“The main impact has been organization. Zelty AI helped us completely refine our billing, support and sales processes. The exact moment I realized it was working was the time it freed up in the first days of the month; the assistant is a huge help with cutoff dates and payment reports. On top of that, their developers' support for any request has been flawless.”
Key performance metrics (June 2026)
| Metric | Result | Operational impact |
|---|---|---|
| Assistant effectiveness | 77.7% | A steady increase from the 65% recorded in May 2026. |
| Total conversations processed | 3,756 | Total volume of interactions handled by the ecosystem. |
| Conversations resolved by AI | 1,157 | 30.7% of the total operational load absorbed 100% with no human intervention. |
| Overall AI handling | 1,683 (44.7%) | Interactions filtered, pre-classified or resolved by the automated agents. |
| Transfer to humans | 923 (24.6%) | Efficient filtering that only routed complex or high-priority sales cases. |
Before implementing Zelty AI, Julián (CEO of Sonytel) faced a critical and common problem for internet service providers: disorganized billing, support and sales processes, combined with an unmanageable flow of messages that overwhelmed the technical team. To solve it, Sonytel tried two earlier solutions on the market:
First alternative
A tool that completely failed to understand the business context and to effectively solve customers' requests, increasing friction instead of reducing it.
Second alternative
A solution that showed acceptable early potential, but whose provider or infrastructure suddenly disappeared from the market, leaving the support operation stranded and forcing the team straight back to manual work.
The biggest operational pain came in the first days of each month. Service cutoff dates, mass payment reports and recurring billing questions created a critical bottleneck. It took up 100% of the agents' time, keeping them from high-value work such as closing sales or solving complex technical incidents in the field.
Organizing processes with a multi-agent system
Sonytel implemented Zelty AI's Multi-Agent System with a dual strategic focus: structuring the administrative chaos and recurring interactions, and automating repetitive high-volume tasks. The rollout wasn't just connecting a generic bot; it involved continuous refinement and agile technical integration, enabling:
Process modeling
Defining and implementing automated service routes that precisely tell apart billing, technical support and sales inquiries.
Automating demand peaks
Setting up precise knowledge bases so the AI could autonomously handle receipt submissions, payment reports and cutoff date checks.
Partner-level technical iteration
A direct channel with Zelty's development team, ensuring immediate responses and integrations to adapt the AI's behavior to Sonytel's real and changing needs.
Time freed up and operational predictability
The real, measurable business impact took hold in June 2026. The AI assistant's effectiveness grew organically to a solid 77.7%. Of the more than 3,700 monthly interactions, the AI directly handled 1,683 and fully resolved 1,157 of them, relieving the human team of almost a third of the total transactional load.
The “aha moment”
For the CEO it came right at the start of the month: the team saw a dramatic amount of time freed up on service cutoff days. Instead of manually processing hundreds of “I already paid” or “what's my balance?” messages, the platform absorbed the transactional volume in seconds, bringing calm and order back to the company.