A customer-service AI agent is more than the chat window a customer sees. Behind every deployed agent, a team needs to configure knowledge, define its behaviour, align it with a brand, choose where it appears, and understand how it responds to users.
Clair AI was designed as a conversational AI-agent platform for customer-service use cases. Desisle created the product design over approximately 14 days, with one designer responsible for the engagement and its design iterations.
Project Overview
- Client: Clair AI
- Industry: Conversational AI and customer service software
- Product: Conversational AI agent platform
- Platform: Web-based product, subject to confirmation
- Primary Users: Teams configuring and deploying AI customer-service agents
- Services: AI product UX and UI design
- Team: One designer
- Duration: Approximately 14 days
- Scope: Product design and design iterations
What Did Desisle Do?
Desisle designed the product experience for Clair AI, a conversational AI platform that helps teams configure, train, customise, and deploy customer-service agents. The work focused on the workflows needed to upload business information, define prompts, apply branding, enable agents across web pages or products, and interact with deployed agents to retrieve information.
The Challenge: Building More Than a Chat Interface
The visible chat experience is only one part of a customer-service AI product. The more complex design problem sits behind it: users need to prepare the agent with relevant information, establish how it should behave, customise its appearance, decide where it appears, and monitor or test its responses.
Clair AI needed an interface that could make this process understandable. Product teams should be able to move from agent setup to deployment without navigating a fragmented collection of configuration screens or depending entirely on technical specialists.
The AI Agent Setup and Deployment Workflow
The product was designed around the lifecycle of creating and using a conversational AI agent. This workflow connects the operational work of preparing an agent with the user-facing experience of asking questions and receiving answers.
- Configure or create an AI agent.
- Add business information and relevant source data.
- Train the agent through prompts and configuration.
- Apply customer-specific branding and customisation.
- Enable the agent across relevant web pages or digital products.
- Test or interact with the agent.
- Review AI responses and retrieve relevant outputs.
Recommended Visual: A workflow diagram showing Configure → Add Knowledge → Train → Brand → Deploy → Test → Review.
Why It Matters: It establishes that Clair AI is an agent-management platform, not only a chatbot interface.
Designing for Brand, Knowledge, and Control
Customer-service agents often need to appear within different customer environments. The platform therefore needed to support customisation and branding as part of the setup experience.
It also needed to support the addition of business data. The supplied product description indicates that teams can upload numerical information, metrics, and other business-related data so the agent can be trained around the information relevant to that customer or product.
The design challenge was to make these configuration activities feel structured. Users need to understand what information has been provided, which settings affect agent behaviour, and what they need to complete before deployment.
Key UX Decisions
1. Separate Setup from Customer Conversation
Problem: The tasks involved in configuring an AI agent are different from the tasks involved in using one.
Decision: Design the product around both an operational setup experience and an interaction experience for asking questions and viewing outputs.
Why: Administrators need control over agent data, prompts, branding, and deployment, while end users need a simple conversation interface.
Trade-off: The platform must maintain a clear connection between setup choices and the behaviour users see after the agent is deployed.
2. Give Users a Structured Deployment Path
Problem: Training, branding, and deployment can become confusing when users do not know which actions are required or what stage they are in.
Decision: Organise the experience as a clear progression from knowledge setup through customisation, training, deployment, and testing.
Why: A visible sequence helps users understand how to prepare an AI agent before making it available to customers.
Trade-off: The product needs to stay flexible enough for advanced users while remaining understandable for teams that are new to AI-agent configuration.
3. Keep AI Outputs Tied to Available Data
Problem: A conversational agent is only useful when its responses relate to the knowledge and business data provided to it.
Decision: Design the product to support uploading and using business information, metrics, and related data in agent training and interactions.
Why: The agent needs context to respond to customer and business questions in a meaningful way.
Trade-off: The interface must make it clear what information the agent can use and avoid implying that outputs are independently verified facts.
The Designed Experience
The product design established an AI-agent workflow for customer-service teams, covering the key activities involved in configuring and using conversational AI agents.
- AI Agent Creation and Configuration: Establish the initial setup and configuration of customer-service agents.
- Business Data and Knowledge Upload: Provide a workflow for adding relevant business information and source data.
- Prompt-Based Agent Training: Support the configuration of prompts that help define agent behaviour.
- Brand Customisation: Allow agents to be tailored to customer environments.
- Deployment: Support enabling agents across relevant customer websites and digital products.
- Embedded Conversational Experiences: Account for the chat experience users encounter on customer-facing platforms.
- Agent Testing and Interaction: Support asking questions and reviewing the agent's responses.
- Output and Data Review: Provide an experience for reviewing responses and retrieving relevant outputs.
The goal was to make a technically complex product feel manageable for teams that need to deploy branded AI customer-service support across multiple customer environments.
Recommended Visuals:
- Agent dashboard showing configured agents and their status.
- Data or knowledge-upload screen.
- Prompt-training interface.
- Branding and customisation screen.
- Deployment or embed configuration view.
- Deployed chat interface showing a customer-facing conversation.
Outcome
Desisle delivered the product design for Clair AI in approximately 14 days, with one designer responsible for the work and design iterations. The project established a user experience for configuring, training, customising, deploying, and interacting with customer-service AI agents.
No verified post-launch performance data was supplied. This case study does not claim support-ticket reduction, automated-query rates, response-time improvements, customer-satisfaction gains, cost savings, adoption, retention, or business impact.
What We Learned
Conversational AI products need to make the relationship between setup and output understandable. Teams need to know what data they have added, which prompts or settings influence agent behaviour, where the agent is deployed, and how users experience it after launch.
The product also reinforced that brand customisation is not only visual. In a customer-service context, teams need confidence that an AI agent can be configured around their business context before it is presented to customers.
Building a Similar AI Agent Product?
Teams building customer-service AI agents need to design more than a polished chat window. They need to make knowledge management, configuration, training, deployment, branding, testing, and user interaction feel like one coherent product experience.
Desisle can support AI and SaaS teams designing complex setup workflows, embedded agent experiences, contextual AI interactions, and development-ready interfaces.
Frequently Asked Questions
What Is Clair AI?
Clair AI is a conversational AI-agent product designed for customer-service use cases. It enables teams to train, customise, brand, deploy, and manage AI agents across customer websites and digital products.
What Did Desisle Design?
Desisle designed the Clair AI product experience, including flows for agent configuration, knowledge upload, prompt training, brand customisation, deployment, and conversational interaction.
How Can Users Customise an AI Agent?
The supplied product context supports branding and customisation of agents for customer environments. Specific branding controls and implementation details should be confirmed before publication.
What Data Can Users Provide to the Agent?
The supplied context indicates that teams can upload numerical data, business data, metrics, and other relevant information for agent training. Supported file types, data sources, and governance requirements need confirmation before publication.
Was Clair AI Deployed as a Live Production Product?
The supplied information confirms product design and iterative design delivery. It does not confirm a live production release, technical integrations, or deployment outcomes.
Are Customer-Support Performance Metrics Available?
No verified performance metrics were supplied. The case study should not claim automated resolution rates, lower ticket volume, faster response time, customer satisfaction improvements, cost savings, or revenue impact.