F.A.Q.

Find answers to your questions here

Solution

What does your solution actually do?

Our solution automates the end-to-end processing of incoming messages. It identifies the type of request, automatically extracts key data, pre-fills fields in your business tools (CRM, databases, etc.), suggests a response (if desired), and routes the message to the appropriate department.

It can also trigger automations such as appointment scheduling or contract termination handling, without any manual intervention. Result: your teams focus on higher value-added tasks, and your ROI is maximized through time savings and reduced errors.

What message volumes can you handle?

Capacity is fully scalable based on your needs and has no strict volume limit (typically from 50,000 to 10 million messages per year)

What is the time to go live?

Our Solutions Consultants configure the AI according to your needs. The average timeline is 4 to 6 weeks, but it may vary depending on the project scope (number of business use cases, integrations, languages) and the desired level of automation.

Is there support during go-live?

Yes, our Solutions Consultants support you throughout the AI configuration phase and continue to support you throughout your entire customer lifecycle.

What are your use cases?

Our use cases focus on automating the processing of incoming messages in sectors such as banking, insurance, tourism, and retail.

For example, our AI automatically processes incoming claims declarations in insurance, common customer requests in banking, bookings and cancellations in tourism, and customer returns and complaints in retail.

It classifies, extracts, and routes information to the right tools or departments to speed up processing and free teams from manual tasks.

What is the analysis time for an incoming message?

The analysis time for an incoming message typically ranges from 1 to 10 seconds.

Which languages are supported by your AI?

Our solution is multilingual and already supports 24 languages:

German, English, Chinese, Korean, Danish, Spanish, French, Hungarian, Italian, Japanese, Norwegian, Dutch, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Swedish, Czech, Turkish…

The effort required to implement a new language is minimal.

Can you prioritize emails?

Yes, our NLU can detect multiple intents within a single message and prioritize emails based on their level of urgency.

Can your AI adapt to our specific business use cases?

Yes, our Solutions Consultants team configures the solution to address your specific business use cases and requirements.

Can your AI be trained or adjusted using our own data?

Our AI works without training; it is configured directly using your data and your business context.

Does your solution integrate with our existing tools?

Yes, our AI is available via API and can connect to your internal tools.

Do you offer a testing phase or a PoC?

Depending on your needs, we can offer customized demos.

How does post-deployment support work?

We provide three levels of support to ensure continuity and ongoing improvement after deployment:

Level 1 : Operational maintenance, auto-scaling (10-minute latency), and deployment of minor versions with no service interruption.

Level 2 : Intervention in case performance drops below the contractual threshold, quarterly performance monitoring, and implementation of corrective actions if needed.

Level 3 : Continuous improvement through feedback, including analysis of user feedback and regular adjustments to enhance and optimize automation.

This approach ensures day-to-day stability while enabling gradual, measurable improvements over time.

Which business KPIs does your solution improve?

Our solution directly improves key KPIs through automation: reduced Average Handling Time (AHT), lower churn, increased customer satisfaction and retention, and fewer human errors.

Can your solution be deployed on-premise?

Yes, our solution can be deployed on-premise under certain conditions.

ROI

What performance guarantees do you offer?

We commit to achieving a relevance rate of at least 85%.

What are the concrete benefits for a company?

Average handling time divided by 2

First-contact resolution time divided by 3

Cost of incoming requests divided by 10

Is there a minimum number of messages to process?

Our solution delivers strong ROI starting at 100,000 messages processed per year.

Technology

What is neuro-symbolic AI?

Neuro-symbolic AI is a hybrid approach that combines the learning power of neural networks with the reliability and explainability of symbolic systems.

Why neuro-symbolic AI?

Unlike LLMs such as ChatGPT, which are designed to generate free-form text and may produce hallucinations, our neuro-symbolic AI is built for the reliable automation of message processing. It combines neural networks with symbolic logic, making it far more explainable and predictable: every decision can be justified and traced.

This transparency ensures a high level of reliability, which is essential for industries handling sensitive data (banking, insurance, etc.). While an LLM behaves like a “black box,” our AI delivers verifiable, auditable results that comply with regulatory requirements. This makes it particularly well suited to environments where trust and compliance are critical.

How is your AI different from ChatGPT or a traditional LLM?

Our AI is not a conversational assistant like ChatGPT. Instead, it understands incoming messages in a structured way—without hallucinating—then classifies them, extracts useful information, and suggests reliable responses.

Can your AI analyze attachments even if the email body is unclear?

Yes. Our AI analyzes attachments and extracts the keywords and data required. It can process 28 different file formats: ( .pdf / .doc / .docx /.ppt / .pptx / .pptm / .xls / .xlsx / .xlsm / .odp / .ods / .odt. / .csv / .eml / .epub / .htm / .html / .json / .log / .msg / .rtf / .txt / .png / .gif / .jpeg / .jpg / .tif / .tiff )

GDPR

Is Miralia GDPR compliant?

Yes. Our solution is hosted in full compliance with GDPR requirements. Our technology is designed to process only the information that is strictly necessary, without any mass collection of personal data. All data is hosted in France on secure, sovereign servers. We guarantee our customers security, transparency, traceability, and full control over their data.

Is Miralia compliant with the EU AI Act?

Yes. Miralia follows an approach that is fully aligned with the principles of the AI Act:

Transparency and explainability : Our AI is designed to be explainable at every stage, enabling human oversight and full decision traceability.

Limitation of personal data usage : Our technology can operate without systematically relying on personal data, reducing compliance risks.

Security : We apply recognized standards (such as OWASP ASVS) to protect against vulnerabilities, in line with the AI Act’s robustness requirements.

Responsible design (“by design”) We aim for a “compliant by design” AI, with an architecture that inherently limits high-risk use cases and facilitates adaptation to regulatory changes.

Does Miralia host data in Europe / on sovereign servers?

Yes. Our AI is hosted by Scaleway in France, a sovereign European cloud provider.

Pricing

How does pricing work?

Our solution is offered as an annual subscription that includes a defined volume of analyses, onboarding and support services, the SaaS infrastructure, and the processing of incoming message flows.

Company

What is Miralia?

Miralia is a French company specializing in explainable and sovereign artificial intelligence. Its neuro-symbolic technology automates the processing of incoming messages (emails, attachments, forms, etc.) to improve productivity while ensuring reliability, transparency, and regulatory compliance.

When was Miralia founded?

Miralia was founded in 2016 and is the result of more than five years of R&D focused on neuro-symbolic AI models.

Who are Miralia’s customers?

We work with a wide range of customers, from SMEs to mid-sized companies and large enterprises, across many industries (Insurance, E-commerce & Retail, Tourism, Defense & Security). We are particularly well suited to sectors handling so-called “sensitive” data or complex use cases.