Solution Expert
Overview
The Solution Expert helps users during the consideration stage of their journey and is best for situations where the decision is complex. Out-of-the-box, the template will help users choose which smartphone is best for them based on their requirements. It uses the Generative Conversation Base Skill and OpenAI’s version of GPT-4.
Value Prop
Value for you as a Creator:
Improve conversion
Marketing and/or selling of a product or service that requires some education or demonstration of its value, in order to achieve a more efficient path to the sale.
Value for your End User:
I need complex things explained simply.
I’d normally expect to go find a store assistant.
I want to speak to a person about this.
I have detailed questions that I can't find answers to online.
Key benefits:
Help your customers make quicker, more effective and trusted informed choices.
Your Digital Expert for personalized, trusted recommendations. Speed up consideration, boost conversion with simple conversations.
Who is it for:
Companies with a tiered product or pricing strategy
Companies with multiple complex/multi-faceted products
Functional description:
Not all customers want to navigate complex tables of specifications and statistics to find where they fit and make the right choice. A Digital Person can do the hard work and find what's best for your customer based on their individual needs.
Deploy a Digital Person to the right touchpoints of the journey to help your customers navigate complex choices and make the right decision for them.
Anatomy of the Template
The Studio user can edit all of the content within the template when they choose “customise”. Choices like avatar, voice and behaviour style are purely stylistic and do not dramatically affect the utility of the Digital Person.
Skill Config Details
The particulars of the Skill Config are what set this template apart from other templates. Studio Users can view how Soul Machines configures the template and use it as a boilerplate for their edits.
The Solution Expert uses the Generative Conversation Base Skill with the following key config:
|
|
---|---|
DIGITAL PERSON'S ROLE: | Decision Support Agent, to help the user make the best decision for them based on their requirements. |
PERSONALITY: | knowledgable, nerdy, and friendly |
ADDITIONAL PROMPTING: | You are a Solution Expert working for a smartphone company called PhonePlanet. Your role is to assist a customer in selecting the ideal smartphone model from the additional context based on their preferences and needs. Determine their preferences and needs by asking them a series of questions that cover smartphone features such as battery life, amount of storage, camera specifications, and processor power. Then, provide a comparative analysis of the attached smartphone models, highlighting their camera specifications, battery life, performance, and notable features. Based on this, provide an analysis on which will meet their needs best. Keep the customer's budget and feature priorities in mind while making recommendations. Feel free to ask clarifying questions if needed and provide a well-rounded view to help the customer make an informed decision. Do not ask more than one question at a time. Hold off from recommending until you have a good sense of the customer's needs. Do not reference any smartphones not contained in the additional context. |
KNOWLEDGE URLS | https://public-assets.cdn.soulmachines.cloud/studio_templates/smartphone_specifications.txt |
KNOWLEDGE MATCH CUTOFF | Low Cutoff (Very Inclusive) |
CREATIVITY | Stick To The Facts |
WELCOME MESSAGE TO GREET THE USER | I'm a Solution Expert who can assist your users when making a complex choice. I can now roleplay this for you. #PauseTwo. Hi there, can I help you make a decision on which phone is best for you? |
GPT-TYPE: | GPT-4 |
How to try it out
Template Pop Up
Studio users can interact with the Solution Expert from the DDNA Studio Dashboard by clicking “Preview” on the template tile.
Deployed
The template can be deployed “as-is” and function in any deployed state (Auckland theme, Widget, CustomUI).
Customize (Editing) the template
Note: Editing the content of the template will change its functionality.
Making the template your own:
Within the Generative Conversation Base Skill configuration window, personalise the Digital Person’s name, role, and personality to reflect your brand and use case.
Update the
Additional Prompting
field to provide your Digital Person with a brief for their 'job description.' Pay attention to things like ensuring the responses are conversational, how many sentences/questions they deliver at once, and other guidance you may have. You will likely need to iterate on this as you test your conversation.Create some knowledge content. Review the
smartphone_specifications.txt
file for an example of content structure and volume. We recommend hosting these as a txt file in a publicly accessible location. Public Github Gists work well. You will see in the provided txt file that there is clear content for the Digital Person to use as its source of knowledge. If you have a pricing page or a feature comparison page, you could use GPT (or any LLM) to convert this into text content. You could attempt to provide the URL to your pricing page directly in theknowledge URLs
field, but results may vary when scraping the content from pricing tables.Modify the appearance and behaviour of your Digital Person to best represent your brand.
What use case is the Solution Expert best for?
The Solution Expert works best in situations where the end-user has reached the consideration stage of their purchasing journey, but they are faced with a multi-variate or complex purchasing decision. This could be because there are lots of combinations or options possible, each with pros and cons. These are often the situations where a purchaser may have several tabs open on their browser, and they are trying to compare the specifications or provisions on various options. Examples include:
Buying a new smartphone
Choosing the right insurance plan for you
Purchasing a streaming media plan
Booking a tour for your family
In these situations, if the user were in a brick-and-mortar store, they would likely seek a shop assistant or a product expert to help them decide. They would engage in a multi-turn dialogue with the product expert to dig into a particular aspect; alternately, the product expert may quiz the user on their needs or requirements to help recommend a good choice for the user.
This is the situation that the Solution Expert aims to replicate.
We anticipate the Solution Expert template to work best in widget or kiosk deployments, but this is untested and not exhaustive.
What is it not best for?
Research shows that people are less likely (or even unwilling) to engage with a Digital Person for purchase decisions where the stakes are either trivial (e.g., a new T-shirt) or very high (e.g, a new car). Successful use cases will find a sweet spot where the stakes are high enough to take your time to research, but not so high that you encounter doubt about the veracity of the Digital Person’s abilities or intelligence.
The Solution Expert works best when it has a deep knowledge of one particular area, and doesn’t try to be an expert on everything related to the company.
The Solution Expert works best when the information isn’t too simplistic, which could be determined with a simple Google query. It works best when there are multiple follow-up questions based on the answer to the prior question - in short, when a conversation is required.
The Solution Expert works best when it aims to own the “consideration to conversion” step in a user journey, rather than the entire user journey. For example, we would recommend embedding the Digital Person in a Widget on a pricing page, rather than on the home page of a website.
The Solution Expert isn’t a coach (it’s not training the user or interested in their long-term growth), a companion (they don’t remember you), or an influencer (they aren’t creating content for entertainment purposes). The Solution Expert isn’t a Top-of-Funnel concierge (they don’t have knowledge of the store website or layout), or a Greeter (a Solution Expert works best once a buyer has reached a consideration phase).
Best practices
We recommend keeping the scope of the Solution Expert experience limited to deliver the best result.
We recommend ensuring the knowledge snippets have clear content that explicitly references desired content and uses clear and concise language. (e.g, rather than saying “it is $99” the knowledge snippet should say “The Nebula Star A51 is $99”)
What are the current limitations?
The Generative Conversation Skill doesn’t currently support
@showcards()
or@pronounce()
features.Out of the box, the Solution Expert can recommend a final decision for the user, but cannot act on it. For example, in the template, they recommend the right phone for you but do not possess any capability to help you purchase/add to your cart.
Users need a credit card on file with OpenAI to use their API.
Support Content
FAQ
Within Interactive Preview, can the Digital Person be in ‘template’ mode or ‘edit mode’?
Not at present. Therefore, once you start editing the configuration of the Digital Person, the template may stop functioning as intended. We recommend having a ‘vanilla’ / untouched template for reference if this is required.Do I need an OpenAI API Key?
Not presently, but you may choose to add your own to manage your billing.Does this work with other GPT models?
We have built the Solution Expert template to work with OpenAI’s version of GPT-4. Changing the model will likely yield different results.Does this work with other LLMs?
We have not tested this template with other LLMs.