EDF
Intelligent routing callbot + self-service chatbot
EDF's callbot and chatbot on Illuin Technology's Dialogue platform cut call abandonment by 16%, improve direct routing by 30% and satisfaction by 7 points, with 100,000 monthly exchanges and more than 25 million cumulative interactions in France.
Key points
- Intelligent routing callbot and self-service chatbot for the customer front desk.
- Illuin Technology's Dialogue conversational platform, in SaaS.
- Call abandonment -16%, direct routing +30%, satisfaction +7 points, 100,000 exchanges per month.
- Evidence level B, confirmed live status.
Objective
Smooth the phone and digital front desk: reduce abandoned calls, send each customer to the right person the first time, and free advisors from simple requests.
The deployment
EDF deployed a callbot and a chatbot on Illuin Technology's Dialogue conversational platform, in SaaS. The callbot interprets the spoken request and routes the call to the right path: an EDF advisor, an external service like Enedis, self-service (bill payment), or a digital resource. The advisor is informed in advance of the nature of the request. The chatbot handles simple requests and manages about 100,000 exchanges per month. The platform, which combines generative AI and more frugal approaches, has been proven by EDF over more than 25 million interactions. EDF's 5,000 advisors, all in France, focus on high-value cases.
Results Proof B
A quantified case study from the integrator (Illuin Technology) on the phone front desk, confirmed by an AFRC article naming EDF and the customer experience director, and by an AFRC session mentioning more than 25M interactions. Concordant sources.
How it works
Documented architectureThe stack in detail
- plateforme Dialogue (Illuin Technology) SaaS conversational platform that carries the routing callbot and the self-service chatbot; proven at EDF over more than 25 million interactions.
- llm Briques genAI et NLP frugal de Dialogue The platform combines generative AI and more frugal methods; the exact models are not publicly named.
- integrateur Illuin Technology French scale-up publishing the platform, supporting EDF's customer experience division.
- infra Telephonie et centre de contact EDF 5,000 advisors based in France; the callbot routes the call and informs the advisor of the request before handling.
How it runs, concretely
For ops teams-
1Listen and interpret AI
The callbot interprets the customer's request on the phone and identifies the target path.
-
2Routing AI
It routes to an EDF advisor, an external service (Enedis), self-service, or a digital resource, informing the advisor of the nature of the request.
-
3Self-service by chat AI
The chatbot handles simple requests and hands off to an advisor when needed.
-
4Human handling customer service
The advisor takes over the high-value cases, already contextualized.
Correctly detecting the spoken intent for routing. If the intent is misclassified, the call goes to the wrong path and the gain on abandonment and routing disappears.
How your customers perceive this type of use
Sourced studiesLes consommateurs n'acceptent pas les chatbots par defaut : 64% prefereraient que les entreprises n'utilisent pas d'IA dans leur service client (Gartner, 2024) et pres d'un utilisateur sur cinq du service client par IA n'en retire aucun benefice (Qualtrics, 2025). L'acceptation se construit sur trois conditions mesurees par Salesforce : savoir qu'on parle a une IA, pouvoir escalader vers un humain, comprendre la logique de l'agent.
Acceptance conditions
- Etre informe qu'on parle a une IA et non a un humain (pres de 75% le demandent, Salesforce 2024)
- Un chemin d'escalade clair vers un agent humain (45% plus enclins a utiliser l'agent IA, Salesforce 2024)
- Une logique de l'agent clairement expliquee (44% plus enclins, Salesforce 2024)
Red lines
- Rendre l'humain injoignable : c'est la premiere inquietude des consommateurs sur l'IA dans le service client (Gartner 2024) et 50% craignent que l'IA les coupe du contact humain (Qualtrics 2025)
- Remplacer le service client par l'IA sans alternative : 53% envisageraient de partir chez un concurrent (Gartner 2024)
Sources: Salesforce 2024 · Gartner 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- a map of intents and journeys
- a directory of services and advisor skills
- a knowledge base for self-service
Org prerequisites
- a contact center structured by skill
- agreement on the routing rules
- an advisor informed of the request upfront
Possible stack
- a voice + chat conversational platform (Dialogue or equivalent)
- an intent-based routing engine
- genAI components anchored on a knowledge base
The plan, step by step
- Step 1Map the most frequent call intents and the target paths: advisor, self-service, external service.Deliverable: An intent reference and validated routing rules
- Step 2Configure the routing callbot on the platform and wire it to telephony.Deliverable: A working callbot in pre-production
- Step 3Launch a pilot on a major call flow, passing the context to the advisor.Deliverable: First figures on call abandonment and direct routing
- Step 4Extend to the self-service chatbot on simple requests (bill payment).Deliverable: Chatbot in production on common requests
- Step 5Measure abandonment, direct routing, and satisfaction; correct the misclassified intents.Deliverable: A quantified readout and an intent improvement backlog
First step: Model the most frequent call intents and wire a routing callbot before adding self-service by chat.
The same pattern, proven in other industries
ADT
genAI customer agent
Air India
genAI customer agent (automated customer service)
AT&T
Governed enterprise genAI assistant: an LLM connected to internal knowledge, open to tens of thousands of employees and called at scale by automated systems
Sources
- S1 EDF transforme son accueil client avec un callbot et un chatbot epaules par l'IA Interested party archive pending
- S2 L'humain et l'IA au coeur de l'experience client (EDF) - AFRC Secondary archive pending
- S3 Dialogue: une plateforme testee et approuvee par EDF sur plus de 25M d'interactions - AFRC Meet-up 2025 Secondary archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.