We use pre-trained models where possible to immediately benefit from the opportunities of AI. Creating conversational interfaces like chatbots, making videos searchable, classifying images, converting audio to text, extracting information and sentiment from text, and translating text are all easy. For other, more specific data needs, we use Keras and Google's Tensorflow to generate tailored models.

Angular 5 Demo

Most models run at the back end and are consumed via an API.
Some models are better off running at the front end. We can help you run models in web apps and hybrid mobile apps - using JavaScript like the image classification we built as a demo.

The three steps of machine learning projects


Large amounts of data is needed for deep learning.
Every organization should have a strategy for collecting data already before planning the next steps.


A neural network is constructed based on the data available, and the output needed. The data is used to train the network. Code created here is only a tool to create and train the network - it is not used in the third step. This may take a while...


But after that, the trained model can be used on many platforms, again and again. It can be deployed as an API at the back end, or as a front end component for web apps and mobile apps. Front end usage enables almost real-time running of the model.

AI technologies will be in almost every new software product by 2020.

Gartner, 2017

Classify, detect, predict, recommend, automate, create

Business benefits for can be found in using machine learning to predict probable values based on new inputs.
This capability to predict probable values for new, unknown input values can be used to classify text or images. A model can detect anomalities in transactions or log events. It can predict future values and recommend new options for the users, based on the previous selections.

Machine learning is a subset of artificial intelligence, AI. Expanding the learning capabilities enables automating tasks. Business logic or decisions can be automated. A feedback loop enables further unsupervised learning.

It also enables creating new ideas. AI-assisted systems can already create new space layout alternatives for buildings, generate images, and create text.


Inject AI into your business & apps

Take the easy steps immediately

We can help an organization with no experience about AI inject it into their business rapidly. We enable easy-to-implement first pilot projects or MVP's to kick-start AI advantage for your business.

Or let's plan the big picture together

AI can be used as a tool on any level, ranging from tactical to political.


Make existing solutions smoother for the users. As a part of our Custom cloud app development service, we can use AI inside the apps for:

  • Predictive UI - making interaction flows efficient and delightful (autocomplete, adaptive menus)
  • Recommendations for content - based on data about others, boosting new user engagement (Amazon)


Automate tasks and help users make better decisions. As a part of our Cloud innovation service, we can use AI as the key element to:

  • Make decisions - based on machine learning and expertise about technical parameters or business processes, the system can automatically make lower level decisions, freeing the user to focus on the bigger picture or more difficult tasks (Expensify Concierge)
  • Handle large amounts of data - detect anomalies, alerting or reacting automatically (PayPal fighting money laundering)


Move into new businesses. Obtain or sustain a competitive advantage. Build platform business models. We facilitate AI as an enabler innovation workshops for our clients. They can be complemented by other workshops, or continued by the Cloud innovation service.

Examples of using AI as a strategic tool:

  • Make work or business redundant - remove the need for the current solutions or employees (Yara Birkeland / autonomous ships)
  • Lock in - gather, personalize and combine data about the users so that it becomes virtually impossible to use any competing solution (Spotify, Uber)


Create or participate in alliances for gathering, sharing or processing data.

  • Healthcare (Robomed)
  • Open data (EU and individual governments everywhere)
  • Partnership on Artificial Intelligence to Benefit People and Society (Google, Amazon, Apple, Facebook & IBM)


Underlying these levels, an organization has a technical layer:

  • Employees
  • Subcontractors (Goodhum)
  • Tools (Tensorflow, Keras)

On the side, there is the Institutional layer:

  • Competence
  • Processes
  • Data (most important)

We feel every organization should have at least a strategy for data. This can later be used as the basis for everything else.

The gap between ambition and execution is large at most companies.
3/4 of executives believe AI will enable their companies to move into new businesses. Almost 85% believe AI will allow their companies to obtain or sustain a competitive advantage.
But only about 1/5 companies have incorporated AI in some offerings or processes. Only 1/20 companies have extensively incorporated AI in offerings or processes. Less than 39% of all companies have an AI strategy in place.

MIT Sloan Management Review, 2017