Our services: artificial intelligence

Neural networks can be just another machine learning approach if you wish but can also power automatic translations, language understanding, face recognition, auto-captioning of images, voice synthesis and much more. Business applications of AI are all around us; human-machine interactions in domotics, consumer goods like Amazon’s Echo (Alexa), autonomous drones, automated support (marketing) systems to name just a few. At Orbifold we have worked in the past ten years on AI by combining classic intelligence (e.g. semantic networks and Bayesian inference) with new neural network topologies (long-short term memory, stochastic layers). Qwiery is our take on AI and is both a showcase of our know-how in this domain as much as it’s a toolbox in our consultancy.

  • Integration of AI in your business

    We can integrate sentiment analysis in your marketing pipeline, extract intents and summarize your documents, infer categories or integrate Microsoft LUIS, couple Google’s AI to your CRM and more.

  • Natural language services

    Language understanding has exploded in recent years and combined with predictive analytics delivers new insights in structured and unstructured data. Trying to extract intents from mails, fine-tune marketing campaigns based on AI and forecasts, couple Twitter sentiment analysis to product releases, auto-detect transaction anomalies or secure voice-based payments…? We’re there to make it happen.

  • Neural networks

    Maybe you understand how your business can benefit from image recognition, face detection or auto-captioning images but you have difficulties to understand when to use LSTM networks or convolutions? How to master TensorFlow, Torch and Theano? Can one plug neural nets into traditional line-of-business apps? What’s the meaning of things like NVidia’s cuDNN, massive GPU processing in the cloud or word2vec?

  • Qwiery

    Qwiery is an open source platform delivering AI services and allows one to create intelligent pipelines. It combines traditional paradigms with modern AI approaches to deliver business solutions. We created Qwiery both as showcase for our AI know-how and as a toolbox for consultancy.

Our Business recipe to embrace AI

  • Business decision

    List the key strategic decisions that affect the business and the related indicators that need improvement (e.g. better customer satisfaction,  increase conversion, reducing claims and so on).

  • Identify AI areas

    Solving a particular business challenge very likely will involve more than one AI area. Ensure that you map all appropriate AI  areas (e.g., NLP, machine learning, image analytics) to the problem you want to address.

  • Think big, start small

    AI’s potential to influence decision making is huge, but you will need to build the right data, techniques, skills, and executive decision-making to exploit it. Have an evolutionary path towards more advanced capabilities. AI’s full power will become available when the AI platform continuously learns from both the environment and people.

  • Build datasets

    Create your own proprietary datasets for training and measuring the accuracy of your algorithms. For example, create your own proprietary database of “crash images” and benchmark the accuracy of your existing algorithms against them. You should consistently aim to improve the accuracy of the algorithms against comparable human decisions.

  • Pilots

    Build a pilot of your AI solution using existing vendor solutions or open source tools. Conduct parallel runs of the AI solution with human decision makers. Compare and iteratively improve the performance/accuracy of AI solution.

  • Scale

    Once the AI solution has proven itself, scale it with the appropriate software/hardware architecture, and institute a broad change management program to change the internal decision-making mindset.