Contact me
Case studies Real project, live

Sales CRM with AI, connected to production

From prospecting to signed contract: an eleven-stage pipeline, follow-ups that schedule themselves and AI that prepares every call. A won contract creates the project in ABRACCIO OS.

For
SMEs and operations, Business teams, Agencies and firms
Service
Build + Automate + AI
Client
Abraccio Studio, Luanda
Connected to
ABRACCIO OS, the management tool that works offline
See it live crm-abracciostudio.online

Team-only access: the sign-in page is public.

Abraccio CRM sign-in: “Da primeira mensagem ao fecho.” and the brand mark in 3D
CRM sign-in on mobile, dark theme
CRM sign-in on mobile, light theme
Light theme and dark theme.
Hover over or tap a block to see its role in the system.

An internal tool in daily use: only the sign-in page is public. The screens behind it are shown in a meeting.

01Problem

The sales team at Abraccio Studio, the agency I run in Luanda, prospected from a spreadsheet: one tab for leads, one for scripts, one for the dashboard. Nobody knew who to follow up, when, or why a contract had been lost.

Cold prospecting needed a measurable method: find companies without a working website, contact them at the right moment, and let nothing slip between two follow-ups.

02Context

Users
Salespeople, sales managers, management, administrative staff, human resources and finance: six profiles.
Current process
A shared spreadsheet, WhatsApp messages and follow-ups done from memory.
Constraints
Contact mostly over WhatsApp, call times suited to Luanda, amounts in kwanzas, the Angolan personal data law (Lei 22/11), and AI that is useful but never sends anything in place of a person.
Tools in place
  • Spreadsheet
  • WhatsApp
  • Email
  • Google Maps

03Solution

A custom CRM that replaces every spreadsheet tab with a page. The built-in guide shows where each old sheet has gone.

The method lives in the software: a pool of companies spotted on Google Maps and checked before they are added, an eleven-stage pipeline with its own rules, follow-ups on day 3 and day 7 that schedule themselves, and an automatic prospecting pause after three working days without a qualified lead.

AI steps in at four moments where the salesperson decides, always from the server and always as a draft: qualifying an imported list, suggesting an angle of approach, preparing a call, writing the meeting summary.

Why this approach. An off-the-shelf CRM would have forced the team to adapt to the tool. Here, the tool applies the agency's sales method, with its stages, its timings and its local constraints.

04Architecture

Input

  • Google Maps pool Companies with no website, or a site that is down or not secure, checked by a salesperson before they enter the CRM.
  • Excel / CSV import Paste from Excel or import a 500-row CSV, duplicates set aside, assignment rules.

Interface

  • React CRM 23 screens: Today, leads, pipeline, calendar, targets, team, playbook… ⌘K palette, light and dark themes.

Automation

  • Node.js API Cookie-based sessions, permissions checked on every request, a log of every change.

AI

  • Server-side AI A, B or C qualification, angle of approach, call preparation, meeting summary: drafts that a person approves, with usage visible.

Data

  • SQLite database Leads, activities, meetings, targets and permissions; full backup and logged CSV exports.

Output

  • WhatsApp and email Messages built from the team's scripts with a note specific to each company; the day 3 follow-up gets scheduled.
  • ABRACCIO OS A won contract creates the project in the production tool: lead, team, schedule, tasks, deadlines and kick-off meeting.

05Features

  • TodayOverdue follow-ups, meetings in the next three days, dormant leads, and the next leads to contact, ranked by priority.
  • Eleven-stage pipelineFrom spotted lead to contract won or lost, drag and drop with the keyboard as well as the mouse, a reason required for every loss.
  • One-click approachA WhatsApp message, email or call prepared from the playbook, with a note specific to the company, and an alert outside the recommended hours.
  • Automatic follow-upsEvery contact logged schedules the follow-up for day 3, then day 7.
  • The AI prepares, a person decidesList qualification, angle of approach, call preparation and meeting summary, as drafts.
  • Performance trackingRevenue against target, paced by working days, weighted pipeline, conversion by stage, reasons for loss.
  • Team discipline built inDaily target, automatic prospecting pause, a Friday meeting that creates itself, monthly ranking.
  • “Do not contact” listFilled automatically, in line with Angolan data protection law.
  • SecuritySix profiles, 32 permissions, lockout after five failed attempts, revocable sessions, a log of every change.

06Result

  • The sales spreadsheet is replaced: every old sheet has its own page.
  • No more forgotten follow-ups: the cadence schedules itself.
  • A won contract moves into production with no re-keying.

How it is measured

  • Qualified leads per working day and per salesperson
  • Stage-to-stage conversion rate
  • Revenue against target

07What I did

  1. Modelling the sales process
  2. UX/UI and design
  3. React front-end
  4. Node.js API and SQLite database
  5. Server-side AI integration
  6. Permissions, security and logging
  7. Connection to ABRACCIO OS
  8. Production launch
Next case study ABRACCIO OS, the management tool that works offline

A system like this one, for your business?

Contact me All case studies