A colleague that happens to be AI.
Squidler gives your company an AI colleague with its own email address and its own memory. You hand it work in plain words, and it does the work, also when you are not there.

You say what you want done. It gets done.
- 01
You hand it work. Tell it what you want done, the way you would tell a new colleague. By email, in Teams or Slack, or in a chat on your phone.
- 02
It does the work. It answers the email, chases the unpaid invoice, books the meeting and keeps the books in order, in the systems you already use. When something matters, it asks you first.
- 03
It learns your business. It remembers your customers, your prices and how you like things done. Correct it once and it keeps it.

How is that different from a chatbot?
Chatbots (ChatGPT, Claude, Copilot, Gemini) are assistants you ask. They are good at it, and we use them ourselves. But when the answer arrives, the work is still yours. A Squidler agent is a colleague with a job.
When it works
A chatbotWhen you open it and ask.
A Squidler agentOn its own as well: every morning, every Friday, when an email comes in.
What you get
A chatbotAn answer. The rest is up to you.
A Squidler agentThe job done: the reply sent, the meeting booked, the invoice chased.
What it knows
A chatbotWhat you remember to tell it.
A Squidler agentYour business. It builds up a memory of your customers, routines and decisions, shared by the whole team.
To be fair: these tools get better every month, and someone technical can make them do part of this. With Squidler it is the whole point, and it is set up for you.
The AI is the talent. Squidler is the employment.
The AI behind ChatGPT and Claude is like a very gifted person: well read, quick, good with words. A gifted person is no use to your company until they have a desk and a job. That is what we add.
An email address of its own
Forward things to it, or let customers write to it directly.
A calendar it works to
It starts its recurring jobs itself. Nobody has to remember to ask.
Keys to your systems
Email, calendar, documents and accounting, as far as you allow and no further.
A memory of your business
What it learns in one conversation is there in the next, whoever is asking.
Rules for when to ask
Before it emails someone new or deletes anything for good, it checks with you.
The longer answers.
If you already know your way around ChatGPT, Claude or Copilot, these are the questions you are likely to have. If you do not, you can skip them. Nothing above depends on it.

Which AI do the agents run on?
Mostly Claude, from Anthropic, which we reach through Amazon Web Services. Images, speech, video and music use models from Google and OpenAI. We choose the model for each kind of work and change it when a better one arrives. Your agent keeps its memory, its connections and its routines when that happens, because they are stored with us and not inside the model.
So how is Squidler different from Claude?
Claude is the model our agents think with, and Claude's own app is a very good assistant for one person at a screen. Squidler is what turns that model into a colleague for a company. The agent has its own email address and workspace, works to a schedule when nobody is at a screen, is shared by the whole team, and comes connected to systems such as Fortnox, Bokio and Spiris.
And from Microsoft Copilot?
Most people who have tried Copilot come away wanting more. It forgets: what you told it last week is gone. It is isolated: it sits inside one Microsoft program at a time and knows nothing of your accounting system, your Google documents or your Slack. And it is constrained: it suggests, where you wanted something done. A Squidler agent is built the other way round. It remembers your business, works across all your systems, and carries a task through to the end on its own.
Couldn't we build this ourselves with ChatGPT or Claude?
A technical person can get part of the way: connect a few tools, write the instructions, schedule some tasks. After that someone has to run it, keep the connections working, decide what it may do without asking, and keep one person's data apart from another's. That is the part we have built and look after, so a company without an IT department gets the same thing.
What can it do without asking us?
What you allow. For each system you connect, you choose how far it may go, for example read only, or read and draft but not send. Whatever you allow, it asks first before emailing someone new, deleting mail for good or using a tool marked as destructive.
Where is our data, and is it used to train AI?
It is stored in Germany, and no AI provider may use it to train their models. One setting keeps all the AI processing inside the EU as well.
What if we want to leave?
You can cancel at any time. Your data belongs to you, and we export it for you when you ask.
There is more on our security page, the list of what agents connect to is under Works with, and larger companies can read about agents for a whole team.
Easier to try than to explain.
From $50 a month. But your first two weeks: a free trial, no strings attached.