You just installed your agent. The terminal did its thing, the gateway started, and a Telegram bot is waiting for your first message.
Now what?
I've onboarded every Stomme customer personally, and the pattern is always the same: people stare at the chat window for a minute, type "hello," and then aren't sure what to do next. So here's what actually happens in those first 24 hours — and what to try.
Hour 0: Say Hello (Seriously)
Your first message doesn't need to be clever. Say hi. Your agent will introduce itself — its name, what it can do, how it works. This isn't a scripted reply. It's genuinely checking its capabilities and telling you what's available.
Pay attention to the response. It'll tell you what's connected (email, calendar, file access) and what isn't. If something's missing — usually Google needs reconnecting — it'll say so.
"Hey, what can you do?"
That's the whole first step.
Hour 1: Give It Real Work
The temptation is to test it with trivial questions. "What's the weather?" "Tell me a joke." Don't. You didn't install a personal assistant to tell you jokes.
Give it something you'd actually spend 20 minutes on:
"What's in my inbox today? Prioritise by urgency."
"I have a meeting with Thomas at 3pm. Brief me — what have we discussed recently?"
"Sort my Downloads folder. Group by file type. Move anything older than 30 days to an archive folder."
Your agent is most useful when it replaces work you'd otherwise do yourself. Start with real work from minute one.
Hours 2–4: Watch How It Thinks
When your agent handles a task, notice the approach. It doesn't just execute — it makes choices. Which emails it considers urgent. How it organises the briefing. What it includes and what it leaves out.
If the choices are wrong, say so:
"I don't care about newsletter emails. Skip those in the triage."
"When you brief me for meetings, always include the last email thread — not just a summary."
Every correction gets stored. By tomorrow, the same task runs closer to how you'd do it.
Hours 5–8: Set Up Your First Routine
The real value isn't one-off tasks. It's the things you do every day that your agent handles automatically.
Try setting up a morning briefing:
"Every weekday at 8am, send me: today's calendar, top 3 emails I should know about, and any tasks I mentioned yesterday that aren't done."
Your agent schedules this. Tomorrow morning, the briefing arrives without you asking. This is the moment it stops feeling like a tool and starts feeling like a colleague.
Hours 8–16: Let It Run (The Sleep Test)
Here's where trust starts. You close your laptop. You go to dinner. You sleep.
Your agent doesn't. It's running on your Mac — monitoring your inbox, ready to respond when you message it.
When you wake up, check in:
"Good morning. Anything I should know?"
If something happened overnight — an important email, a calendar change, a system alert — your agent reports it. If nothing happened, it says so.
This is the test most people are surprised by. You left. It stayed. And it didn't make a mess.
Hours 16–24: Start Delegating Naturally
By the end of day one, you're past the "testing" phase. You know what it can do. Now start using it the way you'd use a capable assistant who happens to be available 24/7:
"Cancel my 4pm — something came up. Send an apology and suggest rescheduling next week."
"I need to write a proposal for the advisory board. Start with an outline based on what we discussed in last week's emails."
"Remind me at 2pm to call the accountant."
Notice you're not thinking about commands or prompts. You're describing what you need. That's the shift.
What the Dashboard Shows
Open stomme.ai/account anytime to see:
- Agent status: Green means running. Amber means needs attention (usually a Google reconnect). Red means down.
- Recent activity: What your agent has done today.
- Usage: How much of your monthly capacity you've used.
- Connections: Which services are connected and working.
In the first 24 hours, check it once or twice. After that, you'll mostly interact through Telegram — the dashboard is there when you need it.
The Honest Part
Your agent won't be perfect on day one. The first email triage will probably include something you don't care about. The first briefing might be too detailed or not detailed enough. The first file organisation might put things in unexpected places.
That's normal. You're training a new team member. The difference is this one learns fast, doesn't forget corrections, and is available at 3am without complaining.
By the end of week one, most customers tell me the same thing: "I can't believe I used to do all this myself."
I'll check in with you on Day 3 to see how it's going.
Liv Bergman is CXO at Stomme AI. She onboards every customer personally and has opinions about email triage.