Tell me what's taking up too much time in your business, where your team gets stuck, or what you'd like to improve.
I'll ask a few questions, help explore practical automation options, and prepare a summary for Nchia Mbong to review. You don't need to know anything about AI to get started.
What would you most like to make easier in your business?
Network error connecting to assistant.
file.pdf
Solutions for Everyday Business Problems
Running a business involves many tasks that take time away from serving customers and growing the business. We help identify repetitive work, improve everyday processes, and use AI and automation to make work easier. Explore the challenges we can help you address and the practical approaches that may suit your business.
What We Do & How We Work
Practical automations designed to solve measurable operational friction, not apply technology for its own sake.
🎯
Who We Help
Small and growing businesses handling repetitive manual work.
Service businesses that need better customer enquiry handling.
Teams that need their information and business tools connected.
Customer enquiries are difficult to organize and prioritize.
Important business information is scattered across separate tools.
Teams lack clear visibility and alerts into routine daily work.
🚀
How We Work
Discover — Understand the core business problem.
Design — Plan the workflow and expected outcome.
Build — Implement the agreed solution cleanly.
Test — Verify that it behaves reliably under edge cases.
Deliver — Explain results, operational limits, and handoff.
Evidence of Work
Explore verified operational prototypes and live interactive case studies.
Demonstrated Capabilities
Production-ready business automations, intelligent client workflows, and verified technical integrations.
Estimate the Cost of Repetitive Work
Evaluate how much manual administration and routine message handling might be costing your operations every month.
3
5 hrs
2,000 FCFA
Weekly team hours:15 hrs
Monthly team hours:65.0 hrs
Estimated monthly labour cost:129,900 FCFA
Illustrative recoverable hours (30%):19.5 hrs
Illustrative potential monthly value:38,970 FCFA
* Note: The 30% recoverable rate is an illustrative planning assumption, not a guaranteed outcome. Real results depend on workflow structure and adoption.
Case Study
STATUS
The Business Problem
How The Solution Works
Verified Real-World Results
Request Consultation & Follow-up
Leave your contact details and business context. Nchia Mbong will review your operational requirements and reach out directly.
⚙️
Solution Title
Project 01 · Demo #001 · Portfolio System Demonstration
LEADFLOW
Automated Lead Triage & Representative Action Pipeline
Built and engineered by Nchia Mbong — AI Automation & Integration Specialist |
Turn incoming enquiries into prioritized, trackable sales action.
LeadFlow connects a website consultation form to a structured Google Sheets CRM, applies deterministic HOT / WARM / LOW triage rules, dispatches priority email alerts, records representative actions, and gives the business owner a consolidated operational view.
≤ 5 sec
Target submit-to-alert dispatch time
3 Tiers
HOT · WARM · LOW rule-based triage
1 Ledger
CRM + representative action log
Live View
KPIs + urgent lead watchlist
● The Business Bottleneck
Speed-to-lead Gaps
Valuable inquiries often sit in shared inboxes without clear ownership or timely action, leading to lost sales opportunities.
Context Loss & Manual Work
Manual data copying and incomplete qualification context slow down routing and increase human data-entry error.
Limited Visibility
Owners and managers lack a consolidated view of acknowledgment status, outreach, follow-up, and deal outcomes.
● How the System Works
01 · Web Intake:A responsive consultation form captures contact details, budget, urgency, service needs, and consent.
02 · Automated Triage + CRM:Google Apps Script processes submission, assigns a Lead ID, applies deterministic HOT/WARM/LOW rules, and records the inquiry into Google Sheets.
03 · Representative Dispatch:A priority email alert provides a lead summary and instant direct access to the representative action workflow.
04 · Rep Action Tracking:A mobile-friendly form records contact status, follow-up needs, deal outcome, and notes.
05 · Owner Visibility & Audit:The Owner_Dashboard summarizes lead volume, priority distribution, acknowledgment, contact outcomes, urgent HOT leads, and recent activity.
● Technical Architecture
Website Intake (HTML/CSS/JS) → HTTPS Request to Apps Script Web App → LeadFlow Processing & Deterministic Triage → Google Sheets CRM → Priority Email Alert → Rep Action Form → Rep_Logs → Owner_Dashboard
⚡ Interactive Qualification Sandbox (Live Demo)
Test the Triage EngineDeterministic Pipeline
Select a preset enquiry or type your own custom message below to watch the triage engine evaluate urgency, volume, and budget in real time:
* Safe simulation: Evaluates input client-side without storing data.
ASSIGNED PRIORITY:HOT PRIORITY (100/100)
Signals Detected:
Immediate implementation timeline detected
High operational friction & message volume identified
Commercial clarity & contact context provided
Automated Dispatch Action:Instant WhatsApp/SMS notification to operations manager; draft tailored proposal within 2 hours.
● System in Action: Verified Implementation Evidence
Figure 1: Client Intake Interface
Website consultation form capturing contact information, budget, and urgency timeline with instant client confirmation.
Figure 2: Priority Email Alerts
Priority-oriented alerts dispatched to assigned sales reps with full lead context and one-click action portal link.
Figure 3: Owner_Dashboard & Audit Activity
Consolidated executive view summarizing total leads, priority distribution, rep acknowledgment, and urgent HOT watchlist.
● Business Value & Performance Target
Organized Intake
Replaces scattered messaging with structured, auditable CRM records.
Consistent Qualification
Deterministic budget and urgency rules eliminate guesswork in prioritization.
Rep Accountability
Tracks acknowledgment timestamps, outreach attempts, and deal outcomes.
Executive Visibility
Owner dashboard flags unattended HOT inquiries before they turn cold.
Performance Benchmark: Designed around a target of under 5 seconds from form submission to rep alert dispatch. For teams previously relying on manual copying and delayed inbox checking, this compresses routine intake from hours into minutes.
Need a more organized way to capture, qualify, and follow up with leads?
Discuss your current intake process, where inquiries get delayed, and how a tailored LeadFlow-style workflow can streamline your sales pipeline.
Project 02 · Portfolio Demonstration
SANAGA ROYAL VOYAGES
AI Customer Support Assistant
Built by Nchia Mbong — AI Automation & Integration Specialist |
A customer-support assistant designed to answer from approved information—and recognize when a person needs to step in.
Sanaga Royal Voyages is a fictional Cameroonian intercity transport operator used to demonstrate how an AI assistant can handle common travel questions with a controlled knowledge base and a defined human-escalation path. The prototype combines a clean chat interface, a Python backend, model inference, and structured sample facts.
10 Prompts
Manual QA suite reported passed
3 Core Files
HTML · JSON · Python
Grounded
Routes, fares, policies in sample data
Fallback
Unsupported or sensitive requests
● The Customer-Support Challenge
Repetitive Routine Inquiries
Transport operators receive constant questions about departure windows, fares, terminal locations, and baggage rules that tie up staff time.
Risk of Hallucinated Information
Generic AI chatbots freely guess missing details, potentially inventing non-existent bus schedules or quoting inaccurate fares.
Unauthorized Policy Promises
Without strict boundaries, an assistant might improperly promise refunds or compensation during breakdowns instead of escalating to human managers.
● Solution Overview & Workflow
01 · User Question:A passenger opens the floating chat widget on the landing page and types an inquiry.
02 · Python API Call:The browser sends the message over HTTPS to the backend server, keeping API keys securely hidden.
03 · Model Inference & Factsheet:The backend loads structured JSON facts and requests model inference via Groq at temperature 0.2 for strict determinism.
04 · Grounded Answer or Support Fallback:The assistant replies using approved facts or smoothly directs unsupported requests and complaints to phone and WhatsApp contacts.
● Technical Architecture
Customer Question → Chat Widget (index.html) → Python API (server.py) → knowledge_base.json + openai/gpt-oss-120b (Groq API, temp 0.2) → Grounded Answer or Support Escalation
• index.html: Landing page and floating chat widget.
• knowledge_base.json: Structured sample data covering routes, departure windows, fares, terminal locations, baggage limits, and policies.
• server.py: Local Python HTTP server and secure backend API endpoint.
• Credential Security: Groq API credentials reside exclusively in server environment variables, never exposed in client code.
⚡ Interactive Support Assistant Sandbox (Live Demo)
Sanaga Royal Voyages AssistantGrounded Knowledge Base
Test how the AI assistant handles approved travel policies and strictly routes out-of-scope inquiries to human support staff:
Hello! I am the automated support assistant for Sanaga Royal Voyages. How can I help with schedules, baggage, or fares between Yaoundé and Douala today?
* Grounded strictly on verified policy factsheet. Answers without guessing.
● System in Action: Verified Implementation Evidence
Sanaga Royal Voyages demo landing page with embedded customer-support assistant running locally at localhost:8000.
Figure 2: Unsupported Route Handling & Escalation
When asked about Bamenda and Maroua, the assistant recognizes the routes are absent from the schedule and provides configured human-support contacts.
Figure 3: Complaint Handling & Staff Handoff
For a breakdown-related refund request, the assistant directs the customer to the terminal station master rather than making an unauthorized refund promise.
● Quality Assurance & Validation (10-Prompt Test Suite)
The project report describes a 10-prompt manual test suite covering the core response boundaries. All 10 tests were recorded as passed:
✓Known Routes & Fares Accurate Yaoundé, Douala, Bafoussam rates
Central Tuition Management, AI Document Intake & Clearance Verification
Built by Nchia Mbong — AI Automation & Integration Specialist
Eliminate manual re-entry backlogs and fragmented payment tracking.
This production prototype demonstrates an administrative workflow that captures student onboarding, assists with reading bank slips or Mobile Money records, checks extracted data with deterministic rules, maintains a cumulative tuition ledger, generates receipts, and enables QR-based verification.
5 Stages
Intake → Extract → Validate → Ledger → Receipt
Deterministic
Zero AI hallucination in financial arithmetic
Dual Receipts
Enrolment & Tuition Clearance Slips
QR Lookup
Direct record status route verification
• The Administrative & Document Challenge
Manual Re-Entry Backlogs
During peak enrollment weeks, bursary desks get overwhelmed by hundreds of paper deposit slips, manual typing, and endless student queues.
Financial Math Risk in AI
General AI vision models can miscalculate financial balances. Strict separation of extraction from deterministic math eliminates errors.
Audit & Tampering Hazards
Paper receipts can be altered or lost. Without instant digital QR verification and locked central ledger updates, verifying authenticity is difficult.
• Solution Overview & Five-Stage Workflow
01 · Web Intake:Responsive portal allows students or intake staff to submit student registration details and deposit slips or mobile-money screenshots.
02 · Multimodal Extraction:Document analysis proposes candidate structured values (payer name, bank transaction reference, amount, and payment date).
03 · Deterministic Validation:Strict Python rules validate transaction format regex, verify fee schedules, separate non-refundable registration fees (25,000 FCFA), and calculate balances.
04 · Central Master Ledger Update:Atomic record write logs the payment, increments cumulative tuition paid, records the balance, and logs an immutable audit trail entry.
05 · Tamper-Evident Receipts & QR Route:Generates official Enrolment & Tuition Clearance slips with an embedded QR code linking to a direct status-verification page.
• Technical Architecture
Web Intake (index.html) → Python Backend API (server.py) → Multimodal Extraction (OCR/Vision)
→ Deterministic Rule Engine (exact integer math) → Central JSON Ledger (atomic writes) → QR Clearance Slip Engine
• Multimodal Extraction Pipeline: Vision model parses candidate text from noisy images without direct write privileges.
• Deterministic Python Rule Engine: Application logic verifies data types, enforces exact integer subtraction, and rejects arithmetic drift.
• Master Ledger Storage: Structured JSON ledger maintaining running balances, transaction history, and audit log entries.
• Receipt Verification Route: Scannable QR code redirects to an authoritative verification route for real-time clearance confirmation.
