About TLD Chat
A domain-focused AI companion that blends conversational workflows with live availability checks and WHOIS lookups, developed by ICI Innolabs.
Overview
TLD Chat helps you validate domain ideas without leaving the conversation. It pairs natural language chat with tooling for availability checks, WHOIS details, and structured task handoffs so you can work faster.
What's new? This release is a refactoring and modernization of a 2023 proof-of-concept, taking advantage of nativestructured outputs and function calling to deliver production-grade stability, scalability, and everyday usability.
Features at a Glance
- Conversational UI with persistent threads plus inline rename and delete actions.
- AI guidance tuned for registrar-style workflows via the Vercel AI SDK and OpenAI models.
- Convex-backed storage for messages, threads, and per-user token usage tracking.
- Clerk-protected authentication keeping each user's research private.
- Direct tool access for availability checks and WHOIS lookups inside the chat flow.
Tech Stack
- Frontend: Next.js 15 App Router, React 19, Tailwind CSS, shadcn/ui, Lucide icons.
- AI Layer: Vercel AI SDK orchestrating OpenAI models.
- Backend: Convex database and serverless functions for persistence and tooling.
- Auth: Clerk for identity, sessions, and secure middleware enforcement.
Evolution: 2023 Prototype vs Modern Refactor
Purpose
- 2023 Prototype: Benchmark function and tool usage under PN 23 38 01 01 at ICI Bucharest.
- Current Version: Provide dependable, production-ready domain research assistance.
Function Calling and Outputs
- 2023 Prototype: Pre-native support, reliant on LangChain orchestration and manual parsing.
- Current Version: Native structured tool calls with deterministic schemas.
UI and UX
- 2023 Prototype: SvelteKit interface tailored to researcher workflows.
- Current Version: Next.js 15 and React 19 chat UI with persistent threads and inline actions.
Reliability
- 2023 Prototype: Demo-grade availability; outputs not guaranteed.
- Current Version: Hardened flows, clearer error surfaces, and production deployment paths.
Auth and Privacy
- 2023 Prototype: Minimal authentication fit for lab experiments.
- Current Version: Clerk-secured sessions with per-user isolation and middleware guards.
Infrastructure and Models
- 2023 Prototype: LangChain plus Node.js with Ollama experiments (Llama 2, Mistral) and GPT-4o-mini.
- Current Version: Convex serverless backend paired with the Vercel AI SDK and OpenAI.
Scope
- 2023 Prototype: WHOIS insights, SEO name generation, and gTLD/ccTLD availability checks.
- Current Version: All prior capabilities with stronger guarantees, tracing, and token tracking.
Historical Context (2023)
The original TLD Assistant was a proof-of-concept showing how LLMs could call external tools such as WHOIS to surface actionable domain insights. It was developed at ICI Bucharest's Department of Software & Complex Systems Engineering as part of the National Core ProjectPN 23 38 01 01. Being a research demo, availability and accuracy were not guaranteed.