Sectors
Answer Engineering for Social Care and Services for Vulnerable People
The short answer
When the person asking is a worried parent, an older person, or someone at risk, a wrong or careless answer can cause real harm. Answer Engineering makes your service information the source answer engines cite, and keeps any answer system on your site to approved wording, with safeguarding language detected before anything is said and personal details kept out of storage.
Where Answers Go Wrong in Social Care
- Families and carers are given eligibility or charging rules from another area, or from pages that are years out of date.
- How to raise a safeguarding concern is paraphrased, incomplete, or missing from the answer someone receives.
- A general chat tool replies to a person in distress with something that reads like advice, and no route to a human.
- Personal details typed into a vendor's chat tool are stored on a platform outside your control and your jurisdiction.
What We Engineer
Safeguarding Detection First
Messages are checked for safeguarding language before any answer is given. A match returns a fixed response your safeguarding lead has approved, with the offer of a person, and the visitor is never cut off.
Personal Details Kept Out of Storage
Email addresses, phone numbers, National Insurance numbers, NHS numbers, and dates of birth are removed from conversation records before anything is written to disk.
The Statements That Must Be Exact
How to report a concern, who is eligible, what a service costs, and how to complain, held in the Verified Answer Bank and given word for word.
Accessible to the People You Serve
Where the visitor's browser supports it, questions can be spoken rather than typed and replies read aloud, alongside the text. Pages we build meet WCAG 2.2 AA.
Working Alongside Your Safeguarding Lead
An answer system is not a safeguarding service and never replaces your procedures. Safeguarding detection works from a defined list of words and phrases that your team reviews. It is a prompt to route someone to a person quickly, not an assessment of risk.
Everything runs in the jurisdiction you choose, in retrieval-only mode by default, so nothing a visitor types is sent to an AI provider. We document how each deployment is configured so your safeguarding lead and data protection officer can review it before launch.
See it in practice
Built and Run by Us
My Chat Assistant
Our self-hosted answer engine, with safeguarding detection, personal data redaction, and human handover built in.
Questions
Frequently Asked Questions
What happens if someone discloses harm to the answer system?
Safeguarding language is detected before any answer is given. The system returns a fixed response your safeguarding lead has approved and offers a person. It is not a safeguarding service, and your own procedures always apply.
Is anything a visitor types sent to an AI provider?
Not by default. The system runs in retrieval-only mode and sends nothing to any third party. Connecting a provider is optional and your decision.
Can people who find typing difficult use it?
Yes, where their browser supports it. Questions can be spoken and replies read aloud, and the text version is always available.
Where is the data held?
In the jurisdiction you choose, including UK hosting or your own infrastructure.
Keep reading
Related
Start with a Confidential Chat
A no-obligation conversation with H D Fraser about how answer engines represent your organization today, and what a privacy-first approach would involve. NDA on request.