Give AI the explanation you would want a customer to read.
Your business already has useful knowledge: the steps you repeat, the policies you explain and the details a new teammate needs to learn. Bringing that information into TruCom gives customers, your team and AI a clearer place to begin.
The work starts with deciding what is true, useful and appropriate to share. Then you can test how that information supports an answer.
Build around the questions you already hear.
Start with a small set of questions your team answers often. How does someone get started? Where do they find a receipt? What should they try when an invitation does not arrive? A focused collection of useful answers is easier to check and maintain.
TruCom can work with AI-enabled help articles, supported website sources, uploaded documents and pasted text. Choose the source that makes the information easiest for your team to keep current.
A public help article is useful when customers should be able to find and read the answer themselves. A source added for AI gives you another way to provide reference material. Check its audience, AI setting and readiness rather than assuming that adding content makes it available for every use.
Include the detail that gets someone unstuck.
Write the question in words your customers use, then answer it directly. Give the steps in order. Name the page, action or document someone needs, and explain what they should expect after taking that step.
Include the conditions that change the answer. If a process depends on the country, product, account type or stage of an order, say so. If a teammate needs to review a request, explain how the customer reaches them and which details will help.
“Find the tracking link in your dispatch confirmation email” gives someone a place to look.
If the email is missing, explain what to check and how to ask your team for help.
Document what your business actually does. A warm tone can make clear instructions easier to read.
Use one consistent explanation for a policy or process. If two sources disagree, resolve the difference in the content so a customer or teammate has a clear answer to work from.
Audience, AI access and readiness each have a job.
The audience describes who a source is intended to help: customers, agents or both. AI access controls whether that source is enabled for AI use. Readiness tells you whether it has been processed and is available for retrieval.
For customer answers, TruCom uses ready, AI-enabled sources intended for customers or both audiences. Linked help articles also need public visibility. Sources intended only for agents are kept out of the knowledge searched for those customer answers.
Instructions you can stand behind publicly.
Product guidance, customer policies and clear steps someone can follow when asking for help.
Context that helps the team do the work.
Internal procedures and review notes that are useful to an eligible teammate through team tools.
A ready source still needs an accurate explanation. Readiness confirms processing; your team checks whether the information is correct. When you update a source, check its sync or indexing state before testing the new content.
Try the questions before relying on the answers.
The AI Playground lets you test customer-style questions and inspect the result alongside the knowledge retrieved. You can see an answer or a handoff decision, then read the source excerpts that informed it.
Ask the same question in a few natural ways. Include an incomplete question and a related question your content does not fully answer. Then try the customer experience with a direct request for a person, so you check the route back to your team as well.
Does the answer address the question? Does the source support the details it gives? Are the conditions and next steps clear? If something is missing, improve the source and test again.
You can choose whether visitors see source citations in supported AI replies. During setup, read the source material yourself as well. A good test gives you evidence about that question and setup; keep checking as the content and customer questions change.
AI access, available models and chargeable usage depend on your plan, credits and workspace settings. The written examples on this page make no AI requests. Tests you run in your own workspace follow its AI usage configuration.
Leave room for questions the content cannot settle.
Help content can explain a process. A specific customer’s situation may still need investigation, judgement or access to another business system. In the delivery example, an article can explain where tracking lives; the team needs the actual delivery details before confirming an arrival time.
When the available evidence is not strong enough, TruCom can request a human handoff. Customers can also ask for a person. AI pauses for the handoff, and an eligible teammate can continue in the same conversation with the messages already in view.
Choose the operating mode and connections that fit your support routine. Make sure the intended department and responders can receive handoffs. A handoff requests help; your team’s availability determines when someone can pick it up.
Give Copilot context. Keep the teammate’s judgement.
The AI support agent can answer eligible customer messages where you enable it. Agent Copilot works alongside your teammates: helping summarise the conversation, find relevant knowledge or prepare a reply suggestion for review.
Those tools use context for different purposes. A summary helps the next teammate understand what has happened. Related knowledge can surface material available to agents, including internal sources. Customer-facing reply suggestions draw on customer-appropriate knowledge.
The teammate reads the suggestion, checks the source and edits the response to fit the situation before sending it. If the evidence is too weak for a useful reply suggestion, Copilot can return without a draft.
Check names, dates, conditions and any claim about an action being completed. Only tell a customer that something has been changed or checked when the team has actually done that work.
Keep the knowledge as current as the business.
A policy changes. A screen moves. Customers find a step you have not explained. Use those moments to improve the content your team and AI work from.
- Choose who looks after each topic.
Have someone check the instructions and policies they know well, especially after a product or process changes.
- Update the source and check it is ready.
Allow the updated content to sync or be indexed. Check the audience and AI setting still match its purpose.
- Test the question again.
Read the answer and source excerpts together. Check a question that needs a person as well as the straightforward one.
- Learn from the handovers.
When the same missing explanation keeps bringing customers back to the team, turn it into a clearer article or source.
Start with one question your team knows well. Write the answer clearly, test it and make sure a person can take over when needed.