From messy requirements into working CRM, automation and AI workflows.

Case 01 · Retail · Voice AI

The flower company: from random calls to a revenue-generating voice robot

One of my clients, a flower company, had no call scripts at all. Their managers were handling incoming calls completely randomly, without any structure.

I developed a professional sales and service script for them, integrated it into a voice robot, and set up full automation between the robot and their CRM — including logging call results and automatic follow-ups.

The resulting system standardized the first line of communication, automated CRM follow-ups and surfaced previously unstructured customer demand. One of the insights from those conversations led the client to introduce balloons as an additional product category.

• Voice robot
• Sales & service scripts
• CRM automation
• Call-outcome logging
• Follow-ups

A chaotic first-line process became a structured, automated workflow with consistent conversations, CRM-recorded outcomes and systematic follow-ups.

Case 02 · Healthcare education · AI chat agent

Prenatal support platform: from a 3-day callback to a one-minute qualification

A company in prenatal education and support for expectant parents came to us because their managers couldn't keep up with incoming leads. Everything was getting stuck, and they assumed the problem was simply low conversion.

I started by reviewing their traffic channels and how they handled them — then went through the process as a real customer. The finding: it took 3 business days from the moment a lead submitted a request until a manager actually called. In their industry, that's far too slow.

We proposed an AI agent for messengers that immediately qualifies the lead and answers questions in real time — literally within a minute.

• Messenger AI agent
• Lead qualification
• Traffic review
• Customer-journey walkthrough

Managers stopped spending first-line capacity on spam, junk and non-target inquiries and could focus on qualified, higher-intent leads. The overloaded manual response process became an automated qualification layer with a first response measured in minutes rather than business days.

Case 03 · Healthcare · CRM rebuild

Dental clinic: rebuilding a CRM that four contractors left broken

A dental clinic came to us after trying 3–4 CRM contractors over a couple of years. Those vendors made daily operations slightly easier but never thought about marketing, conversions, CRM hygiene or analytics. As a result the CRM had become a messy notebook: to pull anything useful out of it you first had to clean the data — and even then you couldn't really trust it.

I started with an AS-IS review of the clinic's lead, consultation, treatment and analytics processes, then designed the TO-BE CRM model around simpler funnels, fewer required fields and automation-first state transitions.

The project was handed to me with full ownership. The target model deliberately reduced operational complexity: simpler funnels, the minimum required fields and controlled state transitions designed to reduce human error.

The whole system became highly automated. Operators only fill in 3–4 required fields and choose from short lists of 5–6 values. Managers can't move deals through the pipeline manually, can't create duplicates, and can't leave a deal sitting unqualified for an extra day — because if a patient with a toothache isn't contacted within minutes, they're gone.

Marketing channels were set up correctly, analytics were connected, and everything is reconciled with the clinic's medical software: when a visit happens and generates revenue, it shows up both in the sales CRM and in the final analytics.

• CRM rebuild
• Error-proof automation
• Marketing channels
• Analytics reconciliation

The result was a CRM where operational data could be trusted for day-to-day work, marketing attribution and downstream analytics instead of requiring manual cleanup before every analysis.