Skip to content

AI chat and search over your data

Assistants that answer from what you have already written.

We connect an assistant to your documents, web pages and video transcripts. It answers from them, shows its sources, and says “I don’t know” when the answer isn’t there.

Illustration: an assistant answering customer questions from a company’s help centre and courier guide. Sample data.

Illustration with sample data.

Scope

What we build.

Who it’s for

  • Support teams answering from a help centre or policy documents.
  • Experts and creators whose audience asks the same questions.
  • Professional services teams searching contracts, notes and files.
  • Answers from PDFs, Word files, web pages and YouTube transcripts
  • The sources behind every answer
  • “I don’t know” when the answer isn’t in your sources
  • Answers in your voice and style
  • A chat page, an embed or a widget on your site
  • Hand-off to your team when the assistant can’t help

Approach

How it typically works. Every project is different. This is the usual shape.

  1. 01

    Choose the sources

    We agree what the assistant should answer from, and what it must never use.

  2. 02

    Index and test

    We index your content and test answers on real questions from your team or customers.

  3. 03

    Tune the answers

    We adjust tone, sources and hand-off rules until the answers hold up.

  4. 04

    Launch where people ask

    It goes live as a chat page, an embed or a widget on your site.

Models and tools

Models and tools we use.

Models

  • Anthropic Claude
  • OpenAI GPT
  • Google Gemini
  • DeepSeek

Tools

  • Supabase
  • Postgres
  • Vercel AI Gateway
  • Next.js
  • TypeScript

Names are trademarks of their owners.

Questions

Common questions.

What if the answer isn’t in our documents?

The assistant says it doesn’t know, instead of guessing. You choose what happens next, such as a hand-off to your team.

Will our documents be used to train AI models?

We don’t use client data to train shared models without your agreement. We can sign an NDA and a data processing agreement before we see your data.

Which kinds of content can it answer from?

PDF, Word, Markdown and text files, web pages and YouTube transcripts. We check your formats during Discovery.

Can it sit on our website?

Yes. As a chat page, an inline embed or a chat widget.

Security and privacy

Careful with your data.

  1. Registered in Ireland

    Foldox Limited is an Irish company and follows the EU GDPR.

  2. Your data stays yours

    We don’t use client data to train shared models without your agreement.

  3. Agreements first

    We can sign an NDA and a data processing agreement before we see your data.

Other services

More we can build.

Start a project

Want answers from your own content?

Tell us what the assistant should know. We aim to reply within two working days.