Not a peep
Hey Developers How about $2950
Moderators: Slitherine Core, Panzer Corps Moderators, Panzer Corps Design
Re: Hey Developers How about $2950
-
bondjamesbond
- Lieutenant-General - Nashorn

- Posts: 3305
- Joined: Mon Jun 27, 2022 10:10 pm
Re: Hey Developers How about $2950
Maybe then at least the source code will be released )
https://github.com/nival/Blitzkrieg
https://mynickname.com/id73473


Re: Hey Developers How about $2950
That makes no sense. Slitherine wouldn't give away their IP for nothing, especially when they supposedly want us to play PzC2 (which doesn't support Mac or iPad).bondjamesbond wrote: Fri Aug 15, 2025 7:17 amMaybe then at least the source code will be released )
https://github.com/nival/Blitzkrieg
Re: Hey Developers How about $2950
Maybe AI could help.
Reverse engineer the binary executable.
Implement a smarter AI opponent.
AI answer on game AI:
Absolutely. And I agree. Using GPT or Claude API calls for every AI turn would be the wrong architecture for this project.
A strategy game can involve hundreds or thousands of decisions. There is no reason those decisions should require a paid LLM call.
I'd build a completely local, conventional AI opponent. No tokens, no internet connection and no running costs.
## How I'd improve the Panzer Corps 1 AI
There are several established techniques from game AI research that are well suited to this type of game.
| Technique | What it achieves |
| ---------------------- | ----------------------------------------------- |
| Heuristic evaluation | Understands which moves are valuable |
| Pathfinding | Finds effective routes across the map |
| Threat assessment | Recognises dangerous enemy positions |
| Tactical search | Evaluates possible sequences of moves |
| Monte Carlo simulation | Estimates outcomes of uncertain situations |
| Strategic planning | Concentrates forces around important objectives |
These are all conventional algorithms. They run locally on your PC.
## An example
Imagine an enemy has three possible actions:
Battlefield decision
[Steam Community :: Panzer Corps](https://images.openai.com/static-rsc-4/ ... ose=inline)
[steamcommunity.com](https://steamcommunity.com/app/268400?u ... hatgpt.com)
The AI evaluates three alternatives:
Attack nearest infantry
Score 35
Attack exposed artillery
Score 85
Move armour towards objective
Score 72
The AI chooses the artillery attack because it offers the greatest strategic benefit.
Those scores are illustrative, but the principle is straightforward.
More importantly, we can make the AI evaluate the consequences of its actions rather than just choosing the highest immediate combat probability.
## What I would build
I'd use three levels of intelligence:
1. Strategic: What is the overall objective? Which victory hexes matter?
2. Operational: Where should forces concentrate? Which units should support each other?
3. Tactical: Which attack, movement or defensive action is best right now?
The challenge is connecting those three levels without making the AI excessively slow.
Your i7-8700 with 32 GB RAM is perfectly adequate for this kind of conventional game AI.
## One important technical question
Before designing this, I'd want to investigate how much of Panzer Corps 1's AI can actually be intercepted or replaced.
There are two possibilities:
- Modify or extend the existing AI through its scripting and modding interfaces.
- Build an external AI decision engine that communicates with the game.
The second is more flexible but considerably harder.
I'd suggest starting by examining the PC1 installation files and available modding tools. We can establish what is possible before writing any code.
And I particularly like the idea of making the AI smarter without making it cheat. It should beat you because it makes better decisions, not because it knows things it shouldn't or receives artificial combat advantages.
Reverse engineer the binary executable.
Implement a smarter AI opponent.
AI answer on game AI:
Absolutely. And I agree. Using GPT or Claude API calls for every AI turn would be the wrong architecture for this project.
A strategy game can involve hundreds or thousands of decisions. There is no reason those decisions should require a paid LLM call.
I'd build a completely local, conventional AI opponent. No tokens, no internet connection and no running costs.
## How I'd improve the Panzer Corps 1 AI
There are several established techniques from game AI research that are well suited to this type of game.
| Technique | What it achieves |
| ---------------------- | ----------------------------------------------- |
| Heuristic evaluation | Understands which moves are valuable |
| Pathfinding | Finds effective routes across the map |
| Threat assessment | Recognises dangerous enemy positions |
| Tactical search | Evaluates possible sequences of moves |
| Monte Carlo simulation | Estimates outcomes of uncertain situations |
| Strategic planning | Concentrates forces around important objectives |
These are all conventional algorithms. They run locally on your PC.
## An example
Imagine an enemy has three possible actions:
Battlefield decision
[Steam Community :: Panzer Corps](https://images.openai.com/static-rsc-4/ ... ose=inline)
[steamcommunity.com](https://steamcommunity.com/app/268400?u ... hatgpt.com)
The AI evaluates three alternatives:
Attack nearest infantry
Score 35
Attack exposed artillery
Score 85
Move armour towards objective
Score 72
The AI chooses the artillery attack because it offers the greatest strategic benefit.
Those scores are illustrative, but the principle is straightforward.
More importantly, we can make the AI evaluate the consequences of its actions rather than just choosing the highest immediate combat probability.
## What I would build
I'd use three levels of intelligence:
1. Strategic: What is the overall objective? Which victory hexes matter?
2. Operational: Where should forces concentrate? Which units should support each other?
3. Tactical: Which attack, movement or defensive action is best right now?
The challenge is connecting those three levels without making the AI excessively slow.
Your i7-8700 with 32 GB RAM is perfectly adequate for this kind of conventional game AI.
## One important technical question
Before designing this, I'd want to investigate how much of Panzer Corps 1's AI can actually be intercepted or replaced.
There are two possibilities:
- Modify or extend the existing AI through its scripting and modding interfaces.
- Build an external AI decision engine that communicates with the game.
The second is more flexible but considerably harder.
I'd suggest starting by examining the PC1 installation files and available modding tools. We can establish what is possible before writing any code.
And I particularly like the idea of making the AI smarter without making it cheat. It should beat you because it makes better decisions, not because it knows things it shouldn't or receives artificial combat advantages.
Re: Hey Developers How about $2950
Probably, advance will be, when we let the AI fix all of this, as doing other people with other games.
-
bondjamesbond
- Lieutenant-General - Nashorn

- Posts: 3305
- Joined: Mon Jun 27, 2022 10:10 pm
Re: Hey Developers How about $2950
If you were to invest $2,950 in AI right now, it would definitely improve the bots' intelligence—and possibly even the graphics and sound, too.
https://problembo.com/ru/services
https://problembo.com/ru/services
https://mynickname.com/id73473


-
terminator
- Field Marshal - Gustav

- Posts: 6196
- Joined: Mon Aug 15, 2011 12:48 pm
- Location: the land of freedom
Re: Hey Developers How about $2950
bondjamesbond wrote: Sat Nov 14, 2026 1:51 am If you were to invest $2,950 in AI right now, it would definitely improve the bots' intelligence—and possibly even the graphics and sound, too.
-
bondjamesbond
- Lieutenant-General - Nashorn

- Posts: 3305
- Joined: Mon Jun 27, 2022 10:10 pm
Re: Hey Developers How about $2950
terminator wrote: Sun Nov 15, 2026 6:40 amaction-nvidia-demission-huang.jpgbondjamesbond wrote: Sat Nov 14, 2026 1:51 am If you were to invest $2,950 in AI right now, it would definitely improve the bots' intelligence—and possibly even the graphics and sound, too.
In the East, the penis is considered the devil’s saber. Meanwhile, in the West, if a man had a male lover, his stock would only go up! But that has very little to do with our modest topic!
https://mynickname.com/id73473






