An AI answer engine, built from your own website
Contexto reads everything you've published, learns how your organisation speaks, and answers visitors' questions with citations — in days, not months. Here's what your visitors get, what your organisation gets back, and exactly what happens under the hood.
Your best answers are already on your site. Your visitors just can't find them.
Knowledge-heavy organisations publish more than anyone can navigate: programmes, policies, events, working groups, research outputs, eligibility rules. As no-click search climbs above 60–70% in knowledge-intensive fields, the visitors who still reach your site are your highest-intent audience — they sought you out deliberately, for something specific.
A search box makes them do the work. A generic chatbot guesses, deflects, or contradicts your own policy. Contexto does neither: it reads your published content, works out who's asking, and gives them the precise answer your best staff member would — sourced, cited, and in your voice.
Not a chatbot. A concierge that actually knows your organisation.
Every one of these works out of the box — configured to your organisation during onboarding, not bolted on afterwards.
A direct, cited answer
Not ten blue links to scroll through — one clear answer, written from your published pages, with a citation to every source it used. If it isn't on your site, Contexto says so rather than guessing.
Page-aware
Contexto knows which page the visitor is reading. Someone on your clinical-trials page and someone on your careers page get different answers to the same words, because context changes meaning.
Audience-aware
A clinician and a patient asking about the same programme get different answers. Contexto asks a natural clarifying question — only when the answer genuinely depends on who's asking.
Remembers who they are
A visitor who identified as a researcher last week isn't re-qualified this week. Session memory picks up where the last conversation left off, so returning visitors skip the small talk.
Guided next steps
When someone signals real intent — a programme, a membership, a trial — Contexto surfaces the right call-to-action at the right moment, not a generic "contact us" in the footer.
Grounded in your content only
Every answer comes from your own website, never the open internet. No invented references, no hallucinated programmes, no answers that contradict your published policy.
Up and running in days. No engineering required.
No developers, no data migration, no lengthy implementation project. Contexto is self-service from day one.
You teach Contexto who you are
A guided onboarding — by voice if you like — captures your audiences, tone, priorities, and the answers that must never be improvised. Fully self-service.
Contexto reads your site
It indexes your published pages via your sitemap and CMS API. No uploads, no migration — your content is already there, Contexto just learns to navigate it perfectly.
One line of code and you're live
A single embed places the widget on your site, fully branded and configured as yours. Visitors start getting precise, cited answers immediately.
Your visitors start telling you things
Every conversation feeds back intelligence — who's asking, what they need, what your site doesn't answer yet — surfaced automatically, with no transcripts to read.
Yes, we require all grant recipients to deposit their research data in an open access repository within 12 months of project completion [1]. If the data contains sensitive human participant information, it must be properly de-identified first [2].
Every question a visitor asks is something they're trying to tell you.
The same conversations that help your visitors quietly become the best audience research you've never had to run. Contexto reports who is asking, what they actually want in their own words, which pages spark the most questions, and — most valuable of all — the questions your site couldn't answer. Those gaps become a content to-do list backed by real demand.
For the technically curious
You don't need to know any of this to use Contexto — but if you're the person who'll be asked how it actually works, here's the honest version.
Retrieval, two tiers
Your published pages are the first and priority tier. Optional trusted external sources are a second tier, consulted only when your own content can't answer. Retrieval boosts the visitor's current page so "what's on this page?" works.
Grounding & anti-hallucination
The model answers only from retrieved passages. It never invents names, dates or URLs, and every link in an answer is checked against your actually-indexed pages — anything unrecognised is stripped before the visitor sees it.
Inline citations
Answers carry numbered citations and source pills. Grouped or duplicate citations are cleaned up automatically, so what the visitor sees is a tidy answer with a clear trail back to your pages.
Smart model routing
Simple messages (greetings, quick replies) go to a fast, low-cost model; substantive questions with strong retrieval go to a higher-quality model. You get quality where it matters without paying for it everywhere.
Always-in-sync indexing
Contexto reads your site via its sitemap and CMS API — no uploads, no migration. A daily reconciliation compares content hashes and re-indexes only what changed, so the answers never go stale.
Built-in transparency
Contexto identifies as AI, carries an accuracy disclaimer and a privacy link, and runs on enterprise AI providers under data-processing agreements — the transparency the EU AI Act expects of an assistant like this.
Grounded answers are also safe answers.
Because Contexto can only draw on content you've already published and approved, it can't leak, invent, or contradict. Visitor data is stored in your region — Australia or the EU — never used to train AI models, and purged on a rolling retention schedule. EU visitors are asked before they're remembered.
How it works — in detail
How long does it take to go live?expand_more
Days, not months. Onboarding is a guided conversation you complete yourself, and indexing runs at up to around 10,000 pages an hour via direct CMS integration. Most clients are answering visitor questions within days of finishing onboarding.
Does it work with WordPress or our CMS?expand_more
Yes. Contexto discovers your content through your sitemap and, for WordPress, the REST API — pulling in taxonomy and custom fields as well as page text. It keeps the index in sync as your team publishes, so new and edited pages become answerable automatically.
Where do the answers come from — does it use the internet?expand_more
Only from content you have published. Contexto retrieves the most relevant passages from your own pages (and any trusted sources you explicitly enable) and writes the answer from those passages alone. It has no open-internet access at answer time, which is exactly what keeps it accurate.
What happens when it can't answer?expand_more
It tells the visitor honestly that it doesn't have that information rather than inventing an answer, offers a handoff to your team where you've configured one, and logs the unanswered question as a content gap you can see in your intelligence reports.
How does it stop the AI making things up?expand_more
Several layers. The system prompt forbids fabricating facts, names, dates or URLs; the model is instructed to only name people or link pages that appear verbatim in the retrieved content; and a post-processing step programmatically removes any link that doesn't match a page Contexto has actually indexed. Grounding is the core of the product, not a setting.
Can we control what it says?expand_more
Yes. During onboarding you set the tone of voice, audience personas, priorities and answer rules — including topics that must route to a human and things that must never be improvised. You can adjust these any time from your dashboard without redoing setup, and you can turn a single flagged conversation into a new answer rule in a click.
How does it know who is asking?expand_more
When an answer genuinely depends on the audience, Contexto offers natural choices (for example, "Are you a clinician or a patient?") as quick replies. The answer is remembered for the session via a cookie, so the visitor isn't asked again, and for logged-in users your site can pass known context so it never needs to ask at all.
What does it cost to run — will AI usage blow out our budget?expand_more
No. Contexto is fully managed: we pool AI usage across all clients, route cheap queries to cheap models, and filter bot traffic before it reaches the model. You pay one predictable monthly fee, never a metered AI bill.
Ready to turn your website into a conversation?
You’ll be speaking with Marameo Design — 15 years building digital platforms for health, research and university organisations.
30 minutes, no obligation