The kind of work we do

The situation

Prospect research gets done by hand: someone works through directories and map listings, copies details into a spreadsheet, and tries to judge which businesses are worth approaching. It is slow, it is the first thing dropped in a busy week, and the list is usually stale by the time anyone picks up the phone.

What we built

A scheduled n8n workflow gathers businesses matching an ideal-customer profile, enriches each one with public web data, scores fit with an LLM, drafts a first-touch message, and writes the result into the CRM with a source link for every claim.

Stack

• n8n
• LLM enrichment
• CRM sync
• Scheduled runs

The situation

Calls arrive while staff are already with a customer, and after closing there is nobody to answer at all. Every unanswered call is a booking that goes to whoever picks up next — and because it is never logged anywhere, the loss stays invisible.

What we built

A telephony voice agent handles greeting, service selection and availability lookup against the live calendar, writes the booking, sends an SMS confirmation, and warm-transfers anything unusual to a member of staff. Every call is summarised into the inbox.

Stack

• Voice agent
• Calendar API
• SMS confirmations
• Human hand-off

The situation

Most incoming messages are the same handful of questions — is it in stock, when does it arrive, how do returns work, are you open today. Answering falls to whoever is free, replies slow to a crawl at peak times, and a genuine problem waits in the same queue as a routine question.

What we built

A WhatsApp Business assistant grounded in the client’s product catalogue and policy documents, with order-status lookups, French/Arabic/English handling, an escalation path to a human agent, and a weekly digest of what customers actually asked.

Stack

• WhatsApp Business API
• Retrieval over docs
• Order lookup
• Escalation rules