Customer support · Production

How Electra's customer care team handles complex support tickets 80% faster with Dust

The problem

Electra's customer care team faced a scaling challenge as ticket volume grew with European expansion, with complex escalated tickets requiring agents to pull information from multiple disconnected systems. The company also needed an AI solution accessible to non-technical staff while satisfying cybersecurity's data governance requirements, which scattered generic ChatGPT licenses could not meet.

First attempt

Electra had previously distributed generic ChatGPT licenses across teams, but cybersecurity had no visibility into what company data was being shared with AI tools and could not enforce compliance policies or measure data access.

Workflow diagram · grounded in source

Escalated ticket arrives

trigger

“When a human agent receives an escalated ticket from Intercom, they can call one of the three Dust agents depending on the ticket type”

Agent selected by ticket type

routing

“The team identified three categories of escalated tickets that consumed the most time from Electra's customer care team: invoice-related questions, refund requests, and general complex inquiries about charging sessions or technical issues”

Conversation thread scanned

ai_action

“The agent automatically scans the conversation thread to understand the customer's issue”

Multi-source data retrieval

ai_action

“it pulls relevant information from three key sources:Slack conversations where teams discuss station issues, customer problems, and operational updatesNotion knowledge base containing documentation, procedures, and troubleshooting guides…”

Pre-drafted response delivered

output

“Within three minutes, the Dust agent delivers a pre-drafted response that includes all the necessary context, relevant links, and specific data points”

Human agent review and send

human_review

“The human agent reviews the response, makes any needed adjustments, and sends it to the customer”

Reported outcome

The three Dust customer care agents reduced time per escalated ticket by 80%, with complex tickets now handled in about 3 minutes.
The agents average 4 new AI conversations per hour. Company-wide, Dust reached 70% weekly active users within the first month of its September 2025 launch, stabilizing at 70–80%.

Reported metrics

Time reduction on escalated tickets80%

specialized customer care AI agents deployed3

AI conversations created per hour4

Weekly active users within first month of launch70%

Show all 8 reported metrics

time reduction on escalated tickets80%

specialized customer care AI agents deployed3

AI conversations created per hour4

weekly active users within first month of launch70%

weekly active users stabilized range70-80% depending on seasonality

Dust agents created company-wide170

time per escalated ticket now3 minutes

initial AI champion licenses distributed50

Reported stack

DustIntercomMCP

Source

https://dust.tt/customers/electra-customer-care-team-faster-support-tickets

Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The three Dust customer care agents reduced time per escalated ticket by 80%, with complex tickets now handled in about 3 minutes.

What tools did this team use?

Dust, Intercom, MCP.

What results were reported?

Time reduction on escalated tickets: 80%; specialized customer care AI agents deployed: 3; AI conversations created per hour: 4; Weekly active users within first month of launch: 70% (source-reported, not independently verified).

What failed first in this deployment?

How is this customer support AI workflow structured?

Escalated ticket arrives → Agent selected by ticket type → Conversation thread scanned → Multi-source data retrieval → Pre-drafted response delivered → Human agent review and send.