Laraue Blog — .NET, Nuxt, Telegram Bots and AI Workflow

Real code, real decisions, real tradeoffs. We write about what we build — .NET libraries, Telegram bots, AI integrations, and the architecture mistakes worth learning from.

DeepSeek API in .NET — an OpenAI-compatible client without an SDK, Ollama for development and AI that tidies up notes
Part 26 of the series about building a Telegram task tracker in a tiny team. How Boards got AI that turns a stream of thoughts from a chat into a neat task: DeepSeek in production, Ollama for local development, our own OpenAI-compatible client on HttpClient and the /aisave command in Telegram.
Read →
How we rewrote the frontend — a Nuxt architecture with explicit dependencies, Vitest tests and rules for AI agents
Part 23 of the series about building a Telegram task tracker in a tiny team. Why the Boards UI had to be rewritten, how the new frontend architecture works with every page declaring its dependencies explicitly, how that helps Vitest tests, and why the repository has a file of rules for AI agents.
Read →
How we run our backlog in Boards with Claude Code over MCP
Part 27 of the series about building a Telegram task tracker in a tiny team. We built an MCP server for Laraue Boards to understand how MCP works, then started using it with Claude Code every day. How the first version came together in two hours, what was wrong with it, and the daily loop it made possible.
Read →
Why reviewing AI-generated code is often more expensive than writing it yourself
The modern approach to development increasingly comes down to one claim — AI should write the code, and the developer just steers it and checks the results. The fact that reviewing generated code well enough to trust it is often slower than writing it yourself gets quietly left out.
Read →
Prototyping the interface with AI before writing code
Part 2 of building a Telegram task tracker in a tiny team. Why we start with the interface, how AI HTML prototypes replaced Figma, and how the first Laraue Boards prototype turned out.
Read →
How we build — engineering principles for working on real products
The engineering principles Laraue projects follow — start from the user path, do not overcomplicate ahead of time, split logic into layers, isolate third-party integrations, and where AI helps and where it is not allowed near the codebase.
Read →
How we decide what to build — validating a product idea in a crowded market
Validating a product idea before writing code — researching the market, testing the idea in public, and using AI to judge whether it is realistic to compete for the keywords you need.
Read →
Building an AI-Powered Real Estate Ranking System with C#, Ollama, and a Custom Crawler
A technical deep-dive into an open-source real estate aggregator for Saint Petersburg — covering the C# / .NET 10 architecture, Ollama vision model integration, custom crawler design, and the ideality scoring formula.
Read →
Using Ollama in C# and .NET — Local LLM Integration With Structured Output
How to call Ollama from C# and .NET: HttpClient with text and image input, structured JSON output, a typed NuGet adapter that generates the schema from a C# class, and how it compares with OllamaSharp and Microsoft.Extensions.AI. No cloud API required.
Read →
Laraue.Apps.RealEstate: Status, Limits and What Remains
The status of Laraue.Apps.RealEstate: an application that crawled two big listing sites and rated flats from their photos with a local vision model. 100,000+ listings, open source; the crawler is not launched now.
Read →
Building a Vocabulary Learning Telegram Bot with C# and .NET 10
A technical deep-dive into the architecture of an open-source Telegram vocabulary bot — covering the C# / .NET 10 stack, AI-powered auto-translation pipeline, data model, and local development setup.
Read →