A live status line, and friendlier messages.

Thu, 29 Oct 2026 00:00:00 GMT

Every SN3 AI Employee now has this update. Each one updated in about 20 seconds, and every update passed. One update reached every AI Employee at once. That’s the reason we moved everyone onto one version last week.

Live status line on Telegram

While your AI Employee works, one short line at the bottom of the chat tells you what it’s doing, like “Checking your email… 1m.” The line disappears when the answer arrives, and the usual “typing…” indicator still shows. It replaces the old “Still working…” messages.

Plain-English system messages

Notices like “busy” and “queued” now read in simple, friendly language. So do error messages.

Cleaner scheduled messages

Scheduled briefs and reports arrive without the technical header. They also show up in your AI Employee’s chat history now.

Better web search

Every AI Employee now uses an enhanced built-in search engine, so results come back faster and more reliably.

A quiet check-in

Say you haven’t messaged your AI Employee for a full working day. It looks over its own work first, then offers one to three things it could help with. If there’s nothing useful to say, it stays quiet.

Check-ins happen on weekdays only, between 9 and 5 in your time zone, and only on Telegram.

A 15-minute schedule minimum

Scheduled jobs now run at most every 15 minutes. A job set to run more often, every minute for example, is adjusted automatically.

What stays the same

Your scheduled jobs, connections, and memory are exactly as they were. AI Employees on Slack get the friendlier messages and better web search. The status line and quiet check-in are Telegram only.

We tested every change on our own AI Employees before it went out to clients. New clients who start this week begin on this version.

One version for every AI Employee.

Sat, 24 Oct 2026 00:00:00 GMT

This is the update we’re proudest of. For the first time, every SN3 AI Employee runs one current version, so from here one update reaches everyone at once. A few of our longest-running AI Employees jumped three months of improvements in a single step.

About the Railway incident

Railway reported a security incident on October 21. We reviewed everything connected to it and found no sign that anything of ours or yours was misused. If anything needed your attention, we’ve already contacted you.

Also in this update

Reliability

Quiet, safe updates. AI Employees update without sending “shutting down” messages. Each update is checked automatically and rolls itself back if anything fails. Conversations, memories, schedules, and connected apps all carry through.
Message search was rebuilt where needed, so finding past conversations is reliable.
Scheduled jobs get checked ahead of time to confirm they’ll run.

Smarter, newer models

Access to the newest models, including Claude Opus 5.5, Sonnet 5.5, Haiku 5.5, and GPT-6.1 Sol, with a very-long-context version for ChatGPT subscribers.

Connected tools

Live progress updates while connected tools work.
Optional cost tracking per tool, and spend limits per chat, for connected tool servers.

Better conversations

Long conversations hold together better, so your AI Employee keeps track over time.
Faster spoken replies. Voice can also use its own model.

AI Employee Changelog | SN3 Group