Live Demo · AI Document Search
This is a working demo of the kind of AI feature we build for clients. Below is a fictional company policy set. Ask it anything in plain English. No search terms, no folder-digging.
Northfield Digital Ltd, a fictional HR policy set. Five documents, plain-English search.
Try asking
Answer
Your question is sent to OpenAI to generate the answer. It is not stored or logged. The documents are fictional.
The same pattern, built into your existing software.
In a real engagement, the documents are yours: policy manuals, product specifications, knowledge bases, PDF archives or support material. The source formats, access permissions and update process are checked during scoping.
This isn't a standalone tool. It's a feature added to whatever you already have. A button in your internal portal, a tab in your CRM, a widget in your support desk.
A real build adds error handling, rate limits, cost controls, source access rules, testing and a fallback when the documents do not contain an answer.
Same process every time.
You describe the problem: what documents, who uses them, what outcome you want. I tell you what I'd build and what it costs.
Fixed price, fixed scope, fixed timeline. No ambiguity before we start.
I build it into your existing codebase or system. Short async check-ins. You stay focused on your work.
Working feature, walkthrough, documentation. One round of revisions included.
The same retrieval and analysis work, already shipped and in use.
Content strategy tool that crawls a site against its competitors and surfaces the exact content gaps costing it organic traffic. No signup, results in under three minutes.
Crawling · Data pipeline · No-auth tool UX
Case study ↗ · getcontentgap.com ↗Built and shipped a production AI platform with CV analysis, document processing, AI-assisted job matching, and career tools. A commercial SaaS with paying users, built entirely by me.
Next.js · TypeScript · LLM APIs · Supabase
Case study ↗ · jobspaceai.com ↗AI reading platform for UK primary school children, with personalised story generation, comprehension tools, progress tracking, and teacher dashboards. Built and launched solo, end to end.
Next.js · Supabase · Azure OpenAI · Stripe
Case study ↗ · primary-story.com ↗Questions about document search
Common sources include policies, manuals, product documentation, support articles and text extracted from PDFs. Suitability depends on document quality, permissions, volume and how often the source material changes.
This demonstration answers from its fictional policy set. A production assistant can be restricted to approved internal sources and should say when those sources do not contain the answer.
Yes. A production version can return document names, sections or quoted evidence so staff can verify an answer against the original material.
Access should follow the organisation’s existing permissions, with suitable authentication, logging, retention and provider settings. Different teams should not automatically see the same source set.
The index needs an update process so amended or withdrawn documents replace stale material. The right approach depends on where the source documents are managed.
Tell me what you have and what you want staff to be able to do with it. I'll tell you honestly whether it's a fit and what it would cost.