Apply
DA
APPLY AGENCY / APPEARANCE

See us through your bias

Pick the colour you would rather think in.

↑ ↓ moves · Enter selects · Esc closes.

We stand on shoulders. And at the bar.

Same Apply Agency. Here are the people we owe a nod, and the bits we have to answer for ourselves.

Apply Agency Original
Apply Agency’s own look. Kind. Skilled. Hardworking. With black as the workhorse and a little colour in the personality.
Tokyo Night
A tribute to Omarchy, DHH and the community. The theme was created by Enkia; the palette here is inspired by folke’s TokyoNight port. They took the night shift. We brought the colours to work.
Sea swim & oat latte
Our own blend of sea air, blue ink and soft corners. For anyone who was in the sea at six this morning and would quite like to mention it.
Pistachio with a misprint
Our own palette with warm paper, green tones and Georgia in the headings. Like a small design magazine that has been given access to production.
Lars’ Friday theme
Our own cocktail, mixed for Lars: blue and white with a nod to F.C. København, lime in the glass and FREITAG by Magnus Millang on the stereo. Lars picked the theme. The calendar will have to fall in line.

Under the bonnet: Geist, Geist Mono and Geist Pixel from Vercel, plus Georgia as a system font. Adapted for Apply Agency with Codex. We take responsibility for the names ourselves.

 Contact
All services

02 · AI & agents

AI that does something, not just answers.

Apply Agency builds agents that can work with your own data and actions. They can help the user in the portal, support the editor or meet the customer in an AI client. It is not a chatbot on the side, but a controlled way into the work that is already happening.

If this is you

When the answer exists, but the work still lands with a person.

Many organisations already have the data and the rules. They are just spread across CMS, CRM, cases, documents and business systems. We make one useful task available to an agent at a time and measure whether it actually makes everyday work easier.

See it in practice

From question to action.

The same agent can help the customer in three ways. Choose a flow and see what makes it work.

Choose an agent flow
The customer asks

“What is the status of my case?”

“Can we move my appointment to Friday?”

“My case doesn't fit. Can someone help?”

Explore the agent's building blocks. Use the arrow keys to choose another one.
The selected flow Needs the customer's yes Choose a circle to read more Read about the selected circle
  1. 01The customer asks
  2. 02The agent looks it up
  3. 03The customer gets the status

An answer from the right source.

The agent finds the case with the customer's permissions and explains where it stands. The customer gets an answer based on a lookup.

  1. 01Find a free slot
  2. 02The customer approves
  3. 03The appointment is changed

A new appointment, once the customer has said yes.

The agent finds a free slot and asks the customer to approve it. Only then is the appointment changed through the controlled access.

  1. 01The customer asks
  2. 02The agent narrows it down
  3. 03A colleague takes over

A person takes over with the context.

When the task needs a judgement call, it goes to a colleague along with the question and the conversation. The customer doesn't have to start over.

An illustrative flow. We adapt sources, permissions and approvals to your task.

See it in Umbraco

AI in Umbraco: three features that give editors time

SEO copy in one click, an image that fits the page and Copilot as a proofreader. Three AI features for editors, shown in a 35-second film.

See the example

Private AI

Our own models on our own DGX Spark cluster, EU hosting at Scaleway and frontier models per task

We have bought a cluster of NVIDIA DGX Spark and run open models on a closed network. Three routes to a model, shown in two diagrams.

See the example

What we take responsibility for

01

One useful task first

We start with a concrete workflow and a result that can be assessed, not with a big AI plan.

02

Answers from your data

The agent looks things up in the sources you have approved and shows what the answer is based on when the task calls for it.

03

Actions with clear limits

Lookups, updates and next steps follow the same roles and permissions as the rest of the solution.

04

A person can take over

Uncertainty and complex cases are passed on with context. The agent should not bluff its way through the task.

The foundation

MCP is the bridge, not the whole product.

An MCP server can give several experiences the same controlled access. The portal is one way in. An in-app agent, an editor tool or the user's own AI client can be others.

  • MCP servers
  • In-app agents
  • AI clients
  • APIs and integrations
  • Knowledge and documents
  • Roles and permissions
  • Evaluation and testing
  • Monitoring

A typical agent flow

Status is only the beginning

A customer asks about their case. The agent looks it up with the customer's permissions, explains the status and can suggest a next step. If the customer approves, the agent can update, book or pass it on. Otherwise it stops and hands the case to a person with the conversation attached.

How we work

From opportunity to an agent in production

We make AI concrete early. Only when the task, the data and the limits are clear do we build the solution that will meet real users.

Find the value → Limit the access → Test against reality → Put it into production The next step follows on Choose 01 Limit 02 Test 03 Run 04 Quality in every change Human review and automated checks Review Test Release From the first decision to the next improvement
Find the value → Limit the access → Test against reality → Put it into production

  1. 01

    Find the value

    We choose a frequent or cumbersome task where better access to data or actions can make a real difference.

    You get A well-defined task and a goal we can test.

  2. 02

    Limit the access

    We decide which sources the agent may read, which actions it may perform and when a person should take over.

    You get Clear permissions and known stopping points.

  3. 03

    Test against reality

    We try the agent on real questions, awkward cases and the situations where a safe no is better than a clever guess.

    You get Documented behaviour before users meet it.

  4. 04

    Put it into production

    We track quality, errors and usage, and only extend the agent once the first task holds up.

    You get An agent that can grow at a pace you can keep up with.

Shall we look at your solution together?

Tell us where the self-service, the platform or the operations are pinching. Then we will have a short chat about what is worth solving, and what is not.

kontakt@apply.agency +45 70 60 55 04