Ecommerce ops · Production

Trendyol scales AI and agentic workflows across 200 teams using n8n: seller chatbot, legal AI, and search relevancy agent

The problem

Trendyol's 2,000-person technology org had far more demand for new tooling and automation than engineering time could absorb — even small tasks rarely made the product roadmap. Non-technical teams had backlogs with no independent way to act on them.

Workflow diagram · grounded in source
1
Seller asks chatbot question
trigger
“Sellers can now ask it things like "what's my revenue over the last seven days?"”
2
Orchestrator classifies intent
ai_action
“the chatbot routes the request through an orchestrator that classifies the intent and calls one of around ten n8n workflows behind it”
3
N8n workflow executes action
routing
“classifies the intent and calls one of around ten n8n workflows behind it. SQL queries, and price analysis all run as actions, not just answers”
4
Employee tags legal bot in Slack
trigger
“employees can tag the bot with a question before reaching out to the legal team directly”
5
RAG retrieves and compares legal answers
ai_action
“n8n retrieves answers from the knowledge base and compares responses across multiple LLMs”
6
Legal team validates responses
human_review
“The international legal team validates answers and uses it to speed up their own responses”
7
Agent scans low-CTR search terms
trigger
“Running periodically on low-CTR (Click-Through Rate) search terms across four storefronts—Romania, Greece, the UAE, and KSA”
8
Relevancy score generated
ai_action
“the agent evaluates the first 40 product results for each search query to generate a relevancy score”
9
Ticket auto-triggered below threshold
routing
“If this score falls below 80, the system automatically triggers a ticket for the respective team”
Reported outcome

In under a year n8n grew from one developer's local POC to over 1,000 active users, 700 active workflows in production, and roughly half a million executions per quarter across 200 team-scoped projects.

Reported metrics
Active usersmore than 1,000
Active workflows in production700
Executions per quarterroughly half a million
Isolated team projects200
Show all 7 reported metrics
active usersmore than 1,000
active workflows in production700
executions per quarterroughly half a million
isolated team projects200
seller chatbot current rollout10%
developer time savedthe real benefit here is the time saved for our developers
time to Enterprise adoptionwithin a month
Reported stack
n8nRAGLLMsSlack
Source
https://n8n.io/case-studies/trendyol/
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

In under a year n8n grew from one developer's local POC to over 1,000 active users, 700 active workflows in production, and roughly half a million executions per quarter across 200 team-scoped projects.

What tools did this team use?

n8n, RAG, LLMs, Slack.

What results were reported?

Active users: more than 1,000; Active workflows in production: 700; Executions per quarter: roughly half a million; Isolated team projects: 200 (source-reported, not independently verified).

How is this ecommerce ops AI workflow structured?

Seller asks chatbot question → Orchestrator classifies intent → N8n workflow executes action → Employee tags legal bot in Slack → RAG retrieves and compares legal answers → Legal team validates responses → Agent scans low-CTR search terms → Relevancy score generated → Ticket auto-triggered below threshold.

Related ecommerce ops cases