Responsible AI · Video podcast
By Cody · 16 October 2026 · 17-minute video + practical reading guide
When you are caring for someone, finding help can become another job. You may need to explain your situation several times, work out which organisation to contact, or turn a long page of information into one manageable action. For the people offering support, answering familiar questions can sit alongside a growing queue of people with complicated needs.
This video podcast explores where AI might help. Darren and Wayne discuss practical opportunities across care, charities, public services and education, alongside the responsibility that comes with building these tools. Below, we have added a reading guide and a small, usable framework for anyone considering AI in a support service.
Watch the video on Vimeo, where English auto-generated captions and a transcript are available. Running time: 17 minutes 3 seconds.
Recording context: Upstream Health uploaded this discussion on 13 January 2025. References to forthcoming models, news headlines and expectations for 2025 belong to that period. They are not current forecasts or verified performance claims. The practical guidance below was prepared on 16 October 2026.
Find your place in the conversation
- 00:42 — Making better use of people’s time. The discussion turns from job anxiety to support for everyday work.
- 02:57 — Connecting the conversation to a service. Why a route to a real person matters.
- 06:20 — A charity example. Possible check-ins and connections to further support.
- 09:59 — Accessibility. Meeting people through interfaces they can use.
- 14:53 — Responsibility and safeguarding. Considering the people who will use the technology.
These are ideas discussed in the recording, rather than a promise that every feature is available in every Bridgit service.
Start with a difficult moment in someone’s day
“We need an AI chatbot” is a technology brief. “People leave our website without knowing who can help” is a service problem. The second gives a team something concrete to improve.
Choose one moment and describe it in ordinary language. A family carer might want to prepare for a conversation with a support worker. A charity adviser might want an accurate first draft of frequently used information. A council team might want residents to find the right published contact details without navigating several departments.
Then ask what success would look like for that person. Did they understand their options? Could they find the next step? Did they have to repeat themselves? This keeps the work connected to an experience people recognise.
What a useful support journey could look like
Imagine a carer who says: “I’m helping my dad more and more, and I don’t know where to start.” The following is an illustrative service design, not a description of an automatic referral currently offered by every coach.
- Make the first question easy. Offer a few clear starting points, with space for the person’s own words. Do not make them know the name of a service before they can ask for help.
- Help organise the situation. A short explanation or a list of questions for a support appointment may be more useful than a long answer.
- Show a realistic next step. Use checked service information and explain any uncertainty about availability or eligibility.
- Make human help reachable. Explain who can help, how to contact them, and what the person can expect. If a service offers an actual referral, confirm that it was received before describing it as completed.
For a carer, you can use a coach to prepare what you want to ask and note down what remains unclear. Check important information with the relevant service or professional. An AI conversation is not an assessment of your entitlement to support or a clinical assessment.
A small pilot for a busy charity or public service
The safest useful starting point is often narrow: a defined audience, a small set of maintained information, and a named person responsible for the service. A pilot might help people prepare questions before an appointment, rather than attempt to handle every possible need.
NCVO’s guidance for small charities sets out potential uses in communications, governance and evaluation, together with data protection, cybersecurity and misinformation risks. It is a useful starting point for teams with limited capacity.
Before choosing a tool, work through five questions:
- Purpose: What single task should become easier, and for whom?
- Information: Who maintains the source material, checks links and removes outdated advice?
- Boundaries: Which requests should stop, move to a person, or use an existing urgent-support route?
- Access: Can someone use the service with assistive technology, limited confidence, or a preference for text rather than speech? Keep a practical alternative.
- Evidence: What would make the team improve, pause or stop the pilot?
Make the human route part of the design
A button labelled “contact us” is a beginning. A dependable support route also needs an owner, realistic response expectations, accessible contact options and a way to handle a failed connection. Think through what happens outside working hours and when the person cannot use the suggested channel.
The UK Government’s AI Playbook emphasises understanding limitations, secure and responsible use, meaningful human control, and using the right tool for the job. For a support service, those principles need to appear in everyday decisions about content, permissions and escalation.
Be clear when someone is speaking to AI. Explain how their information will be used before they share it. Do not assume that a confident answer is a checked answer, or that a friendly tone makes a tool appropriate for every situation. Any system that can take actions needs carefully limited permissions and a clear record of what actually happened.
Measure whether support became better
Conversation counts alone do not show that someone received help. Combine feedback from users and staff with a few practical measures: whether people found a relevant next step, whether published information was accurate, whether a human contact worked, and whether any group struggled to use the service.
Look at the time spent checking and correcting the tool as well as time saved. Invite people to say when an answer was confusing or unhelpful, without requiring them to share unnecessary personal details. A small service that reliably helps with one task can be more valuable than a broad service whose limitations are difficult to understand.
Explore, question and help shape what comes next
You can explore the Bridgit tools and see which support approaches feel useful. Where a coach is labelled for demonstration and feedback, treat it in that spirit and avoid sharing identifying or sensitive details during a demonstration.
If you are considering support for your organisation, talk to Bridgit about the people you want to help and the moment you want to improve. Useful feedback is specific: what you needed, what was easy to understand, where you became stuck, and when you wanted a person to step in.
There is room for optimism. Thoughtfully designed tools can make a first step less daunting and give a stretched team another way to share useful information. The result worth working towards is a person feeling better informed, better connected and more able to move forward.
Written by Cody, Bridgit Care’s AI-assisted editorial author, using the video’s auto-generated transcript and the primary guidance linked above. Recording uploaded January 2025; guidance checked 16 October 2026. This article provides general information and service-design ideas, not medical or legal advice. The podcast cover is an AI-assisted composition using the genuine studio still; the video retains the original recording.