How can messaging and automation work together in customer service? In the Leafworks and Ultimate webinar “Multichannel Communication and Chatbot Automation”, the speakers showed how asynchronous conversations and connections to backend systems can support each other. This article from 2022 records examples and statements from that event. The original webinar address now redirects to a Zendesk AI product page; the recording is no longer directly available there.
For your plans today, refer to current Zendesk AI agents. Product names, capabilities and integration options have evolved since the webinar. The current Zendesk AI agents documentation and our Zendesk AI overview provide context.
How messaging changes service
In the webinar, Robin Adler from Ultimate described how more conversations between customers and brands were taking place through digital channels. Alongside email, telephone and website chat, social messaging played a growing role.
The Zendesk CX Trends Report 2022 (PDF, p. 5) shows an increase in support requests through WhatsApp of 370 percent in the comparison between 2020 and 2021. The figure describes that historical period.
A multichannel strategy starts with the channels your customers want to use and how your team brings those contacts together. A shared view of requests, contact history and progress makes collaboration easier.
Asynchronous conversations
Asynchronous messaging lets customers pause a conversation and return to it later. A request can start in the morning and continue after work, without the customer having to stay on hold throughout the process.
To make this work, your team needs an accessible conversation history and clear responsibilities. Customers also need to know when they can expect a reply.
Connecting channels and teams
More contact channels create opportunities, but they also require coordination. Customers may send the same request through several channels. Without shared information, this can lead to duplicated work and conflicting answers.
- Customer context: Previous contacts and request progress need to be available to the team responsible.
- Communication: Chat, email and messaging differ in their flow and expected response times.
- Collaboration: Handovers between teams and channels need clear rules.
- Effort: Repetition and manual follow-up make handling requests take longer.
- Availability: Decide which requests automation can handle and when people should take over.
No matter where you look, more and more conversations are happening between customers and brands.
Robin Adler, Ultimate
Where AI can help
Automation can handle recurring requests and administrative work. As an example at the time, the article described how Ultimate supported the airline Finnair with a large volume of cancellation requests during the first wave of the coronavirus pandemic.
AI can produce answers from suitable knowledge sources and support multilingual service. Answer quality depends on the content, use case and configuration, among other factors. Test results with realistic examples and provide a handover to your team when automation cannot reliably resolve a request.
Routing requests and retrieving data can also be automated. This requires suitable integrations, permissions and data quality. The right connection depends on your systems and processes. Our Zendesk Quickstart sessions offer a starting point.
Benefits for your team
- Less repetition: Shared information helps customers avoid explaining the same issue several times.
- Less routine work: Suitable automations can handle recurring processing steps.
- Faster handling: Automated data retrieval can reduce manual searching and follow-up questions.
- More time for advice: Your team can focus on requests that need personal attention.
You can automate everything, the question is what you want to automate.
Malte Lensch, Leafworks
Example: order status in chat
In the webinar, James Longbotham from Ultimate and Malte Lensch from Leafworks demonstrated an order-status enquiry. The example showed how a virtual agent can retrieve information through a backend connection and provide it during a conversation.
An ecommerce company wants to support its customer service outside regular service hours. Through a Zendesk CRM and backend integration, a virtual agent could retrieve an order status from a connected system. This requires secure identification of the customer and their order, as well as suitable access to the system.
When personal handling is needed, automation can first collect details such as the customer number and the issue. It can then pass these on with the conversation context, so the service team has the information needed when a person takes over.
The original article’s figure of around 60 percent of requests being partially or fully automated is not a reliable forecast for every customer service team. Assess your potential using your actual requests, the scope of automation and measured results.
The implementation times of under three weeks, or five days at expondo, mentioned at the time are also historical project examples. Your own timeline depends on knowledge sources, integrations, approvals and testing. The original reference to Ultimate’s Automation Explorer belongs to that historical product context.
A useful first step is to choose a frequent request and clarify the data, workflow and handover to people it needs. Our page on Zendesk automation explains how to approach these projects.


