Skip to content
AtheronLABS

You're visiting from the United States. Prices are shown in US dollars. Not right?

labs@atheron:~/industries/gaming/ai$ train agent --grounded

AI for game studios

Support agents and chat moderation models.

Game studios get two clear wins from AI: answering the flood of repeat questions after a patch, and reading chat in the slang your players really use. The first suits a frontier model working from your help content. The second needs a model fine-tuned on your own reviewed chat.

Start a projectSee the example projects

// what matters here

What is different in this industry.

01

Two jobs, two routes

A support agent on a frontier model starts fast and speaks many languages. A moderation model is fine-tuned on your chat, because only your logs teach it your game's slang.

02

Moderators decide the hard cases

Borderline messages go to a person. Their decisions are recorded and become the next round of training data.

03

Scored before it ships

Each model is measured against a held-out set of your own support threads or chat, and does not go live below the agreed line.

04

Fast enough for live chat

The moderation model is served close to your game servers so a message is scored before it reaches other players.

// example projects

Priced examples for this service.

Each one is hypothetical, labelled as such, and priced live from our rate card. Open any of them in the estimator and make it yours.

Example projectWeb appBot with a web dashboardAIFrontier model

Player support agent in Discord

Hypothetical. Not a client, not a result.

// The problem

After every patch, the support channel fills with the same questions about known bugs, refunds and account recovery, and players with real problems wait behind them.

// What we would build

A support agent in Discord and on the help site that answers from the studio's own knowledge base and patch notes, checks known issues and opens a ticket with the details filled in when a person is needed. It uses a frontier model through its API, which is the fastest start for public help content in many languages.

// What is in it

  • Answers from your help centre, known issues list and patch notes, with links
  • Questions answered in the player's own language
  • A ticket opened with platform, build and logs attached when a person is needed
  • Account actions only through your tools, after the player signs in
  • An evaluation set from real support threads, run before every change

// Stack

  • Frontier model API
  • discord.js
  • Node.js
  • PostgreSQL with pgvector
  • Helpdesk interface

// estimate

Build
≈ US$97,200 to US$149,000, delivered within 23 weeksCAD 138,400 to 211,700
Hosting
≈ US$558 a monthCAD 795 a month
Support
≈ US$2,500 a monthCAD 3,565 a month

Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.

AI route
Frontier model
Model running cost
≈ US$383 a monthCAD 545 a month

Timeline by milestone

Discovery
2.9 to 3.4 weeks
Commands agreed
0.4 weeks
Specification and evaluation plan
0.4 weeks
Design approved
2.4 to 3.4 weeks
Core features
1.4 to 1.9 weeks
Full build
0.9 to 1.4 weeks
Working in a test server
0.9 to 1.4 weeks
Working pilot
3.4 to 6.4 weeks
Testing and fixes
0.9 to 1.9 weeks
Launch
0.9 to 1.4 weeks
Production
0.9 to 1.4 weeks

Outside our hands, and added to the calendar

API access from your game team
1 to 4 weeks

How it is paid

Deposit 20%
CAD 27,680 to 42,340
Discovery 1.9%
CAD 2,629.60 to 4,022.30
Commands agreed 1.3%
CAD 1,799.20 to 2,752.10
Specification and evaluation plan 9.4%
CAD 13,009.60 to 19,899.80
Design approved 1.9%
CAD 2,629.60 to 4,022.30
Core features 5.7%
CAD 7,888.80 to 12,066.90
Full build 3.8%
CAD 5,259.20 to 8,044.60
Working in a test server 2.6%
CAD 3,598.40 to 5,504.20
Working pilot 28.3%
CAD 39,167.20 to 59,911.10
Testing and fixes 1.9%
CAD 2,629.60 to 4,022.30
Launch 3.2%
CAD 4,428.80 to 6,774.40
Production 10%
CAD 13,840 to 21,170
Holdback, 30 days after launch (10%)
CAD 13,840 to 21,170
Start from this example
Example projectWeb appAICustom LLM

Game chat moderation model

Hypothetical. Not a client, not a result.

// The problem

In-game and community chat is full of slang, misspellings and coded harassment that word lists miss, while harmless banter gets blocked.

// What we would build

A custom moderation model fine-tuned on the studio's own reviewed chat logs, so it learns the game's slang, scoring messages in real time and sending borderline ones to human moderators, served in the studio's own cloud account.

// What is in it

  • Messages scored in real time for harassment, hate, spam and scams
  • Fine-tuned on your own reviewed chat, in the languages your players use
  • Borderline messages queued for a moderator, not silently removed
  • Moderator decisions recorded and used for the next round of training
  • Scored against a held-out set of your chat before every release

// Stack

  • Fine-tuned open-weight model
  • PyTorch
  • Python
  • Kafka
  • AWS

// estimate

Build
≈ US$137,000 to US$209,000, delivered within 31 weeksCAD 194,900 to 297,300
Hosting
≈ US$7,020 a monthCAD 10,000 a month
Support
≈ US$2,500 a monthCAD 3,565 a month

Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.

AI route
Custom LLM
Model running cost
≈ US$6,530 a monthCAD 9,300 a month

Timeline by milestone

Discovery
2.9 to 3.4 weeks
Data and evaluation plan
0.4 weeks
Design approved
2.4 to 3.4 weeks
Core features
1.9 to 2.9 weeks
Full build
1.4 to 1.9 weeks
Training
6.4 to 11.4 weeks
Testing and fixes
0.4 weeks
Evaluation and red-teaming
3.9 to 4.4 weeks
Launch
0.9 to 1.4 weeks
Production
0.9 to 1.4 weeks

Outside our hands, and added to the calendar

API access from your game team
1 to 4 weeks

How it is paid

Deposit 20%
CAD 38,740 to 59,220
Discovery 1.5%
CAD 2,905.50 to 4,441.50
Data and evaluation plan 7.9%
CAD 15,302.30 to 23,391.90
Design approved 1.5%
CAD 2,905.50 to 4,441.50
Core features 4.6%
CAD 8,910.20 to 13,620.60
Full build 3.1%
CAD 6,004.70 to 9,179.10
Training 23.9%
CAD 46,294.30 to 70,767.90
Testing and fixes 1.5%
CAD 2,905.50 to 4,441.50
Evaluation and red-teaming 15.9%
CAD 30,798.30 to 47,079.90
Launch 1.5%
CAD 2,905.50 to 4,441.50
Production 8.6%
CAD 16,658.20 to 25,464.60
Holdback, 30 days after launch (10%)
CAD 19,370 to 29,610
GPU time for training and testing, at cost
CAD 1,200. Billed up front, at cost, outside the milestones
Start from this example

// questions

Questions we get about this.

Why fine-tune instead of using a frontier API?

Chat volume is high and the language is your game's own. A fine-tuned model in your cloud is quicker and cheaper per message at that volume, and it learns your slang. For a smaller game, a frontier moderation endpoint can be the cheaper start, and we price both.

What chat data does a moderation model need?

A sample of your chat that moderators have already reviewed, in the languages your players use. We help you label more where the model is weak.

Can the agent issue refunds or restore accounts?

Only through your own tools and rules, after the player signs in, and with a person approving anything you choose.

Back to the industry: GamingAbout the service: AI agents and models

// next

Not quite your project?

Tell us what you have in mind. We will come back to you with a range and the questions that would narrow it. Or book a call and talk it through.

Start a projectBook a call
AI for games: Discord support agents and moderation models | Atheron Network Labs