Note: this is an illustrative example, based on typical commercial process automation projects. The data is representative and the company has been anonymized.
A B2B SME automated the cycle that runs from a request received by email through to the order being logged in the ERP. Rather than replacing the existing tools, the solution handled the repetitive part, triaging the mail, reading the requests and filling in the quotes and orders, leaving the commercial decision with the team. The gain was in response time and reliability.
The context
Customer requests arrived in a shared mailbox, in free text and often with lists or attachments. The sales team read each message, interpreted the request, checked prices and availability, drew up the quote and, once it was approved, logged the order in the system. It was continuous work, dependent on people's availability and hard to keep up with on the busiest days.
The challenge
The process combined characteristics that made it heavy and error-prone:
- Shared mailbox: requests were mixed in with the rest of the correspondence, with a real risk that a message would go unanswered or take too long to be dealt with.
- Free-format requests: each customer wrote in their own way, with no fixed format a rigid bot could parse.
- Manual transcription: copying items, quantities and terms into the quote and the ERP took time and introduced errors.
- Response time: the longer a quote took, the higher the risk of losing the business to a faster competitor.
The solution
Engibots automated the cycle by combining mail triage, AI-based reading and integration with the management system (see ERP and CRM systems integration).
- Mail triage: the solution monitors the shared mailbox, identifies the messages that are quote or order requests and separates them from the rest of the mail.
- Intelligent reading: artificial intelligence reads the body of the message and the attachments, interprets the free text and extracts the items, quantities and terms, even when the format differs from customer to customer (see artificial intelligence in document processing).
- Quote and validation: the data is cross-checked against pricing and availability information, the quote is prepared using the company's template, and the salesperson reviews and approves it, with doubtful cases flagged.
- Order logging: once the quote is approved, the order is logged in the ERP with no further data entry, and the customer is notified.
The decision to automate the administrative part instead of imposing a new platform reduced resistance from the team and focused the investment on the point of greatest waste.
Results
Illustrative results of this approach:
- Response time: quotes prepared the same day, instead of sitting in a queue waiting for availability.
- Errors: fewer errors in quotes and orders, thanks to eliminating manual copying.
- Traceability: no request lost in the shared mailbox, with every step logged.
- Sales focus: the team started spending its time on the customer relationship and selling, instead of admin work.
- Capacity: demand peaks are now absorbed without adding headcount.
Lessons
- The bottleneck is at the entry point: in a sales operation, it's in the mailbox that requests pile up and get lost. That's where automation pays off fastest.
- Free format calls for artificial intelligence: requests written by different people have no fixed structure, which is why AI is better suited than a rigid bot.
- The decision stays with the human: automation takes care of the repetitive work, commercial approval remains with the team.
- Integration closes the loop: the value is only realized once the order enters the ERP without being typed in again (see how to calculate the ROI of automation).