Skip to content
RIWA
Case study — AI System

Lythe AI staffing agent

The MVP they pitched with: a WhatsApp agent that talks to candidates and matches them to jobs.

Lythe — system architecture
Client
Lythe
Sector
Staffing — Singapore
Timeline
8 weeks
Built with
n8n · OpenAI API · Supabase · RAG · Redis

The problem

Lythe is a Singapore startup automating the staffing agency model with AI-driven job matching. To get an MVP in front of investors and early clients they needed the first piece working end to end: a WhatsApp agent a candidate could hold a real conversation with, which then connected them to suitable roles.

What we built

The agent, and the intelligence behind it. This was not a matter of wiring services together — we designed the conversation flow, wrote and iterated the system prompt, tuned the model for the recruitment domain, and put retrieval over their own data so answers came from their roles rather than from the model's imagination. Redis holds conversation state, so a candidate picks up where they left off.

The result

There is no before-and-after number here and we will not manufacture one — the product was new. What it did was give Lythe something real to pitch with instead of a deck. It was also the first automation this studio shipped, and the reason the AI side of the business exists.

New to us?

Let us prove it first

We started in 2025, and we know that is a fair thing to weigh up. So before money changes hands, we will build you a working demo of the thing you are actually asking for. If it convinces you, we carry on. If it doesn't, you have lost a week of our time and none of yours.

REPLIES WITHIN 1 HOUR · SUN–THU, 9:00–23:00 KUWAIT CITY