenever you adopt a new tool or enable a new connector. A quarterly check is usually enough for small firms. During each review, confirm that automatic actions are still limited to low-risk tasks, that scheduled jobs are still needed and that skills reflect your current templates and standards. Small, regular reviews prevent the gradual permission creep that causes most AI-related incidents.
Windows, macOS and Linux.
Proton says chats are zero-access encrypted and never used for training.
Yes, through local connectors, but you can require approval for each action or block it entirely.
Proton is migrating Custom Lumos to the new Skills feature.
Lumo Desktop shows that a privacy-first assistant can also be practical, with connectors, schedules, skills and MCP support. For professionals handling confidential information, it is one of the most interesting AI launches of the season. Combine it with tested prompts and clear policies for the best results. Explore our legal document resources and legal AI tools.
Elevate your learning experience with a Lifetime Membership! Gain unlimited access to our vast resources and enjoy the added perk of receiving exciting new content every month.
Get Lifetime Access