You want to put Machine Learning to work in your game, to increase players' engagement and LTV.

You are looking for a Real-Time solution ("in-game" actions based on real-time prediction & personalization), because acting while your users are playing drives the most efficient results:
- Actions personalization for a player with a high risk of churn
- Difficulty personalization for Casual games
- Ad Pressure personalization for Hyper Casual games
- Offer personalization (content, pricing...)

Your staff is already fully booked, and you are looking for a "work-for-hire" partner that can design, develop, test & tune the ML solution you need (and host the live solution if needed).

Your infrastructure is already fully loaded, with user acquisition management, ads, remote content, analytics, live ops... And you don't want to put it at risk working on a new ML project.

=> Our team can design and develop the ML solution you need, using our own askblu.ai ML-Ops platform.

=> When the solution is validated by your team, you can integrate it in your own infra-structure, or we can run it on the askblu.ai platform (pricing based on the DAU of your game).

1 - Data collection

We already have a working SDK (Unity/iOS/Android) that collects key events (sessions, stages, monetization, check the SDK documentation here).

If your project needs additional events specific to your game, we create a custom version of our SDK for your game, with the new events added.

Features Engineering

We already have the infrastructure to receive events and process them as player variables in real-time. (PREP module).

A specific PREP module will be create for your game, based on your player variables selected during the design phase.

2 - Machine Learning model

Our platform structure is modular: each solution for each game has its own ML model, that can be updated (re-trained) every day, week... depending on what's best for the model.

So our platform will host the ML model that we will develop specifically for your game, during the development and testing phases, and more if you wish (live usage of the solution).

3 - Real-Time inferences

Our platform contains all the modules needed to receive requests from our SDK (prediction request, personalization requests...), run the inference on the ML Model, and send the answer back to the game, and so on, for each type of request.

So a specific set of these modules will be developed for your game (and its custom SDK), and directly hosted on the askblu.ai platform, to process the solution during all the tests / debugging / data gathering / exploration phases.

4 - Delivery

When the project is finished, you get all the documentation and code (events -> features description,  ML model description, gitub access...).

Then you can decide how you want to run the live solution:
- you can host it on your own (we will assist you on the integration of the solution in your infrastructure)
- we can host it on the askblu.ai platform, and you will be billed based on your game's DAU on the platform.