Custom AI CPQ development for complex quotes
We design and develop Configure, Price, Quote systems with custom AI integrations built around the company’s needs.
These are true AI-assisted quoting and configuration systems: product selection, compatibility and discount rules are handled by the software, so they are thoroughly tested and their behaviour is reproducible; AI instead interprets customer requests and enters them into the calculation system.
What is an AI CPQ?
A CPQ guides product configuration, applies the rules that determine the price and generates the commercial proposal. In an AI CPQ, natural language accelerates data extraction and retrieval, but does not replace the engine that governs compatibility and calculations.
It is suitable when the quote depends on compatible components, bills of materials, dimensional variants, commercial terms and margin thresholds.
This is a perfect solution when the sales team’s workload is heavy and represents an obstacle to growth, or when the company realises that response times for quote requests are too long and result in lost opportunities.
Deterministic engine, AI at the surface
Configuration and pricing rules live in versioned, testable and explainable code. AI interprets unstructured requests received via email, WhatsApp or dedicated forms and translates them into standard input sets to pass to the configuration engine, which returns configurations and makes the catalogue searchable, without inventing prices or feasibility.
The number of iterations, the direction of the flow, human review methods and integration are decided case by case around the company’s needs, objectives and technology environment.
Request extraction, analysis and categorisation
This part of the flow is shared with qualification agents: information is extracted from every RFQ (request for quotation) received in natural language, and missing information is identified. Depending on the company’s requirements, an automatic response can be designed to obtain all the data needed to prepare a quote, including in the form of a dynamic questionnaire containing only the relevant fields, to facilitate the categorisation of responses and process the data without sending it unnecessarily to the AI model.
Again, according to the criteria agreed with the company, confidence levels are established below which the request is immediately forwarded to the sales team or ignored.
We can integrate different channels for receiving requests: email providers such as Google Workspace, Microsoft Outlook or Italian providers such as Aruba, WhatsApp messages, or requests submitted through forms on the website.
Examples of AI CPQ systems for different company objectives
Very different flows can be built depending on the objective.
Example 1 A company may be interested in giving the customer a price range before arranging a meeting with a salesperson or a technical site visit, thereby avoiding investing time and resources in an out-of-target lead. To handle a case of this kind, we can create a response generated through a template in which the CPQ engine is responsible for identifying configuration candidates and adding tolerances and safety percentages. In this case, after an initial testing phase and if the confidence percentage stated by the model is sufficient, the response could be forwarded directly.
Example 2 A company may instead want to retain full control over its offers and respond to customers only with finalised proposals. In this case, the AI CPQ acts as an assistant and uploads draft quotes to the CRM or ERP for approval and sending by the team.
CPQ and 3D configurator
When geometry helps customers choose or validate a product, a 3D configurator can become the interface for the same rules engine, accompany the quote sent to the customer, and be used to refine the proposal and convert the order. We offer AI CPQ and 3D Configurator as a single project, not as two estimates to be added together.
When a CPQ is not needed
If the price list is simple and the work mainly consists of reading emails and collecting missing data, an agent for quote requests is more proportionate.
An agent dedicated exclusively to qualification is often more suitable when the company does not rely on product codes, but calculates custom quotes, and quote requests are not too numerous. In these cases, the return on investment of the configuration and pricing engine is lower than what can be achieved by first investing in speed-to-lead and information gathering.
Pricing
From €15,000 one-off. The typical range is €15,000–40,000, depending on the catalogue, rules and integrations.
From €450/month for hosting, monitoring, support and oversight of the rules and integrations that change along with the catalogue.
Frequently asked questions
What is an AI CPQ?
It is a Configure, Price, Quote system in which a deterministic engine governs configuration and pricing, while AI facilitates information gathering and use of the system.
Does AI decide the price?
No. Prices, discounts, compatibility and margin thresholds remain in versioned, testable rules. AI is used at the surface, where it handles language and unstructured data.
Is the 3D configurator a separate project?
No: when needed, the 3D interface, product rules and quoting are designed as a single system and receive a single estimate.