# I have a personal assistant. Why am I still doing everything alone? > I have a personal assistant. And by the time I remember to ask him for something, I've already done the work alone. How to stick a request to an action that already exists in the day, check a small sample, and stop managing every shaliach one by one. - URL: https://shluchim.shinlamed.com/en/posts/2026-10-26-still-alone/ - Author: Seffy Fisher (ספי פישר) - Published: 2026-10-26 - Language: English (en) - Topics: Everyday AI, Grok Bot - Hebrew version: https://shluchim.shinlamed.com/posts/2026-10-26-still-alone.md - Arabic version: https://shluchim.shinlamed.com/ar/posts/2026-10-26-still-alone.md I have a personal assistant. And by the time I remember to ask him for something, I've already done the work alone. (Shluchim, singular shaliach, is Hebrew for "emissaries". It's what I call my AI agents.) You can give him work. You can talk to him from the phone. You can show him a screen. And still, inside the day itself, it's very easy to keep going exactly like before. Not because there are no tools. Because the request didn't stick to an action that already exists in the day. Without that, every time you have to rethink whether to turn to him at all. The day is full. There are messages, meetings, something stuck. The habit of asking for help has to get in there too, on autopilot. And when you do ask, sometimes what comes back isn't good enough. And then you have to explain again, fix, check. Great, more work for me. So the question is how you make daily use actually stick. ![Basher swallowing a screenshot sent to him midday](https://shluchim.shinlamed.com/images/posts/still-alone-20261026-tip-midday-1600.webp) ## The morning started without him Say I'm preparing for a meeting. I open tasks, look for what moved, remember I still haven't read that update. Only when I'm almost done do I notice I could have asked for help. That's the place where I want to bring Grok Bot in. Into an action that already exists in my day. In the [Grok Bot guide](https://shluchim.shinlamed.com/en/posts/grok-bot-guide/) I suggested a two-day experiment, with questions at times you set ahead: What are you doing right now? Can I help with one thing? You need to set the times and check that the routine was actually created. You can also do it without reminders. Every time I start preparing the weekly meeting, send him one message. That way the habit doesn't depend on my memory. It catches onto something that already happens. ## Midday, we stop for a second: A reminder, or a routine, that stops me in the middle. Not at the end of the day, when it's already too late to think. Take a screenshot on the phone or the computer. Send it as-is to the shaliach. And ask: "How can you help me with this? This is what I'm doing right now." No cleaning, no wording, no waiting for the perfect moment. That's how the reflex sticks. ## I don't have the energy to write him an operating manual Now I already remember he's there. But explaining what I'm doing feels like a project. And if it feels like a project, it won't stick. So you can start from a screenshot cleaned of sensitive info: "This is what I'm doing right now. I need to get to the meeting with a clear picture. What of this can you take off me?" Or from a voice note on the phone. With the confusion, the unfinished sentences and the "wait, actually". You can ask him to return his understanding in one sentence, and ask only what's missing to start. I don't have to arrive with a ready brief. ## I asked for a brief, I got a scroll Say a long, pretty document comes back. But for the meeting I need to know what's stuck, who's waiting on a decision, and what else to ask. "This is too long" is the start of feedback. You can push a little further: "Show me a sample on one person. Three sections: what moved, what's stuck, and what you need from me. For each section a link to the source. If there's no info, say so." Now there's something small to check and fix, before you prepare everything. And after it works, you repeat the same request without inventing it again. In [my Linear brief](https://shluchim.shinlamed.com/en/posts/2026-10-05-a-linear-brief-before-the-weekly-sync/) Basher prepared a sample. I approved it, and only then was the routine built. The quality of the output still needs checking. A good sample lets you find out early if we're aiming at the same thing. And then the routine already runs without me having to remember to ask. ## Wait, I already explained this At midday I come back to the task. And now I need to remind what we decided, where the file is, and which version was good. Here I'd ask at the end of the work to save a small summary inside the task: the goal, the decisions, the latest output and the next step. For me Linear (task software like Asana, Monday, Jira and the like) can be that place. Next time you can give a link and ask to read the comments too before starting. That way you don't start from zero every time. ## I became the work manager of the assistants If every request forces me to decide which shaliach to turn to, who designs and who checks, I'm in the middle of everything again. In the guide I wrote about the personal assistant as a central address. For me that's Basher. The request can stay simple: "I need to arrive at the meeting prepared. Pull the material together and come back to me if a decision of mine is missing." ![Basher beside a tidy kanban board: manage the work instead of managing every shaliach](https://shluchim.shinlamed.com/images/posts/still-alone-20261026-work-mgmt-1600.webp) ## I got an answer. The headache is still open After the meeting there's a draft, a decision waiting, and someone I need to get back to. In the chat it already says "done". In life, not yet. Here you can ask for a short summary: what was actually completed, what's waiting on me, and what's still blocked. Set a follow-up check for a specific task, and make sure it was set. Before sending to someone or taking a meaningful action, ask for approval. It matters that the assistant also knows where the work stops. ## And please, don't make me more noise If every little thing gets an update, now I have another chat to read. And then I simply stop opening it. You can set it ahead: Update when there's a result, a block, or a decision you need from me. Roll the rest up. If we set a daily check, come back with the things that need action. ![Kinneret with stickers on her face and hat: habits that stick](https://shluchim.shinlamed.com/images/posts/still-alone-20261026-habits-stick-1600.webp) ## So here's what I'm suggesting: Pick one piece of work that already repeats in the day. Stick a fixed request to it, small enough that you won't drop it. Check the result, fix it, then let it run again without thinking from scratch. You can start now: "This is what I do every time at this point in the day. What can you take off me, and how do we check you did it well?"