Thermal insulation manufacturing · Astratek
How Astratek made sales autonomous in thermal calculations - and removed the load from a key expert
TLDR
Astratek is a liquid thermal insulation brand. Every sale requires a thermal calculation for the customer's specific facility. The calculations were done by one specialized expert, so the sales team depended on him completely and waited for hours. BackendFlow developed an AI assistant: the manager enters request data in a chat, the system validates the inputs, and returns a ready calculation in 2 minutes.
Result: the sales team is autonomous. Calculation preparation time dropped from 3 hours to 2 minutes. The expert's routine load was removed.
Context
Astratek's product is technically complex: liquid thermal insulation is used on industrial facilities, and before purchase the customer always requests a thermal calculation for their specific object. Without the calculation, there is no commercial offer.
The calculations were done by one person - a business partner of the company with deep domain expertise. The sales team passed him the request data and waited. On average, one calculation took 3 hours. During peak periods, the queue stretched out.
Two consequences of this setup:
For the sales team: response speed to the client depended not on the manager, but on the workload of one specialist. Sending a commercial offer quickly was impossible.
For the business partner: a significant part of working time was spent on standard calculations.
"When there is too much routine, there is simply no time to think about strategic development"
Task
Give the sales team a tool for preparing thermal calculations independently, without involving the expert in every request. Response speed to the client should not depend on the expert's workload.
Solution
BackendFlow developed an AI assistant integrated into the sales team's messenger.
Workflow:
- A customer request arrives
- The manager enters the facility data into the chatbot
- The system validates the inputs: checks completeness and correctness of parameters, and requests missing data if needed
- The bot performs the thermal calculation using the company's methodology
- The ready calculation is returned to the manager in a commercial-offer-ready format
The business partner is involved only for non-standard facilities with special conditions - cases where expert judgment is truly required.
Insights
Input validation turned out to be the key step.
A significant part of delays in the old workflow came from incomplete request data. The manager passed along whatever they had, the expert asked clarifying questions, and the cycle stretched out. Built-in validation removed this loop: the system highlights missing information to the manager before the calculation starts.
The sales team gained a new negotiation tool.
When a calculation takes 2 minutes, a manager can prepare several options during a call with the client. Speed became a competitive advantage.
The expert's methodology now scales.
The business partner's knowledge, which previously existed only in his head, has been formalized and made available to the entire sales team.
Results
| Metric | Before | After |
|---|---|---|
| Calculation preparation time | ~3 hours | 2 minutes |
| Dependence on the expert | Complete | Only non-standard cases |
| Sales team autonomy | Low | Full |