# What can an AI personal assistant do for a business?

> Justas Butkus designs and deploys AI systems inside companies, mainly the company brain. He is the founder of AINORA, MB – the company behind the Ainora and Impetora brands – based in Vilnius, Lithuania, a graduate of Vilnius University, and a fractional AI officer for a limited number of companies.

Justas Butkus is a fractional AI officer based in Vilnius, Lithuania, working with mid-market companies and scale-ups across the UK, EU and US at two to four days a month. He builds and operates the production AI systems he advises on, and takes no commission, referral fee or revenue share from any vendor.

Video: AI Personal Assistant for Business: It Drafts, You Sign Off (https://youtu.be/TcYvYUVqM-Y)

**An AI personal assistant for a business reads every email, flags what is urgent, drafts the reply and logs the routine work, and a person signs off anything that affects a customer. We built one for our own team. Small decisions it makes on its own; customer-facing actions need an approval, and it only reaches what you give it access to.**

Canonical: https://justasbutkus.com/answers/ai-personal-assistant-for-business/
Last updated: 2026-10-30

---

## What is an AI personal assistant for a business, and what do the big ones miss?

There is a lot of hype about the big companies' personal assistants. They are amazing for simple use cases. If your business is a bit more complex, you might need to tweak one to get it to the level where it needs to be.

So I will tell you about our AI manager, which I think is truly next level: what it does, how you could build one yourself, and what it should look like if you were picking a vendor.

## How does it sort the inbox and flag what is urgent?

We use it to monitor all the activity, including all the mailboxes. It filters out spam, promotional emails and anything not relevant. Then it takes the urgent emails and prioritizes them.

If something is not very urgent, it goes into a queue. When a person asks for a report, the assistant walks them through the queue with suggested replies. If something is urgent, a team member gets pinged in our team chat, like a push notification, and goes to the app to check it.

## What does it do when it spots a problem?

We sell AI solutions, so we have automatic monitors on our own platform. They check that it is not down, that everything works, and the quality of people's experience: what was said on the calls, which automated actions ran, and all the errors.

If the manager spots an error, it can trigger a developer on its own to investigate and propose a fix, or to disregard it. If something is really urgent, it also pings us. For example, when a client's question keeps failing, it drafts an email to the client asking to expand the knowledge base, sends it with our sign-off, and pings us to check that everything is all right.

This way we are not stressed 24/7 about a new urgent inquiry or complaint. We know beforehand what breaks, we usually fix it ahead of time, and the customer experience is better.

## How does it help sales reps and keep every follow-up promise?

You can do the same in a traditional company, even one that is not as technological as ours. The least creative way is to connect your CRM and your email, and have your sales reps use it.

The CRM is filled in instantly. Emails are drafted in each rep's style, because it learns from the way you communicate, so you get consistent, concise writing in your own voice instead of generic chat-tool bloat. Reps probably draft emails with a chat tool already, but now it has proper context, and it logs in the CRM what was done and what was not.

It also tracks follow-ups. We track every promise we make: if we said we would reach out in the second part of the month and ask how it is going, we will do that. Our AI does it, and asks us first whether it is okay.

## What does the person still do?

The only thing the human needs to do is sign off: approve sending the email, or approve an update. For the small decisions the AI acts on its own. For actions that can have an impact on our customers, we use approvals.

The human is not out of the loop. They check the activity, and if something is wrong they give feedback, so the system is more precise next time. It is a recurring learning loop.

**What the assistant does alone and what needs a person**

| Task | Who acts |
| --- | --- |
| Filtering spam and promotional email | The assistant, on its own |
| Queuing non-urgent email with suggested replies | The assistant; a person goes through it on request |
| An urgent email or an error | The assistant pings the team |
| An email to a client | Drafted by the assistant, sent with our sign-off |
| A follow-up we promised | The assistant asks first, a person approves |
| A CRM update that affects customers | A person approves |

## What else can it take over, and where do I start?

Next are sales calls. Record the video or phone calls, put the transcriptions into the AI, and it writes down the key points, puts them into the CRM and schedules the follow-ups or agreed actions. It can inform the legal team of needed agreements too: in our case AI drafts them and a lawyer reviews. Anything your people do next to a screen can be automated.

It stays safe because it has no access to what you do not give it. It cannot send information out or do anything sensitive. You will not trust it on day one; over time you train it and put guardrails in place, until it is close to a virtual copy of you or your company's best practices.

If you do not know where to start, download the free prompt linked in the video's description and drop it into your chat tool. It asks about your current processes and bottlenecks and what you can automate. Our goal as a company is the most revenue per employee: with the same headcount, each rep more productive, you serve more customers and keep more profit.

## Frequently asked questions

### Can an AI personal assistant send emails on its own?

For small decisions, yes. Anything that has an impact on customers needs a person's approval: they approve sending the email or the update.

### Does this only work for a technology company?

No. You can do the same in a traditional company by connecting your CRM and your email.

### How can I keep it from doing things it should not?

It has no access to what you do not give it, so it cannot send information out or do anything sensitive. You also add guardrails over time, and you do not trust it fully on day one.

### How does it write emails in my team's style?

It learns from the way you communicate, so the drafts are consistent and concise and sound like you rather than like a generic chat tool.

### How do I find out what to automate first?

Use the free prompt linked in the video's description. It asks about your current processes and bottlenecks, and writing them down shows the highest return actions.

## Related

- [A company brain versus an AI assistant](/company-brain/vs-an-ai-assistant/) – Where a general assistant stops and a company system starts.
- [The 3 levels of AI adoption](/answers/3-levels-of-ai-adoption/) – Chat tools, systems on top of your work, and a company brain.
- [Who can see what](/company-brain/permissions/) – Access limits for an AI system, person by person.

## About the author

**Justas Butkus** – designs and deploys AI systems inside companies, mainly the company brain. He is the founder of AINORA, MB – the company behind the Ainora and Impetora brands – based in Vilnius, Lithuania, a graduate of Vilnius University, and a fractional AI officer for a limited number of companies.

## Want to see what an assistant like this would do in your business?

Describe your inbox and the follow-ups you keep dropping in a short email, and we will look at it together.

Contact: justas@ainora.lt · [LinkedIn](https://www.linkedin.com/in/justas-butkus/)