Our Company
A small team that believes AI in customer service works best when it stays in the background.
Back to HomeOur Story
How Eveilsynthet came to be
Eveilsynthet grew out of a simple observation: most customer service teams who tried AI tools in their workflows found that the tools arrived faster than the trust did. Agents felt uncertain about what the system was doing; managers got dashboards they couldn't interpret; and customers, occasionally, received replies that didn't sound like anyone in particular.
We started working on this problem in Kuala Lumpur in 2022, running small studies with support teams across a few industries — reading their ticket samples, sitting in on QA sessions, and asking frontline staff what was actually useful and what wasn't. What we found was consistent: the teams that got the most from AI drafting were the ones where a person had spent time listening first.
That's the core of what we do. Before we suggest any technology, we listen. We read your conversations, understand your tone, and map the places where a draft suggestion would actually reduce friction — rather than just add noise.
Our Mission
Helping support teams stay human while working alongside AI
We don't believe AI should handle customer relationships on its own. We believe it can help agents handle them better — if it's introduced carefully, configured honestly, and reviewed with the same rigour applied to any other quality standard.
Our work is bounded on purpose. We audit, we pilot, we review. We write plain-language reports rather than live dashboards. We hold quarterly conversations with frontline staff. We move at the pace your team sets, not the pace a product roadmap demands.
Our Approach in Brief
- Listen before recommending
- Keep agents central to every decision
- Write findings in plain language
- Agree data handling before any work begins
Our Team
The people at Eveilsynthet
Nadia Rashid
Founder & AI Integration Lead
Nadia spent eight years in customer operations before moving to AI research. She set up Eveilsynthet to bridge those two experiences — bringing a practitioner's patience to every audit and pilot engagement.
Azri Kamarudin
Helpdesk Systems Specialist
Azri handles the technical side of every Draft Assist deployment — configuring tone models, setting escalation rules, and ensuring the assistant layer sits cleanly inside existing helpdesk tools.
Shireen Lim
Quality Review & Reporting
Shireen writes the weekly conversation samples and QA reflections for the Quiet Quality Companion retainer. She has a background in editorial work and brings that discipline to every report.
How We Work
Our operating standards
Written Data Agreement First
No ticket, chat log, or email thread is reviewed until a data handling agreement is signed. Scope, anonymisation method, and retention terms are all defined in writing beforehand.
Agent Consultation Built In
Frontline staff are consulted during the pilot and retainer phases — not as a courtesy, but as a required input. Their observations shape what we tune and what we escalate.
Documented Methodology
Each service follows a written process. Clients receive a clear brief at the end of an audit, structured weekly notes during a pilot, and formatted QA reflections under the retainer.
Confidentiality Standards
All customer data accessed during engagements is handled under strict confidentiality. We do not retain, share, or use client data for any purpose outside the agreed scope of work.
Iterative, Not Waterfall
We adjust based on what we hear. Weekly clinic sessions during the pilot let us tune tone models and escalation rules as agents learn how the assist behaves in practice.
Plain-Language Outputs
Reports are written for people who manage teams, not for people who read data science papers. Every finding is accompanied by a clear explanation of what it means for daily operations.
Context & Expertise
AI integration for customer service in Malaysia
Customer service operations across Malaysia's financial services, e-commerce, and telecommunications sectors have been evaluating AI drafting tools since 2021. Many teams found that the technology was capable but the deployment lacked structure — tools were switched on without prior study of ticket patterns, tone calibration took months by trial and error, and there was no systematic way to understand whether the AI assist was helping or creating additional editing work for agents.
Eveilsynthet addresses this by starting where the work actually happens: the inbox. Our audit methodology reads a defined sample of real support conversations — with consent and under a data agreement — and produces a written brief that maps recurring topics, common friction phrases, and the sentences agents already handle well. This brief becomes the reference point for any AI configuration that follows.
For teams in Kuala Lumpur and across Malaysia working on Zendesk, Freshdesk, or HubSpot Service, our Draft Assist Pilot introduces a suggestion layer that respects the existing escalation model. Sensitive topics — complaints, account security, legal queries — are flagged for human handling directly, without the AI offering a draft. The six-week supervised period gives agents and managers time to calibrate their own judgement about when the assist is useful and when it isn't.
Quality oversight is the part of AI integration that most implementations underinvest in. The Quiet Quality Companion retainer exists specifically for teams that have passed the pilot stage and want a structured, ongoing review of how the assist is performing — written for QA committees and operations directors, not for system administrators.
Ready to start with a conversation?
We're happy to talk through your team's situation before recommending anything. There's no obligation in an initial call.
Get in Touch