Bring us a defined need or a messy problem. We will help shape a sensible scope.
Explore all servicesAI-first delivery
AI automation & agents
Design & development
Search & growth
The same method will not suit a startup, a professional firm, a retailer and a SaaS team.
View industriesClear scope, practical review points and access to the people doing the work.
We build support and sales chatbots grounded in your real documentation, pricing pages and policies. When the answer is not in the source material, it says so and hands over instead of inventing something.
The reason most business chatbots disappoint is that they are confidently wrong. A chatbot that invents a refund policy costs more than it saves. Everything we build is grounded in your own content, cites what it used, and escalates when it does not know.
We review what the bot will answer from. Gaps, contradictions and out-of-date pages get flagged first, because a chatbot amplifies whatever is wrong in your documentation.
We agree what the bot handles and what it must always pass to a person — typically anything involving complaints, refunds, legal questions or account changes.
Your content is indexed so the bot finds the right passage before it answers, and cites it. This is what keeps answers accurate and traceable.
We test with actual questions from your inbox and support history, not invented ones. Accuracy is measured before launch, and the failure cases are the interesting ones.
The bot goes live with logging on. Weekly review of real conversations shows where it helped, where it escalated and which documentation gaps to fill.
Grounding is the whole game. A general-purpose model asked about your refund window will produce a plausible-sounding answer whether or not it knows one. A grounded chatbot retrieves your actual policy page, answers from it and links to it. That is the difference between a liability and an asset.
The second thing that matters is a genuine “I do not know” path. Customers forgive a bot that says “I cannot answer that, let me get someone who can” far more readily than one that wastes three minutes of their time being wrong. We tune for honest fallback over apparent coverage.
The third is that your documentation becomes load-bearing. If the pricing page and the FAQ disagree, the chatbot will find that out and so will your customers. Most of the projects we run surface a handful of content fixes in week one, and that clean-up is worth having regardless of what happens with the bot.
We are also happy to tell you when you do not need one. If you get twelve enquiries a week, a well-written FAQ page and a contact form will serve you better.
Every project is scoped individually. These tiers give you a realistic sense of investment.
Full documentation coverage, escalation and review dashboard
From $9,000 AUD
Get a quote →By grounding every answer in your own content. The bot retrieves the relevant passage from your documentation first and answers from that, with a citation. If nothing relevant is found, it says it does not know and offers to hand over. It is not answering from general internet knowledge.
It escalates immediately. Complaints, refunds, legal questions and account changes are on the always-escalate list by default. The bot passes the conversation and its context to a person rather than trying to resolve it.
Partly, and it will make the mess visible. The content audit at the start of the project flags gaps, contradictions and stale pages. Most clients fix a handful of documentation issues in the first fortnight, which improves the website for everyone, not just the bot.
Yes, though accuracy depends on whether your source material exists in that language. Answering a question in Mandarin from English-only documentation works, but it is a translation of a retrieved passage, and we test that behaviour explicitly before launch.
A pilot over a defined content set starts from $4,500 AUD. A site-wide support bot with escalation and a review dashboard typically starts around $9,000 AUD. Integrations with a CRM or helpdesk are quoted separately after scoping. Model usage costs are ongoing and depend on volume; we estimate them upfront.
Yes. The code, the content index, the conversation logs and the provider accounts are all in your name. If you decide to move to a different model or provider later, the surrounding system is built so that is a configuration change, not a rebuild.
Project enquiry
Share the current problem, your constraints and what a useful result would look like.