November 12, 2025

ServiceStack v8.10

First-Class React + Tailwind for AI-First Development​

We're witnessing a fundamental shift in how applications are built. AI code generation has evolved from a novelty to a productivity multiplier that's become too significant to ignore. While AI models still require oversight for production backend systems, they excel at generating frontend UIs—compressing development timelines that once took months into days.

The Rise of Vibe Coding​

AI can now generate complete, production-ready UI code. This enables an entirely new development workflow that Andrej Karpathy has termed "Vibe Coding"—where developers iteratively guide AI agents to implement features through natural language instructions, where features can be iteratively prototyped, refined and improved within seconds instead of hours.

This AI-first approach is rapidly maturing, with tools like Cursor, Claude Code, and Codex becoming the preferred platforms for this new paradigm with new tools designed to get maximum effectiveness of AI models with sophisticated planning tools, focused models optimized for code generation and edits and agentic workflows that's able to solidifying each new feature iteration with tests, along with detailed documentation, planning, migrations and usage guides.

React & Tailwind: The AI Development Standard​

React and Tailwind have emerged as the de facto standards for AI-generated UIs. Every major platform for generating applications from prompts has converged on this stack including Replit, Lovable, Google's AI Studio, Vercel's v0 and Claude Code Web.

TypeScript​

Whilst TypeScript is often excluded in one-prompt solutions catering to non-developers, it's still a critical part of the AI development workflow. It provides a type system that helps AI models generate more accurate and maintainable code and TypeScript's static analysis also helps identify errors in the generated code which AI Models have have become really good at correcting—as such it's an integral part in all our React templates.

How ServiceStack Excels in AI-First Development​

Context is king when developing with AI models. The better the context, the higher the quality of generated code and ServiceStack's architecture is uniquely suited for AI-assisted development:

Declarative Typed APIs​

All ServiceStack APIs follow a flat, declarative structure—The contract is explicit and consistent and LLMs don't need to guess what APIs accept or return.

End-to-End Type Safety​

Context quality directly impacts generated code quality. ServiceStack's TypeScript integration provides complete static analysis of what APIs accept, return, and how to bind responses—giving AI models the full context they need. The static analysis feedback also directs models to identify and correct any errors in the generated code.

Zero-Ambiguity Integration​

AI models thrive on consistency. ServiceStack removes guesswork with a single pattern for all API calls:

• One generic JsonServiceClient for all APIs
• Consistent methods used to send all requests
• Consistent Typed Request DTO → Response DTO flow
• Uniform error handling

Intuitive Project Structure​

ServiceStack's physical project structure provides clear separation of concerns, with the entire API surface area contained in the ServiceModel project—making codebases easy for AI models to navigate and understand.

Minimal Code Surface​

Less code means fewer opportunities for errors. ServiceStack's high-productivity features minimize the code AI needs to generate:

• AutoQuery APIs - Flexible, queryable APIs defined with just a Request DTO
• AutoQueryGrid Component - Complete CRUD UIs in 1 line of code
• Auto Form Components - Beautiful, validation-bound forms in 1 line of code

These components are ideal for rapidly building backend management interfaces, freeing developers to focus on differentiating customer-facing features.

Modern React Project Templates​

We're introducing three production-ready React templates, each optimized for different use cases:

Comprehensive React Component Library​

All three templates leverage our new React Component Gallery—a high-fidelity port of our proven Vue Component Library and Blazor Component Library. This comprehensive collection provides everything needed to build highly productive, modern and responsive web applications.

This is enough to generate a complete CRUD UI to manage Bookings in your React App with the React AutoQueryGrid Component. or with ServiceStack's built-in Locode UI:

This will generate an empty MyApp.ServiceModel/<Table>.d.ts file along with stub AutoQuery APIs and DB Migration implementations.

Regenerate AutoQuery APIs and DB Migrations​

After modifying the TypeScript Data Model to include the desired fields, you can re-run the okai tool to generate the AutoQuery APIs and DB Migrations (which can be run anywhere within your Solution):

After you're happy with your Data Model you can run DB Migrations to run the DB Migration and create your RDBMS Table:

Making changes after first migration​

If you want to make further changes to your Data Model, you can re-run the okai tool to update the AutoQuery APIs and DB Migrations, then run the rerun:last npm script to drop and re-run the last migration:

Removing a Data Model and all generated code​

If you changed your mind and want to get rid of the RDBMS Table you can revert the last migration:

Which will drop the table and then you can get rid of the AutoQuery APIs, DB Migrations and TypeScript Data model with:

AI-First Example​

There are a number of options for starting with an AI generated Application, with all the Instant AI App Generators like Google's App Studio able to provide a great starting point. Although currently Professional Developers tend to use Cursor, Claude Code or Codex as their day-to-day tools of choice.

Use GitHub Copilot when creating a new Repository​

If you're using GitHub Copilot you can also use it to generate a new App from the Vite React template :

- React 19, TypeScript, TailwindCSS v4
- Persistence in IndexedDB/localStorage
- Recharts
- Vitest with React Testing Library

## Features
Dashboard
- Overview of total income, expenses, and remaining budget
- Monthly summary chart (line graph)
- Expense categories (pie chart)

Transactions
- Add/Edit/Delete income or expenses
- Date filtering/sorting

Budgets
- Set monthly budget goals per category
- Progress bars for spending vs. budget

Reports
- View past months
- Export

Implement Category Auto-Tagging

Allow specifying tags when creating a new transaction.
When users add a transaction, try to predict the tag from the Description, e.g:

Input: “Starbucks latte” → Suggests category: Food & Drinks
Input: “Uber to work” → Suggests category: Transport

Implementation:

Maintain a small local list of common keywords + categories.
Pre-fill category in the transaction form as the user types in the Description.

We use Python for rapid feature development, then sync the latest improvements to ServiceStack's AI Chat implementation via an automated script.

Beyond llms.py's core features, ServiceStack's AI Chat adds integrated authentication, allowing you to offer ChatGPT-like capabilities to your users while maintaining complete control over API keys, billing, and approved prioritized providers—ensuring Fast, Local, and Private AI access within your organization.

AI Chat Admin Analytics​

ServiceStack's AI Chat feature provides a unified API for integrating multiple AI providers into your applications. To gain visibility into usage patterns, costs, and performance across your AI infrastructure, the platform includes comprehensive chat history persistence and analytics capabilities.

Or by referencing the ServiceStack.AI.Chat NuGet package and adding the ChatFeature plugin:

services.AddPlugin(new ChatFeature {
    EnableProviders = [
        "servicestack",
    ]
});

AI Chat History Persistence​

Enabling chat history persistence allows you to maintain a complete audit trail of all AI interactions, track token consumption, monitor costs across providers and models, and analyze usage patterns over time that captures every request and response flowing through AI Chat's UI, external OpenAI endpoints and internal IChatStore requests.

Database Storage Options​

ServiceStack provides two storage implementations to suit different deployment scenarios:

DbChatStore - A universal solution that stores chat history in a single table compatible with any RDBMS supported by OrmLite:

services.AddSingleton<IChatStore,DbChatStore>();

PostgresChatStore - An optimized implementation for PostgreSQL that leverages monthly table partitioning for improved query performance and data management:

services.AddSingleton<IChatStore, PostgresChatStore>();

Both implementations utilize indexed queries with result limits to ensure consistent performance even as your chat history grows. The partitioned approach in PostgreSQL offers additional benefits for long-term data retention and archival strategies.

Admin UI Analytics​

Once chat history persistence is enabled, the Admin UI provides comprehensive analytics dashboards that deliver actionable insights into your AI infrastructure. The analytics interface offers multiple views to help you understand costs, optimize token usage, and monitor activity patterns across all configured AI providers and models.

The analytics dashboard includes three primary tabs:

• Cost Analysis - Track spending across providers and models with daily and monthly breakdowns
• Token Usage - Monitor input and output token consumption to identify optimization opportunities
• Activity - Review detailed request logs with full conversation history and metadata

These visualizations enable data-driven decisions about provider selection, model usage, and cost optimization strategies.

Cost Analysis​

The Cost Analysis tab provides financial visibility into your AI operations with interactive visualizations showing spending distribution across providers and models. Daily cost trends help identify usage spikes, while monthly aggregations reveal long-term patterns. Pie charts break down costs by individual models and providers, making it easy to identify your most expensive AI resources and opportunities for cost optimization.

Token Usage​

The Token Usage tab tracks both input (prompt) and output (completion) tokens across all requests. Daily usage charts display token consumption trends over time, while model and provider breakdowns show which AI resources consume the most tokens. This granular visibility helps optimize prompt engineering, identify inefficient usage patterns, and forecast capacity requirements.

Activity Log​

The Activity tab maintains a searchable log of all AI chat requests, displaying timestamps, models, providers, and associated costs. Clicking any request opens a detailed view showing the complete conversation including user prompts, AI responses, token counts, duration, and the full request payload. This audit trail is invaluable for debugging, quality assurance, and understanding how your AI features are being used in production.

FREE Gemini, Minimax M2, GLM 4.6, Kimi K2 in AI Chat​

To give AI Chat instant utility, we're making available a free servicestack OpenAI Chat provider that can be enabled with:

services.AddPlugin(new ChatFeature {
    EnableProviders = [
        "servicestack",
        // "groq",
        // "google_free",
        // "openrouter_free",
        // "ollama",
        // "google",
        // "anthropic",
        // "openai",
        // "grok",
        // "qwen",
        // "z.ai",
        // "mistral",
        // "openrouter",
    ]
});

The servicestack provider is configured with a default llms.json which enables access to Gemini and the best value OSS models for FREE:

{
  "providers": {
    "servicestack": {
      "enabled": false,
      "type": "OpenAiProvider",
      "base_url": "http://okai.servicestack.com",
      "api_key": "$SERVICESTACK_LICENSE",
      "models": {
        "gemini-flash-latest": "gemini-flash-latest",
        "gemini-flash-lite-latest": "gemini-flash-lite-latest",
        "kimi-k2": "kimi-k2",
        "kimi-k2-thinking": "kimi-k2-thinking",
        "minimax-m2": "minimax-m2",
        "glm-4.6": "glm-4.6",
        "gpt-oss:20b": "gpt-oss:20b",
        "gpt-oss:120b": "gpt-oss:120b",
        "llama4:400b": "llama4:400b",
        "mistral-small3.2:24b": "mistral-small3.2:24b"
      }
    }
  }
}

Clean, Lightweight & Flexible AI Integration​

ServiceStack's AI Chat delivers a production-ready solution for integrating AI capabilities into your applications with minimal overhead and maximum flexibility. The llms.json configuration approach provides several key advantages:

Unified Provider Abstraction​

Define the exact models you want your application to use through a single, declarative configuration file. This thin abstraction layer eliminates vendor lock-in and allows seamless switching between providers without code changes, enabling you to:

• Optimize for cost - Route requests to the most economical provider for each use case
• Maximize performance - Leverage faster models for latency-sensitive operations while using more capable models for complex tasks
• Ensure reliability - Configure automatic failover between providers to maintain service availability
• Control access - Specify which models are available to users in your preferred priority order

Hybrid Deployment Flexibility​

Mix and match local and cloud providers to meet your specific requirements. Deploy privacy-sensitive workloads on local models while leveraging cloud providers for scale, or combine premium models for critical features with cost-effective alternatives for routine tasks.

Zero-Dependency Architecture​

The lightweight implementation adds minimal footprint to your application while providing enterprise-grade AI capabilities. No heavy SDKs or framework dependencies required—just clean, direct performant integrations.

The servicestack provider requires the SERVICESTACK_LICENSE Environment Variable, although any ServiceStack License Key can be used, including expired and Free ones.

Learn more about AI Chat's UI:

Effortless AI Integration​

In addition of providing UI and ChatGPT-like features, it also makes it trivially simple to access AI Features from within your own App that's as simple as sending a populated ChatCompletion Request DTO with the IChatClient dependency:

class MyService(IChatClient client)
{
    public async Task<object> Any(DefaultChat request)
    {
        return await client.ChatAsync(new ChatCompletion {
            Model = "glm-4.6",
            Messages = [
                Message.Text(request.UserPrompt)
            ],
        });
    }
}

It's also makes it easy to send Image, Audio & Document inputs to AI Models that support it, e.g:

var image = new ChatCompletion
{
    Model = "qwen2.5vl",
    Messages = [
        Message.Image(imageUrl:"https://example.org/image.webp",
            text:"Describe the key features of the input image"),
    ]
}

var audio = new ChatCompletion
{
    Model = "gpt-4o-audio-preview",
    Messages = [
        Message.Audio(data:"https://example.org/speaker.mp3",
            text:"Please transcribe and summarize this audio file"),
    ]
};

var file = new ChatCompletion
{
    Model = "gemini-flash-latest",
    Messages = [
        Message.File(
            fileData:"https://example.org/order.pdf",
            text:"Please summarize this document"),
    ]
};

Ask ServiceStack Docs - Introducing AI Search​

We're excited to announce the new Typesense-powered AI Search, a powerful new feature bringing conversational AI capabilities to ServiceStack Docs.