Viking Chess (Hnefatavl) AI

Note: This analysis is based entirely on self-reported information from the project description. No third-party verification or historical data is available.

The description states that Viking Chess (Hnefatafl AI) is a digital version of Hnefatafl, also known as Viking chess. It is described as an asymmetrical strategy game where one side defends a king who must escape the board, while the opposing side has more pieces and aims to surround and capture the king.

The product includes two types of computer opponents:

1. A traditional rule-based AI that evaluates moves using programmed strategies.
2. A modern learning AI that trains locally on the player’s own computer using the graphics card.

It is not evidenced whether either version is currently functional or deployed for public use.

The author claims that this project differs from existing implementations by training locally rather than relying on centralized servers, allowing the AI to learn and adapt during gameplay.

They state their goal is not only to create a strong opponent but also one that continues to develop its playing style as it is used — making the player and AI improve together.

The positioning evolves from being a simple digital recreation of Hnefatafl into an attempt to bring modern AI attention to a lesser-known classical strategy game, aiming for strategic depth comparable to chess or Go.

This claim is self-reported and lacks evidence of traction or adoption.

Not evidenced. The description does not specify who the target users are beyond "players" of Hnefatafl or those interested in historical strategy games. No segmentation, personas, or user behavior data are provided.

Not evidenced. There is no mention of pricing models, monetization strategies, or any indication that the product has a commercial structure beyond personal development.

The author reports building the system using:

• Tools: ChatGPT, Codex
• Technologies: CUDA, DirectML, PyTorch, NumPy, SQLite, Tkinter, Python
• Algorithms: Alpha-Beta Pruning, MCTS (Monte Carlo Tree Search), Gumbel-MCTS, Reinforcement Learning, Neural Networks, Minimax

They also mention using self-play and feedback from human matches to train the AI.

However, there is no evidence of deployment, scalability, or production readiness beyond personal development.

Not evidenced. The description contains no data on:

• Number of users
• Game sessions played
• Revenue generated
• Customer retention
• Product usage metrics

The project appears to be in early-stage development, with the author describing it as a solo effort and an experimental prototype.

Not evidenced. No information is provided about competitors or market positioning within the broader strategy game or AI gaming space.

• Solo Development: The project was built by a single developer (Carsten Cederholm), raising questions about scalability, long-term maintenance, and resource allocation.
• Lack of Traction: No evidence of users, adoption, or revenue generation.
• Unverified Claims: All claims are self-reported without external validation.
• Unclear Commercial Viability: No indication of monetization strategy or business model beyond personal interest.

1. What is the current status of the AI training process? Is it fully functional?
2. Has the local AI been tested with real players, and what feedback has been received?
3. Are there any plans to release the product publicly or make it available for download?
4. How do you plan to monetize this project if at all?
5. What are the technical limitations of the current implementation?
6. Have you considered integrating multiplayer features beyond peer-to-peer?
7. What is your roadmap for future development and user engagement?