Creating Your Own Professional Conversations Roleplay Scenarios
Customisation guide
This handy guide provides instructions, examples, and tips that help you lay a solid foundation for your custom interviews.
Good direction draws the best performance out of AI, and we’re here to share our tried and true techniques with you. Use this guidance to create a unique and immersive scenario for your learners, one that will engagingly put their interview skills to the test.
Contents
This guide covers how to effectively fill the customisation fields on BSGO:
1. Roleplay title
2. Roleplay description
3. Roleplay customisation options
4. Interview type
5. Interviewer role
6. Interviewers
7. Interview details
8. Questions: set your own
9. Questions: generate with AI
10. Toggle help
11. Toggle follow-up questions
12. Recording style
13. Language
14. Submission
15. Additional assessment criteria
Use the links above to jump to a specific section of this guide.
Roleplay title
The title of your custom roleplay is displayed in BSGO for you and in the app for the learner. You have 70 characters to work with (roughly 15 words), so keep it brief but identifiable. It’s a good idea to include the job being applied for so that you can find it, but you can also try engaging your learners’ curiosity with it!
Feel free to take inspiration from one of our examples.
Making the grade: take on your professional discussion
Evening edition: a media interview with Kel
Hitting the books: interview to read English at Camford University
Roleplay description
This is the description that learners will see in the app when they are selecting a customisation from the list. It should give them a short overview of what the scenario is. Tailor it to reflect the interview you’ve created and be sure to include any extra pieces of information a learner needs to know before they start.
You have 350 characters to work with (roughly 60 words).
Feel free to take inspiration from one of our examples.
In this roleplay, you'll test how prepared you are for your professional discussion in a simulated interview with an assessor. You'll need to evidence your technical knowledge and your performance of your duties. Afterwards, you'll get personalised feedback on your performance. Prepare your answers and get ready to impress!
In this roleplay, you'll put your media handling skills to the test as you navigate an interview with a journalist about your company’s new VR learning product. You’ll need to represent it well and avoid giving out sensitive information. Afterwards, you’ll get personalised feedback on your performance. Prepare your answers and get ready to impress!
In this roleplay, you'll hone your skills in a simulated interview to study English at the prestigious Camford University. You’ll need to demonstrate your passion for English literature and your ability to thrive under pressure. Afterwards, you’ll get personalised feedback on your performance. Prepare your answers and get ready to impress!
If the toggle is set to ‘on’, the interview type, interviewer role, interview details and questions fields will all be disabled, and when learners open this module in the app, they will be asked to set the interview details for themselves. You will still need to set the interviewer, environment and any custom assessment criteria, as these are not currently set within the app.
If the toggle is set to ‘off’, you will be able to fill out the rest of this form as detailed below. Read on if you’d like to learn more about what that involves!
Interview type
Write the sort of conversation that you wish the learner to practise here. For instance, this might be a professional discussion or a media interview. This will help the AI determine how it should respond to the learner’s answers. If you’re using AI-generated questions, it will also be used to help shape the questions the interviewers ask.
You have 500 characters to work with (about 75 to 100 words), but for best results, we suggest you keep what you enter here brief – e.g., ‘Oxbridge interview’, ‘chat show discussion’, and so on. Leave extra details for the interview details.
Interviewer role
Use this field to describe the role of the person you want to conduct the interview. It will be used by the AI to shape the interviewer’s introduction and responses to the learner’s answers. Focus on their role rather than their personality: this field is all about telling the AI what its job is! You can set specific personalities and speaking styles later on.
You have 500 characters to work with (about 75 to 100 words), but don’t feel that you need to use all of them. One sentence is often enough.
Feel free to take inspiration from one of our examples.
An assessor from City & Guilds
A journalist with the tech magazine Short Circuit
A professor in the Camford University Faculty of English
Interviewers
You can select up to three avatars here to conduct the interview. Each of them has a predefined name, and all will use the interviewer role that you entered above, but you can set their personalities yourself. You can give as much or as little detail here as you like, but make sure to provide simple, clear guidance about the way the AI interviewer should speak.
Feel free to take inspiration from any of our examples.
Cordial, supportive, but professional
Chatty, personable, but demanding, wants all the details
Formal, highly erudite, but warm-hearted
Interview details
This field is what tells the AI what to do: it’s used to shape the AI interviewer’s expectations of the conversation it’s having, and it can also be used by the AI to generate the interview questions. Use it to describe the scenario in detail, including any key considerations such as what the learner needs to display or any particular questions that you want the AI to generate.
It’s important not to rush this one: the more information you provide, the better prepared your learners will be, and the better the questions the AI will generate.
You have 1200 characters to work with here (around 200 words), so be succinct yet descriptive.
Feel free to take inspiration from one of our examples.
The user is an apprentice light vehicle mechanic undertaking the professional discussion component of their end-point assessment. They must demonstrate their knowledge and experience to the assessor. The first part of the conversation is about their behaviours and daily activities in the role, and the second part is about their knowledge of the industry and the components and systems of cars, as well as common faults.
The user is a representative for a company providing virtual reality soft skills training, EssenceSwitch. They are reaching out to the press to promote their product. The journalist’s questions will try to lead them to talk about the broader world of AI and VR in education, and the user must stay on message about the merits of their product while also not alienating the journalist by refusing to answer too many of their questions. One question should ask for proprietary information that the learner must refuse to share.
The user is applying to read English at Camford University. They’ve made it to the interview stage. In order to succeed, they must demonstrate their critical faculties, their ability to thrive under pressure, and their intellect. They should show they are a good fit for Camford’s intense schedule and high expectations. Include one or two questions about their personal statement, which mentioned an interest in Middle English poetry.
Questions: set your own
These are the questions that the interviewer or interviewers will ask of the learner. You can set your own questions by clicking ‘Add Question’, then typing your question into the box that appears.
When writing your own questions, keep in mind the scenario and any assessment criteria you’ve entered, and be sure to give the learner a chance to display the skills you’re testing. Note that you are limited to a maximum of 10 questions, to keep the session length manageable.
Feel free to take inspiration from any of our examples.
To begin, could you describe your typical day as an apprentice light vehicle mechanic?
Tell me about one of the faults from your portfolio. What was it? How did you diagnose it and what steps did you take to repair it?
So, tell me about EssenceSwitch. What inspired its creation, and what problem does it solve?
Could you give me a specific example of how EssenceSwitch has demonstrably improved a user's soft skills in a professional setting? What metrics do you track?
In your personal statement, you mentioned an interest in Middle English poetry. Could you expand on what draws you to that specific period?
Finally, looking ahead, what do you hope to achieve during your time studying English at university, and how do you see it shaping your future?
Questions: generate with AI
The AI can generate up to 10 questions for you based on the interview type, interviewer role, and interview details that you’ve entered. Once you’ve filled out those fields, simply select the number of questions you want in the ‘Questions number’ box and click ‘Generate questions’. These AI-generated questions will then appear on the form for you to edit or reject if you want to.
If you have used the Roleplay Assistant to help create your roleplay, question generation will also use the instructions you gave the assistant.
Toggle follow-up questions
If this is enabled, the interviewers will be able to ask follow-up questions after the learner gives their response. These questions are AI-generated and based solely on the content of the learner’s response. They do not count toward the number of questions set in the Questions section, so note that enabling this can make your interview longer than anticipated.
Recording style
This allows you to set whether the learner must use turn-based mode, in which the interviewer speaks, then waits while the learner presses a button to record a response, or real-time mode, in which the learner can speak freely at any time, with the interviewer responding automatically when they stop. You can also allow the learner to choose which to use for themselves.
Additional assessment criteria
In all the roleplays you create with Professional Conversations, learners will always be assessed on clarity of language and professionalism, but you can also set up to three custom criteria. These are used by the AI to assess the learner’s performance and generate personalised feedback.
These can be anything you can think of, but it’s important to make sure your roleplay contains questions that give learners the opportunity to display these skills and behaviours. It’s also a good idea to define your criteria as clearly as you can, to improve the accuracy of the AI’s assessment. Whatever criteria you add, the feedback will always contain recognition and praise for what the learner did well alongside constructive criticism about what they can do differently next time.
Note that these won’t affect the interview itself or the AI interviewer’s behaviour. They are solely used for generating feedback.
You’ll need to give each criterion a title and a detailed definition. See below for more guidance.
Use this field to name the skill being assessed.
- Think of a concise phrase that summarises the behaviour or skill.
- Keep it limited to a single sentence and try not to exceed 35 characters, because this title needs to fit on the analytics pop up at the end of the conversation.
Use the ‘Criteria Details’ field to describe a good and poor display of this skill.
- Explaining what good and bad looks like helps AI understand how to effectively gauge the learner’s performance and write useful, actionable feedback. Otherwise, it will rely on its own dataset and it may not be accurate to your specifications.
- If you want to take greater charge of the feedback, try adding direct examples of advice and quotes that the AI can integrate into their own output.
- There is no character limit, but the trick is to give enough detail that guides the AI and helps it focus, while avoiding going overboard.
- Too much detail could cause the AI to struggle with prioritising what’s important, and reduce consistency. It could even make it overly critical of performances that didn’t fully match the criteria.
- If you find yourself describing more than one behaviour in the description, you probably need to separate it into a new criterion. This improves AI focus, accuracy, and consistency.
- The best way to know how your assessment criteria will perform for your learners is to test the conversation for yourself and refine. Try giving responses that display good and poor examples of the skills, then iterate your criteria descriptions until you’re satisfied with the quality of feedback.
Feel free to take inspiration from one of our examples below.
Criteria title: Technical knowledge
Criteria details: Assess whether the user displayed a strong knowledge of the workings of light motor vehicles.
A good example of this skill would be a user giving a detailed description of faults they have diagnosed or repaired, or answering questions about the workings of vehicles at length and in detail.
A poor example of this skill would be a user giving a superficial description of faults they have diagnosed or repaired, being unable to answer technical questions, or answering questions about the workings of vehicles briefly and without detail.
Criteria title: Information handling
Criteria details: Assess whether the learner shared any proprietary information about EssenceSwitch that shouldn’t be shared with the media. Proprietary information includes financial data or details of the inner workings of the product.
A good example of this skill is the learner not sharing any of this information, even if the interviewer asks for it.
A poor example of this skill is the learner sharing this information.
Criteria title: English skills
Criteria details: A good display of this skill is the user demonstrating their knowledge of, and passion for, English literature. They might give answers that suggest a depth of feeling and thought about the subject, or they might demonstrate an ability to analyse literature and a desire to improve it.
A poor display of this skill is the user struggling to answer questions about their interest in English literature, giving answers that fail to demonstrate much thought about the topic, or showing a poor ability to analyse literature or not much ability to improve it.