All Cases
Flagship Case Active Project Construction · International

Enterprise AI System for KOLEO Group

KOLEO Group — international construction company specializing in wooden and stone construction, exporting to USA, Canada, and EU. Website in 13 languages. CE-certified.

200+
Documents
48
API Functions
3+
Access Levels
4 wks
To Production

The Challenge

Knowledge scattered, AI potential untapped

The CEO needed a system that would structure all company knowledge — technologies, pricing, processes, response templates — into a single, searchable source of truth.

The system had to integrate with Google Workspace (Drive, Sheets, Calendar, Gmail), work on multiple devices with different access levels, and enable AI assistants to answer questions based on real company data.

Previous attempts with off-the-shelf AI tools fell short: ChatGPT didn't know the business, and manually searching through scattered documents took 15-30 minutes per query.

What We Built

Five interconnected systems

1

Corporate Knowledge Base

200+ documents structured across 11 categories. 3+ scalable access levels: EXECUTIVE, STAFF, PUBLIC. Single Source of Truth on Google Drive with MASTER_INDEX navigation.

2

CEO AI Assistant

Claude Desktop with custom system prompt, connected to KB via Project Knowledge. Knows prices, technologies, processes, and response templates. Answers in 3 languages (UA, RU, EN).

3

MCP Bridge — 48 API Functions

Custom Google Apps Script bridge connecting AI to Google Workspace: Drive (read/write/search), Sheets (data access), Calendar (event management), Gmail (read-only for security).

4

Multi-Device Access Control

3+ scalable access levels across multiple devices. Staff version with restricted access (no finances, no margins). Control through separate Project Instructions per role.

5

Dashboard & Automation

Looker Studio dashboard with key business metrics. Apps Script automation for data updates. Auto-backups of critical files.

Results

Before & After

Metric Before After Impact
Information search time 15-30 min 10-30 sec 30×
Client response preparation 1-2 hours 5-10 min 10×
New employee onboarding 2-3 weeks 2-3 days
Structured documents ~20 scattered 200+ unified 10×
Tool integrations 0 48 API functions

Timeline

4 weeks from kickoff to production

01
Week 1

Discovery & Audit

Analyzed existing knowledge sources, mapped data flows, identified gaps and opportunities.

02
Weeks 1-2

KB Architecture & Build

Structured 200+ documents across 11 categories with 3+ access levels. Built MASTER_INDEX navigation.

03
Week 2

MCP Bridge Development

Built custom Google Apps Script API bridge with 48 functions: Drive, Sheets, Calendar, Gmail integration.

04
Week 2

AI Assistant Configuration

Configured Claude Desktop with custom system prompts, connected to KB, deployed on 3+ devices.

05
Weeks 3-4

CEO Training & Refinement

Hands-on training, feedback loops, refinement of AI behavior and knowledge base structure.

06
Week 4

Dashboard & Handoff

Looker Studio dashboard, documentation, troubleshooting guides, final handoff.

What Happened Next

The empowerment effect

After delivery, the client independently created company-wide SOPs in just 2 days. They then integrated n8n workflow automation on their own — without any developers.

This is what we build for: systems that don't just solve today's problem, but empower teams to keep extending and improving independently.

Technology Stack

Claude (Anthropic)
AI Platform
MCP (Model Context Protocol)
Integration
Google Apps Script
API Bridge
Google Workspace
Data Layer
Looker Studio
Analytics
Node.js
Runtime

Want Similar Results?

Book a discovery call and we'll show you what's possible for your business.

Book Discovery Call