# AI Hosting UK: Apps, Chatbots, Models and Automation Source: https://dijituldns.co.uk/ai-hosting/ Updated: 2026-10-08 > dijitul hosts AI applications for UK businesses. Small apps that call a hosted AI API from PHP or Node.js can run on dijitul Complete Hosting at £1 a day + VAT. Larger workloads are quoted: RAG apps with vector databases, n8n automation, queue workers, and GPU servers for self-hosting open-weight models. API keys always stay server-side. ## Key facts - Small AI API apps can run on the £1 a day + VAT plan - RAG, vector databases, n8n and queue workers are quoted on VPS or dedicated servers - GPU servers for open-weight models are quoted by VRAM requirement - API keys stored server-side, never in browser code - Supports PHP, Node.js, Python and Docker deployments - dijitul's agency can build the AI application as well as host it ## Most AI apps don't need a GPU A common assumption is that AI hosting means expensive GPU servers. For most business uses, it doesn't. If your app sends requests to a hosted model through an API, such as those from OpenAI, Anthropic, Google or Mistral, the heavy lifting happens on their hardware. Your server just needs to run your code, keep your API key secret and store your data. That kind of app can often run on [dijitul Complete Hosting](https://dijituldns.co.uk/complete-hosting/) at £1 a day + VAT. See [AI chatbot hosting](https://dijituldns.co.uk/ai-chatbot-hosting/) and [API hosting](https://dijituldns.co.uk/api-hosting/). ## AI workloads we host | Workload | What it needs | Offer | | Website chatbot calling an AI API | PHP or Node.js, server-side key, rate limits | £1 a day + VAT | | LLM-powered web app | App server, database, background jobs | Quoted | | RAG over your documents | Vector database, embedding jobs, more RAM | Quoted | | n8n automations | Always-on process, database, secure credentials | Quoted | | Queue workers for AI tasks | Redis, process supervisor | Quoted | | Self-hosted open-weight models | GPU with enough VRAM | Quoted | ## Why the setup matters - **Keys stay secret.** An API key in browser JavaScript can be copied and used to run up your bill. We keep keys in server environment variables, outside the web root. - **Long tasks don't block visitors.** AI calls can take many seconds. Background workers and queues stop them tying up your web server. - **Costs are capped.** Rate limits and usage logging stop a bot or a bug spending your API budget. - **Data is handled properly.** You need to know what personal data goes to an AI provider and say so in your privacy notice. See [UK GDPR and hosting](https://dijituldns.co.uk/gdpr-compliant-hosting/). ## Questions we'll ask - Are you calling a hosted AI API, or running a model yourself? - What's the app built with: PHP, Laravel, Node.js, Python? - Do you need retrieval over your own documents (RAG)? How many documents? - How many users, and how many AI requests a day? - Does any personal or confidential data go into prompts? - Do you need us to build it, or just host it? ## Build and host in one place dijitul is a web agency as well as a host. Our team at dijitul.uk builds custom software, including AI features, so we can design the hosting and the application together. Call 01623 650333 or email info@dijitul.uk. Read more on [LLM app hosting](https://dijituldns.co.uk/llm-app-hosting/) and [GPU servers](https://dijituldns.co.uk/gpu-servers/). ## FAQs ### Where can I host an AI app in the UK? dijitul hosts AI apps for UK businesses. Simple apps that call an AI API from PHP or Node.js can run on its £1 a day + VAT plan. RAG apps, n8n, queue workers and GPU servers for self-hosted models are quoted. ### Do I need a GPU to host an AI chatbot? No, not if the chatbot calls a hosted model through an API such as OpenAI's or Anthropic's. The model runs on their hardware. You only need a GPU to run an open-weight model on your own server. ### How do I keep my OpenAI API key safe on my website? Never put it in front-end JavaScript. Store it in a server environment variable, call the API from your PHP or Node.js backend, and add rate limiting. dijitul sets hosting up this way so keys stay server-side. ### Can I self-host a large language model? Yes, using open-weight models with tools such as Ollama, vLLM or llama.cpp. You need a GPU with enough VRAM for the model size you choose. dijitul quotes GPU servers sized to the models you want to run. ### What is RAG in AI? Retrieval-augmented generation finds relevant passages from your own documents, usually using a vector database, and passes them to a language model with the question. It lets AI answer from your content. dijitul hosts RAG apps on quoted servers. ### Can dijitul build an AI app for my business? Yes. dijitul is a UK web agency at dijitul.uk as well as a host, and builds custom software including AI chatbots and LLM-powered tools. Hosting can then be specified alongside the build. ## Pricing and ordering This is quoted to fit. Request a quote at https://dijituldns.co.uk/quote/ or call 01623 650333.