Tell him what you do every day. He builds it, tests it and runs it for you.
The same Sykik answers your questions, does the work in your apps, and turns what keeps coming back into a job that runs on its own.
Tell Sykik once what keeps coming back. He takes it off your day, so the work you love fits again.
His jobs start on their own: on a schedule, when a mail arrives or when a website changes. He works overnight on his own computer, so nothing waits for you.
No migration, no data export. Sykik signs in to the accounts your team already uses.
You decide what he may do on his own. Every step is visible, and you can pause any job at any time.
Sykik runs on Google Cloud in the EU. Switch your workspace to the EU region and every model call runs on EU infrastructure too — chats, reviews and scheduled runs alike.
Every team hands him its own routine: sales, service, finance, people, the warehouse. And everyone gets their day back.
You choose which tools each person’s Sykik can use and how much it does alone. They sign in and it’s ready to work, in their own accounts.
Marie’s Sykik sees Marie’s inbox, not Tobias’s.
Finance gets QuickBooks, Sales gets HubSpot. Change it any time.
Seats, tools and modes for the whole team, on one page.
Mention @Sykik in Google Chat. He does the work in your tools and answers in the same thread, with the result attached.
Models, data, approvals and prices.
A platform for AI agents that work inside the tools your team already uses. Everyone gets their own assistant that drafts and carries out work — on request, on a schedule, or triggered by an event such as a calendar entry or an incoming mail.
A chat assistant answers you in a tab. Sykik works in your accounts: it reads the thread, pulls the file from Drive, writes the draft and leaves it ready for approval. And Sykik is not tied to one model: a provider API, open weights, or your own — you decide per job, and it can run EU-sovereign where that matters.
Three routes, and you mix them freely: provider APIs (OpenAI, Google, Anthropic) on Sykik credits, open-weights models, or your own fine-tuned model on infrastructure you control — which is how the whole thing can run EU-sovereign. One thing worth knowing about credits: we do not send everything to a frontier model, we pick the model that suits the job. Rewriting a subject line does not need a top-tier model; analysing a contract does.
Sykik runs on Google Cloud in europe-west regions, encrypted with AES-256. Where inference goes to a provider outside the EU, the transfer is governed by standard contractual clauses and named in the privacy policy. With your own key or your own model, inference stays wherever you want it.
Every tool call is classified before it runs. Reads always run, reversible work from the draft tier up, irreversible work only on autopilot — otherwise Sykik stops and asks. Anything the classifier does not know asks in every tier.
Yes, in two ways: on a schedule (every Friday at seven) and on an event (a calendar entry approaching, a mail arriving). What comes out follows the same autonomy tier as everything else — on draft you find a draft in the morning, not a sent result.
No. Inference runs through a gateway we configure to disallow training use, and providers whose terms permit training on submitted data are excluded. Where providers offer zero data retention we use it; otherwise they typically hold data for 0–30 days for abuse monitoring. With your own key or your own model, whatever you agreed with your provider applies.
Two paid plans, both per seat. All-In includes the model bill as credits, including the work that runs when nobody is there. Bring your own model costs less and has you pay your provider directly. Current prices are on the pricing page at /landing/pricing.
Start with one thing that keeps coming back. He sets it up, tries it first and asks before anything unusual.

We currently onboard a small number of partners directly and set up the first workflow together.