← Back to perspectives
/Intellicon FDE Team

Expanding Services with AI Starts with What Customers Need Before and After a Purchase

Beyond buying a wedding dress, couples still need to find vendors and organize wedding preparations. Map out customers’ difficulties from purchase to use, then pilot services for which your company has the information and a team ready to take responsibility.

AIEgentWrX人機協作導入
Expanding Services with AI Starts with What Customers Need Before and After a Purchase

The equipment has been sold, but the customer still does not know what to prepare before installation, so the salesperson has to bring in the after-sales team for help. If a company wants to use AI to turn this kind of support into a service, should it start with customer service Q&A or build a service platform right away? Intellicon Solutions recommends first looking at what customers still need to get done before and after purchasing a product, and how the company’s existing information and experience can help. Only then should the company decide which steps are suitable for an AI Agent.

TechOrange reports that David’s Bridal has expanded beyond selling wedding dresses to helping couples plan their weddings, including adjusting timelines, shortlisting vendors, and scheduling appointments. This also gives wedding vendors opportunities to reach customers who need their services. The company has incorporated customer data and the experience of its in-store consultants into its system, extending its services beyond the dress purchase. However, the reported results for conversions to store visits come from company executives, with no specific figures or comparison samples provided. Businesses can draw ideas from the service design, but should not assume that following the same approach will generate new revenue.

What Do Customers Still Need Help with After Buying a Product?

Take pre-installation preparation in the equipment industry as an example. Start by defining the service clearly: help customers confirm site details and list what they need to prepare, then have service staff identify any conditions that require further clarification. This is a scenario businesses could pilot, not a validated customer case. Whatever is ultimately delivered to the customer should make it clear whom to contact and what information to prepare, rather than simply summarize the product manual.

Sales and after-sales teams can work together to map the customer journey from product selection through installation and use. At each stage, document where customers encounter difficulties, who currently handles them, what information the company has available, and what the customer should receive when the service is complete. If meeting a particular need requires finding a partner first, leave it for later evaluation rather than promising it in the initial pilot.

When choosing a pilot, prioritize work that is directly related to existing products, has accessible information, and falls under a clearly designated department. Then check with customers: would this preparation checklist actually help, or do they need someone to inspect the site? If a customer needs an on-site assessment by a qualified professional, the company should arrange that first. Automatically generating a document does not mean the problem has been solved.

Document How Experienced Staff Ask Follow-Up Questions for the AI Agent

Product specifications can provide dimensions and operating requirements, but they do not necessarily tell service staff what to ask customers first. When interviewing experienced colleagues, ask them to walk through common cases and exceptions: what the customer initially asked, which follow-up questions they asked, and why they ultimately adopted or rejected a recommendation. This captures more than just the final outcome.

For example, if a customer simply says they want to move equipment to another factory, the team can ask an experienced colleague to demonstrate how to confirm the equipment model and site conditions, and which documents to consult. When documenting the process, distinguish between requirements explicitly stated in the documents and advice based on an individual’s experience. If any conditions remain unconfirmed, specify who needs to provide the missing information rather than drawing conclusions from incomplete data.

With EgentWrX, for example, companies can place documents, tables, and standards in a departmental knowledge base for an AI Agent to consult. They can then document the sequence of follow-up questions, applicable conditions, and ways to handle exceptions as skills: reusable methods and rules. Documents give staff sources they can verify, while skills explain how to work through a task step by step. Both require ongoing maintenance by staff.

Once this material has been organized, ask another colleague to review it using de-identified cases, meaning cases with personally identifying information removed. Check for missing questions about essential conditions, incorrectly applied rules, or unsupported recommendations, and make revisions before trying it out. A custom-built skill must pass the platform’s skill review process before it can be used company-wide. An individual’s experience cannot simply be treated as a company standard.

Pilot One Part of the Service: Who Reviews It Before It Reaches the Customer?

For the first pilot, we recommend having service staff use the AI Agent rather than opening it up to customers directly. Continuing with the pre-installation example, the process could start by confirming the conditions provided by the customer, consulting the applicable documents, and drafting guidance. After-sales staff would then review the content and deliver it to the customer. Define clearly what must be done for each service engagement to count as complete.

This process can be organized through tasks and workflows in EgentWrX. Tasks are reusable work cards, while workflows connect tasks in sequence. The team can set up information retrieval and drafting as separate tasks, with human approval required before moving to the next step. This pauses the process until a staff member has reviewed and approved the work.

The review should go beyond whether the writing reads well. Staff also need to check whether the cited documents apply, whether the customer has provided all necessary information, and whether the content exceeds the service scope the company has committed to. If information is missing, service staff should return the work to the relevant colleagues for further questions. Issues involving site safety or special conditions should be referred to the person responsible for that work who also has the relevant expertise. Before the pilot begins, managers therefore need to designate who reviews the content, who delivers it to the customer, and who handles exceptions.

Once Customers Are Using the Service, How Do You Decide Whether to Charge?

We recommend evaluating the service’s usefulness and customers’ willingness to pay separately. A customer’s willingness to accept a free guide does not mean they will pay for ongoing support. At the same time, even if the company does not charge for now, it can assess whether this assistance makes delivery of its existing products smoother. There is no need to rush into making it a standalone business.

During the pilot, the team should record whether customers complete their preparations using the guidance and what further help they need. It should also track how much time staff spend finding information, revising content, and following up on progress. AI usage fees and the cost of handling exceptions need to be included as well, so that faster drafting does not obscure the staff time and effort required for subsequent review.

To evaluate charging for the service, first define its scope and delivery standards. Distinguish between work that is already part of after-sales responsibilities and additional assistance, then ask customers whether they would be willing to buy it. At the next project meeting, start by deciding which service to pilot, what customers will receive, and who will be responsible. Agree on which records will inform the decision to continue, adjust, or stop. Discuss expanding the service only after these points have been settled.

FAQ

Does a company need to build a platform from the outset to expand its services with AI?

No. Start with a clearly scoped service related to existing products. Sales and after-sales teams can work together to identify what customers still need help with, confirm where the information will come from, and define what customers will receive. Service staff can then use drafts in a pilot to see whether customers actually put the guidance to use.

How should experienced staff’s service knowledge be organized for an AI Agent?

Start by documenting the sequence of follow-up questions and the reasoning behind decisions, rather than simply saving past answers in documents. Ask experienced colleagues to use common cases and exceptions to explain which conditions lead to different recommendations, then have other colleagues check whether the rules apply. Where information is missing, specify what else needs to be asked and who is responsible for following up.

If customers are willing to use a free service, does that mean the company can start charging?

No. Willingness to use a service for free and willingness to pay for it are different things. First confirm whether customers actually use the material provided, then clearly define the paid service’s scope and ask whether they would purchase it. Companies also need to track the time and cost of human review, ongoing maintenance, and exception handling to assess whether the service is worth pursuing.

References

30 minutes to map out which work to hand to AI first

Want every employee to have their own AI teammate?

A consultant will be in touch shortly.

ISO/IEC 27001:2022 Certified ISMS