DialOnce

Our expertise in artificial intelligence

Because a trustworthy AI is above all an explainable one, DialOnce designs high-performance and transparent solutions, serving an artificial intelligence that is useful, measurable, and controlled.

AI serving customer relations

Blend AI architecture combining NLU and LLM to understand and resolve customer requests with precision and context

At DialOnce, artificial intelligence (AI) makes it possible to understand and guide every customer request with precision.

 

Our conversational engines are built on a hybrid architecture that combines Natural Language Understanding (NLU) with the most advanced Large Language Models (LLMs) available on the market.

 

This proprietary approach, which we call Blend, merges the rigor of NLU, used to detect intent and emotion, with the flexibility of LLMs, which generate natural, accurate, and contextualized responses.

An adaptable and trustworthy AI

Moving towards multi-agent AI

The main area of development at DialOnce lies in the transition to a multi-agent architecture, built on two pillars: A2A (Agent-to-Agent) and MCP (Model Context Protocol).

 

A multi-agent system enables multiple autonomous, specialized agents to collaborate and exchange information in order to accomplish complex tasks together.

 

Thanks to the A2A protocol, agents can collaborate and share information with one another, coordinating their actions to deliver a seamless user experience.

 

The MCP protocol, on the other hand, enables them to interact securely with third-party systems (CRM, business tools), ensuring that each agent only accesses the data strictly necessary.

 

This evolution allows each stage of a conversation to be handled by a dedicated agent, whether it is identifying intent, processing a request, or activating a business tool. For example, during the creation of a customer service ticket, the conversation can be automatically routed to the specialized agent in charge, ensuring both precision and continuity.

 

This architecture will enhance the relevance and reliability of interactions, while laying the foundation for an ecosystem of interoperable, secure, and trustworthy AI.

AI agent system enabling seamless and secure customer request handling through A2A and MCP protocols
AI with transparent review, conversation sampling, and automated suggestions to enhance service quality

A trustworthy AI

DialOnce’s AI is built on three essential principles: transparency, human oversight, and compliance.

 

Our solution incorporates the concept of “LLM as a Judge”: an AI agent configured as a reviewer, responsible for evaluating the quality of interactions a posteriori based on key indicators such as resolution rate, user satisfaction, and answer accuracy.

 

This agent regularly reviews a representative sample of conversations (around 10%) to identify those that did not meet the expected outcomes or generated user dissatisfaction.

 

But its role goes beyond analysis: it actively contributes to the continuous improvement of the solution. Based on observed trends, it generates automated recommendations to optimize conversational flows, response phrasing, and routing logic.

 

The insights collected then feed into a joint improvement loop, shared between our teams and our clients’ teams during feedback sessions and steering committees.

 

This collaborative process ensures an AI that is evolving and contextualized, able to adapt to real-world usage, environments, and business objectives.

FAQ - Directus

FAQ

All DialOnce products (omnichannel AI agent, mailbot, augmented advisor) are built on explainable, transparent and supervised AI. Our hybrid NLU and generative AI architecture accurately understands intents and generates natural, reliable responses, powered by RAG technology to eliminate hallucinations by leveraging your document corpus. Quality is continuously monitored through automated controls and key indicators: perceived sentiment, resolution rate, and response compliance. The AI always clarifies its role and limitations and can transfer requests to a human advisor for complex or sensitive cases. The goal is to deliver high-performance, controllable and predictable AI that serves your teams.
At DialOnce, RAG (Retrieval-Augmented Generation) is integrated directly into the core functioning of the AI agent. Before formulating a response, the AI queries your knowledge bases and internal documents to retrieve accurate, up-to-date information. This ensures that every response reflects your business repository without requiring model retraining when things evolve. RAG thus strengthens the accuracy, compliance and relevance of responses.
LLM-as-a-Judge is integrated into the DialOnce platform to evaluate, control and improve AI-generated responses. This second model assesses each response according to three indicators: perceived sentiment, resolution, and response compliance. These "trustworthy AI" indicators are visible directly in the DialOnce console and replace heavy manual audits with continuous, standardized supervision. They enable journey adjustments, improvement of AI agent responses, and consistent quality over time.
Fine intent understanding is the first step in how DialOnce products operate. The AI agent precisely identifies through semantic analysis what the customer is trying to accomplish, even when the request is vague or emotional. It detects, disambiguates and collects the necessary information. This understanding enables triggering the right action from the first contact, thus increasing the First Contact Resolution Rate, producing an appropriate response or routing to the right channel.
Semantic disambiguation is natively integrated into DialOnce agents. It effectively processes the imprecise and frequent requests common in customer relations. When intent is unclear, the AI asks the necessary questions to complete the information, rephrases if the request is too vague, and relies on your business content to clarify the situation. This avoids processing errors, unnecessary back-and-forth or unneeded transfers to an advisor. The customer saves time, and advisors receive already-clarified requests.
DialOnce tools (AI agent and chatbot, visual IVR, mailbot and augmented advisor) integrate easily with your ecosystem, whether your CRM, business tools, internal APIs or contact platforms. You can also use the LLM of your choice, sovereign or not, while maintaining security mechanisms, disambiguation, RAG and trustworthy AI indicators. This flexibility ensures operational continuity regardless of your technical environment.
The multi-agent approach is at the heart of the DialOnce platform. Multiple specialized agents collaborate in real time to analyze intent, verify information, retrieve data from your IT systems, generate a response or create a ticket... Thanks to MCP (Model Context Protocol), these agents can exchange structured information in a standardized way, securely call business tools and access only the data relevant to their task. The orchestrator then coordinates these agents (via MCP) to deliver seamless handling, even when requests are complex. For the customer, the exchange remains continuous and uninterrupted. For your teams, this translates into more precise actions, faster processing and better journey control.
See our AI agents in action
Book a demo