151
Countries on the automated email path
Multilingual inbound, one retrieval-and-draft stack.
Case Study
RAG: Automation e-mail response with AI and LLMs
Senetic's Outlook inbox took product questions in every language the company sells in. An employee then searched thousands of SKUs by country and wrote the reply by hand. Vstorm put RAG on that path so the draft comes back in real time.
The outcome
The inbox still lands in Outlook. The reply does not wait in a queue.
Inbound mail is recognised as a product question, matched against live catalogues, and answered in the customer's language — with complementary items when they belong. The published coverage is 151 countries. That is Senetic's footprint on this pipeline, not a Mixam conversion rate.
2 million+ customers a year and 27 subsidiaries describe Senetic's scale. They are not Vstorm throughput. We do not publish minutes saved per mail; the published result is a fully automated RAG response path across 151 countries.
Thousands
Products in the live catalogue retrieval
Country, SKU, and complementary items — not a frozen prompt.
Outlook
Inbox the team already had
The mail still lands where it landed. The lookup and the draft do not.
About the client
Senetic is a global IT solutions provider. Since 2009 it has sold networking, servers, software, and hardware to small and mid-size businesses and to public institutions. It is a Microsoft partner. The operating picture is 27 subsidiaries, sales in 151 countries, and more than two million customers a year.
AI was new to the firm when they came to Vstorm. The first job was to name a process that was actually burning people: inbound product email, not a slide about transformation.
Vstorm's impact
Vstorm's impact, the TL;DR
- Entire inbound email response path automated with RAG
- Real-time replies across 151 countries and multiple languages
- Outlook inbox wired to live product catalogues — thousands of SKUs
- LangChain orchestration plus LlamaIndex retrieval over unstructured sources
- Drafts follow Senetic's structure, including complementary items
- Sales time moved from routine replies to account management
The challenge
One inbox, many languages, thousands of products
A customer email landed in Microsoft Outlook and waited. Someone then searched a catalogue of thousands of products, matched country, SKU, and language, thought about add-ons, and wrote a reply that still had to meet Senetic's communication rules. That loop does not scale across 151 countries.
The brief was not a generic chatbot. It was semantic search and question answering over Senetic's own catalogues, with enough training-data discipline that answers stay accurate and unbiased.
How we delivered
RAG on the mail they already receive
The inbox stayed Outlook. The retrieval layer is what changed.
Name the queue that was eating the day
Workshops mapped inbound mail: language mix, country, product lookup, add-on logic, and the written standard for a Senetic reply. AI was new here, so the first deliverable was a shared picture of the loop, not a model card.
- Inbox path in Outlook
- Language and country mix
- Catalogue lookup pain
Proof of Value on live catalogues
RAG over databases of unstructured product data. LangChain for the generation path. LlamaIndex for retrieval. Semantic search to recognise a product question before anyone drafts.
- RAG over catalogues
- LangChain generation
- LlamaIndex retrieval
Wire Outlook and let the draft go out in real time
The existing inbox talks to LLMs and to catalogues that update live. The model picks options, adds complementary items when they belong, and writes to Senetic's structure.
- Outlook integration
- Live catalogue hook
- Structured multilingual draft
How it works
Recognise, retrieve, then write the mail Senetic would write
Generative models draft the letter. RAG keeps that draft on real SKUs by pulling from connected databases at request time — not from a frozen prompt. LangChain holds the chain. LlamaIndex speeds retrieval over messy product text.
Inbound customer email Lands in the existing Outlook inbox, now wired to LLMs
↓
Product inquiry recognised Country, product, and language identified
↓
Live catalogue retrieval RAG and LlamaIndex over thousands of SKUs
↓
Best-fit options selected Matches plus complementary items
↓
Structured email generated LLM drafts to Senetic guidelines
Senetic × Vstorm — RAG email response pipeline
Results
The reply is automated. Account work is not.
The email response process at Senetic runs on RAG end to end. Clients get answers in real time instead of waiting on a manual search through the catalogue. The team that used to write those mails can spend the hours on account management. We do not publish a dollar figure for that shift. The published facts are full automation of the path, 151-country coverage, and personalisation from live retrieval — not from a template farm.
Work with us
Ready to see how agentic AI transforms inbound sales email?
Meet directly with our founders and PhD AI engineers. We will walk through real implementations from 30+ agentic projects and the practical steps to integrate them into your workflows.