Brand and Project Overview
ITURRI, a global manufacturer of protective equipment and emergency response solutions, was looking for a better way to manage complex B2B quotes associated with its catalog of over 80,000 products. Sales representatives were manually reviewing emails, spreadsheets, PDFs, and forms to identify products and prepare quotes within HubSpot.
Cyberclick designed and implemented a custom AI-powered quoting engine connected to the HubSpot Sales Hub, ERP, ecommerce, and a custom middleware architecture built in Django and Docker. The platform automatically interprets customer requests, recommends products, generates HubSpot line items and quotes, and gives reps a structured workflow for review and approval.
The result was a scalable commercial operation that reduced manual quoting work from hours to minutes while keeping HubSpot at the center of the customer experience.
The Main Challenge
ITURRI manufactures protective equipment for people and vehicles, including uniforms, technical footwear, emergency vehicle bodywork, and medical transport solutions. The company manages more than 80,000 SKUs across the ERP and ecommerce platform, serving customers like fire departments, healthcare providers, military organizations, and public institutions.
As ITURRI expanded across product lines, countries, and customer segments, quoting became one of the company’s biggest operational bottlenecks. Requests came in through emails, PDFs, spreadsheets, forms, and tickets, often with incomplete product details. Reps had to manually search thousands of products, check stock and pricing in their ERP, and build quotes line by line.
The process was slow, inconsistent, and difficult to scale. ITURRI needed a way to reduce manual work without replacing their ERP or changing the way sales teams already worked inside of HubSpot.
The Solution: A Custom AI Quoting Engine Built Around HubSpot
Cyberclick built a custom quoting engine connected to the HubSpot Sales Hub, ERP, ecommerce platform, OpenAI models, and a custom Django application running in Docker containers on Azure.
HubSpot remained the commercial workspace for the sales team. Their ERP was kept as the system of record, while the ecommerce platform provided key product catalog information. Cyberclick connected both systems through its proprietary ERP Connector and a custom middleware layer.
Cyberclick built a multi-system architecture designed to keep HubSpot as the core commercial control center while allowing the ERP and ecommerce platform to remain the systems of record.
The new system included:
- A Django application deployed in Docker containers
- Azure SQL and PostgreSQL databases
- HubSpot APIs for deals, line items, quotes, notes, forms, and tickets
- GPT-based AI models hosted in Azure
- Jenkins CI/CD pipelines
- Nginx reverse proxy with HTTPS
- ERP Connector with retry queues, monitoring, and alerts
Two AI models power the quoting process. The first model reads customer requests from emails, PDFs, forms, notes, and spreadsheets, and then converts them into structured product requests. The second model matches those requests against ITURRI’s catalog and recommends the best products based on category, stock, pricing logic, and historical quote data.
When confidence is low, the system falls back to a similarity algorithm that returns the closest product matches instead of generating an unreliable AI prediction.
How the Process Works
The full quoting workflow now happens directly inside of HubSpot and follows the outlined steps:
- Reps review customer requests inside a Deal.
- The middleware extracts and analyzes the request.
- AI recommends products and quantities.
- Reps review or edit the suggestions.
- HubSpot automatically generates line items and a draft quote.
- Reps finalize, electronically sign, and send the quote.
This project went far beyond a standard HubSpot implementation. It combined custom middleware, AI orchestration, ERP integration, HubSpot automation, Docker infrastructure, observability dashboards, and long-term AI learning from expert feedback.
Results
The solution created measurable operational improvements for ITURRI’s sales process, including:
- Centralized access to more than 80,000 SKUs from HubSpot
- Automated quote creation from unstructured customer requests
- Reduced manual product matching and spreadsheet work
- Full traceability of AI recommendations, rep edits, and final quotes
- Real-time monitoring and logging across integrations
- Activity dashboards for validated and pending quotes
Business Impact
What previously took hours of manual work can now be completed in minutes with rep validation. Sales teams spend less time searching catalogs and more time reviewing customer needs.
The new design also created a continuous learning loop. Every correction made by a sales representative feeds future model improvements, helping the AI become more accurate over time.
The project was designed to scale across countries, product lines, and sales channels without adding significant administrative overhead.
Practical AI Built for Real Operations
AI played a central role in the solution.
Cyberclick used Azure-hosted GPT models paired with custom prompt engineering to process customer requests written in natural language. The models analyze emails, notes, PDFs, spreadsheets, forms, and tickets. They then go on to transform that information into structured product recommendations.
The AI layer also:
- Categorizes products against the ecommerce Platform and ERP taxonomies
- Detects missing or unclear information
- Flags stock or availability issues
- Learns from historical quote corrections
- Stores error cases for future retraining
The use of AI in this project was not simply for show or random testing. The system was designed strategically with operational controls, fallback logic, human review, persistent logging, and long-term learning capabilities suitable for a high-volume industrial sales process.
Why This Project Stands Out
This project demonstrates how HubSpot can become the commercial layer of a highly technical industrial business when combined with strong engineering and AI.
Cyberclick connected HubSpot, ERP, ecommerce platform, Azure AI services, PostgreSQL, Azure SQL, Docker infrastructure, and a proprietary ERP Connector into a single quoting experience that works for the sales team.
Now, the only thing visible to the sales team is a clean workflow inside of HubSpot, but behind the scenes, the platform coordinates APIs, AI models, product catalogs, monitoring systems, retries, and quote generation logic in real time.
What We Learned
The project also reinforced three important lessons for complex industrial businesses.
First, HubSpot does not need to replace ERP systems to create operational value. When every platform performs its specific role well — ERP as the transactional system, ecommerce platform as the product information layer, and HubSpot as the central sales platform — the result is faster, more scalable, and easier for teams to manage.
Second, AI only works efficiently when it is backed by a strong plan and execution. The solution combined GPT models, fallback similarity algorithms, persistent logging, human validation, monitoring, and continuous learning from rep feedback to keep recommendations reliable and traceable.
Third, the best technical systems are often invisible to the end user. Sales reps simply review recommendations and validate quotes within HubSpot, while the underlying infrastructure manages integrations, AI orchestration, ETL processes, retries, observability, CI/CD pipelines, and security controls in the background.
For Cyberclick, the first HubSpot Elite Partner agency in Spain and one of the top-rated Spanish companies in the HubSpot Solutions Directory, the ITURRI project reinforces a core belief that HubSpot can become the commercial backbone of a highly complex enterprise operation when combined with thoughtful architecture and practical AI adoption.
For ITURRI, the project created a faster, smoother, and more scalable quoting process while preserving the technical rigor and customer focus that define the company’s work in safety, protection, and emergency response.
Alessandra Valqui
Leave your comment and join the conversation