Skip to content

A brand that shows up every week.

Posts, carousels and visuals drafted in your voice and your brand kit, scheduled across your channels and reported on. Nothing is published until you say so.

A campaign idea, from brief to scheduled postExample
You

Announce the spring range. Friendly, short, one clear call to action.

You

You give a one-line brief. That is all the agent needs.

Consistency is the hard part.

Everyone agrees the brand should show up every week. Then the week happens. Ideas stall in drafts, visuals wait on design, and the calendar goes quiet exactly when it matters.

The work it takes off your team

  • 01Turning ideas into drafts for each channelBy hand today
  • 02Producing visuals and carousels on brandBy hand today
  • 03Checking every claim before it goes outBy hand today
  • 04Keeping the content calendar fullBy hand today
  • 05Scheduling and publishingBy hand today
  • 06Reporting what workedBy hand today

A team of agents and workflows, each with one job.

Each has only the tools its job needs and a budget checked every time it runs. None of them decides what a person should.

  • Content agentDrafts posts and captions for each channel, in your voice.
  • Visual makerProduces images and carousels in your brand kit.
  • Content calendarPlans, schedules and publishes what you approved.
  • Performance digestTells you each week what worked and what did not.
Connects to
  • Social channels
  • Brand kit
  • Image generation
  • Analytics
  • Shared drive
  • Newsletter

The calls that stay yours.

The agents do the work. These are the decisions they bring to a person, with the evidence beside each one.

  • 01

    Every post

    You see each post as it will appear on its channel, and approve or reject it.

  • 02

    The calendar

    What goes out, where and when, is yours to set.

Controls built into the work.

Not policies in a document: properties of how this department runs.

  • 01

    Claims checked against your sources

    A draft may say only what a source you gave says. A figure or claim it cannot ground is refused by name, before you ever see it.

  • 02

    On brand by construction

    Colours, type and layouts come from your brand kit, held as data, not from a prompt asking nicely.

  • 03

    Nothing published unapproved

    Publishing goes through a port that refuses anything without its approval.

Measured on the record, not estimated.

  • 01

    Posts approved

    Per week and per channel.

  • 02

    Brief to published

    How long a post takes to go out.

  • 03

    Drafts refused

    What could not be grounded, and why.

  • 04

    Engagement

    By channel, week on week.

Every run is recorded with its time, its steps, its decisions and its cost, so these are read from the record, not estimated.

The questions operators ask first.

Something we haven’t answered?

hello@dephrast.comAsk on a call
What exactly do you build?

Five sizes of the same thing. An automation takes one repetitive process off your team. An AI agent handles a defined business function with judgement. An AI system connects several agents and workflows. An AI department runs a whole function end to end, and an AI operating system puts every department under one orchestrator. All of them run on the same platform, with the same approvals and the same record.

Do we have to change our tools?

No. That is the point. We build against the CRM, accounting system, inboxes and spreadsheets you already run, through their own APIs. If something genuinely cannot be automated where it stands, we tell you before you pay for it.

What happens when an agent gets it wrong?

Anything that would reach a customer or leave your business starts behind a person’s approval, so a wrong draft stops there, not in someone’s inbox, unless you have chosen to let that kind of work run on its own. Every run is recorded step by step, so a wrong call is traceable the same day. Where a fact is missing, the work stops and names it instead of guessing.

Will the AI act on its own?

Only where you have allowed it, and only once its record shows it can. Every agent starts supervised; anything that leaves your business starts as a draft or a proposal a person approves. Running unattended is earned task by task, on your own records.

How much does it cost?

Builds start from €750 for a focused automation, €2,500 for an AI agent, €5,000 for an AI system and €10,000 for an AI department; enterprise programmes are priced per scope. Keeping a system monitored and improving starts from €250 a month. The real figure depends on scope, integrations, data, rules and support, so after a free assessment call you get a written scope and a fixed price before anything starts.

Find the first thing worth automating.

Bring us one process your team loses time on. We'll map it, show you what can be automated and what a Dephrast system would look like, whether or not you build it with us.

Free, 30 minutes, and a reply from a person within one working day.