The cost of ML
For most processes Machine Learning is not a feature, but an implementation detail. We believe that automation should not disrupt operations, increase costs or force the talent out of their comfort zone.
For the last ten years, every ML solution became obsolete within 6 months after being published. The only possible strategy, when using Machine Learning for work rather than research, is to constantly upgrade and remain flexible.
When choosing the right ML technology, it is necessary to consider and test every possible model and compare the results, rather than tweaking the data to match the expected results. Real data is complex but it should always be used unaltered.
AI Agents have no bias towards the ML technology that they use. They have no “favourite” algorithms and they are not afraid of failure. But they are relentless in finding the best solution for each statistical problem.
We explore the problem during several hour-long executive meetings, paying specific attention to the implementation issues and the process’ unique characteristics. Once a possible solution is found, we begin the cooperation.
We develop a general solution internally, using sample or public data. During this week we create the specific decision-making rules that allow the AI to understand the problem and find the right solution.
For the next month and a half we create a perfectly simulated work environment for the AI Agent, which utilises real data. There, the results can be observed, problems are corrected and the finished product is perfected.
Cost-saving and Reliable
Waypoint AI Agents are gradually deployed and scaled during a three to six months period, while the simulated work environment continues to operate as training for the AI when new Machine Learning models are published.
“ADA” is the first Agent developed by Waypoint AI, she is a specialist in Predictions using Live Cross-Sectional Datastreams (LCSDs). She is capable of autonomous reasoning when it comes to “connecting the dots” between independent processes and she is best used for predicting outcomes of situations where the user has limited or no control.
“BEN” is the newest Agent to enter production at Waypoint AI. He is most comfortable with Forecasting (What will happen) and Diagnostics (Why did it happen) of Time-Series Data. He is popular with consultants and engineers for the ability to independently “dive” into very large, messy Datasets and to find the dominant causes of a certain event.
A pure decision-making AI, rather than the more problem-solving oriented ADA and BEN, HAVOC replicates the ability of a human observing, deciding and acting. She is capable of activating and instructing the other Agents to generate predictions based on the decisions that she needs to make, acting as the Leader in any AI-controlled environment.
“PAN” explores the Problem-Space described by the works of A. Newell and H. Simon since 1972. He is based on the Delphi Method, also known as Estimate-Talk-Estimate, and he is the result of the founder’s work with prototype quantum computing systems.
Flexible Subscription Model
With our subscription model, you can use Waypoint AI as much as you want and when you need it. Everything is included by one simple monthly fee.
Pay as You Use
If you don’t want financial commitments our Pay as You Use provides flexibility and full control over spending. You will pay only for what you use and nothing more.
If you need something more complex and integrated with your systems, all you need to do is show us the problem. During the partnership we build and maintain a bespoke solution tailored to your needs.
Do you want to use AI in your business but not sure where to start and what benefits it can provide? We are here to help. We will work with you to find the best ways your business can grow using AI.
Datum Technology Services LTD
6th Floor - 2 Kingdom Road
Registered in England & Wales No. 12431124