Workflow automation to Sephyre with TRIARE team

Sephyre is a Swiss-based logistics company that offers a comprehensive range of services for dangerous goods transportation: consulting, support, packaging, delivery, and document preparation. Having numerous complex logistics processes interconnected with each other, the client wants to automate core processes, integrate them with delivery services like DHL, FedEx, and DB Schenker, and significantly reduce execution time.

Thus, TRIARE’s remit was to work on AI workflow automation to turn a manual, multi-system order process into one connected pipeline where data is entered once and flows automatically to every downstream system. Our team suggested building a CRM system to address the issue.

The CRM was deliberately positioned as a bridge/integration layer between all internal processes, not just a records database. That was an initial concept for the CRM – it was supposed to store orders received from the website and facilitate communication between the website, the system for generating dangerous goods documentation, and delivery services.

We started with an MVP build and an end-to-end workflow creation – from the customer placing an order to tracking it after it has been handed over to the carrier. The whole process includes:

– Website order form + order tracking page.

– Automatic order and company creation in the CRM.

– Creating consignments, addresses, and products in DGAssistant.

– Fetching DG transport documentation from DGAssistant.

– Generating a Proforma invoice.

– Shipment creation in DHL, FedEx, and DB Schenker.

– Pulling tracking info from those carriers back into the CRM.

– Automated customer email notifications (order created, confirmed, shipped, tracking).

As a result, the CRM became a central hub where data flows through each stage of order processing. This has made it possible to reduce the number of manual operations and avoid having to re-enter the same information into multiple systems.

We also want to emphasize that automation plays a great role in the dangerous goods transportation sector, as it requires even stricter rules for classification, packaging, shipment preparation, documentation, and the carrier’s work.

​​UNECE documents explicitly state that the required information on dangerous goods may and should be transmitted via electronic data interchange (EDI/EDP) instead of paper documents.

The Sephyre challenge: Manual, fragmented logistics processes before TRIARE

Before the TRIARE team automated workflows for Sephyre, their departments processed orders within three separate systems that don’t exchange data automatically. All goes manually at every step – receiving an order, preparing documentation for dangerous materials, and creating a shipment.

The customers’ requests were incoming through email or company messengers as a PDF file or a few attached documents. Then, the operator manually entered the same order data at least three times. First, into the Bexio CRM (order management + accounting). Then re-entered into the DG documentation software (DGAssistant) to generate transport docs. And finally, re-entered everything again on the carrier’s website to create the shipment.

What’s more, every time a new system was integrated into the process, the data had to be re-entered again. It’s a major part of the work, handled entirely by hand.

Such a situation causes a few issues at the same time. The biggest one is the time operators spend on transferring information instead of working on the order itself. With every new data entry, there was a chance to make a mistake. For dangerous goods, even small details like the UN number, shipping name, packing group, and cargo quantity can affect transport and documentation. Another bottleneck was the lack of a single source of up-to-date information. Order status was not updated automatically, tracking information was not synced back to the CRM, and customers had to be notified of the delivery status manually.

Simply put, the main gap was having no end-to-end workflow where data could be entered once and automatically passed through the entire process.

How TRIARE analyzed Sephyre’s dangerous goods logistics workflow

TRIARE experts kicked things off with a discovery phase to get a detailed picture of Sephyre’s operational processes, data requirements, and integrations with external systems. This helps us to create automation logic based on real company workflows.

Once we finish this stage, we get the project’s formalized database in hand. It includes a User flow diagram (Miro), UI wireframes for both the CRM admin panel and the website (Figma), a C4 architecture diagram, a database structure, a Site Map, and the SRS itself.

Within the discovery phase, we broke down the transportation process into separate stages and mapped out how data should flow between them.

It looks like this: order intake → company/contact creation → DG classification & products → document generation → carrier booking → tracking → invoicing → customer notification.

Special attention TRIARE experts put on the external systems’ requirements. The team analyzed exactly what data is needed by DGAssistant, by each carrier’s API, and by the Proforma invoice.

The analysis covered regulatory frameworks such as ADR for road transport, IATA/DGD for air transport, and the IMDG Code for maritime transport. Logistics-related concepts and data were also incorporated into the process model – Incoterms, HS commodity codes, EORI, TIN, and MSDS/SDS. This gave us a way to build a system that reflects real-world terminology and domain requirements.

We also clearly defined the MVP scope: the team focused on the administrative side, while the client portal and additional features were planned for subsequent phases. We also identified key assumptions and risks in advance, including responsibility for any changes to third-party service APIs after launch.

TRIARE’s automation strategy: Creating one unified CRM ecosystem

The key idea of the document workflow automation comes down to creating one unified CRM ecosystem. Operators enter order data once, and it automatically transfers to all systems involved in further processing.

The CRM system worked as a central operational hub, connecting the website, DGAssistant,  carriers, invoicing, and customer email. Besides this, it also becomes a system of record where the following data are stored with updated statuses and used: companies, orders, products (DG goods), documents, delivery, and tracking.

One more thing we also automated is the process of creating and searching for companies. Whenever a new order came in from the website, the system checked the name of the sending company against its existing records. If the company existed, the order was automatically linked to it, and the contact information was updated. If no match was found, the CRM automatically created a new company.

Every CRM module has its own API, through which the data could be exchanged with other components of the ecosystem.

This architecture laid the foundation for the current automation and for the system’s future development. Later, TRIARE developers added an MCP layer to the existing backend API, making its data and operations available through an AI assistant using natural language.

How TRIARE designed a user-friendly workflow for complex logistics operations

Since logistics operations include managing tons of data, documents, and regulations, the TRIARE team strove to make complex workflows more understandable for operators.

The order page was logically divided into several blocks: Company, Date & Status, General Info, Delivery Info, Shipper, Consignee, Products, and Documents.

Then, we moved to the status logic. Within the CRM system, orders come through the predefined lifecycle: New → In Progress → Processing → Shipped Out → On the Way → Delivered. For specific scenarios, the Declined and Rejected statuses were also added.

All the actions in this chain depend on the current operation status. For example, an operator sees “Accept,” “Generate Transport Documents,” or “Mark as Delivered” only when the corresponding action is relevant.

Orders placed on the website are automatically created in the CRM and already contain the data received from the customer. At the same time, the operator can verify and supplement any information that is insufficient for generating DG documents or creating a shipment.

Other UX solutions also reduce the number of manual steps:

– Address fields use Google API autocomplete.

– Insurance is automatically calculated based on the cost and goods quantity.

– The shipment date is pulled from the pickup date.

– The default currency is set to CHF.

The system also has built-in guardrails: validation of required fields, notifications about unsaved changes, deletion confirmation, and success notifications. For example, the pickup window cannot be shorter than two hours.

To avoid overloading the form with unnecessary fields, the system displays additional parameters depending on the selected shipping method. For Air Freight, displayed fields related to IATA requirements, as well as IMDG-IMO fields for Sea Freight.

TRIARE’s best practices for API integrations and automated document processing

For the Sephyre project, the TRIARE team built integrations so that external systems do not bypass the core business logic. All transactions go through the backend, where dangerous goods validation, invoice calculations, and business rules for working with carriers are executed.

We start by working on handling errors in external systems. If DGAssistant rejects the data, the CRM shows the operator the specific fields containing errors. These can be corrected, and the document can be regenerated. If the external service is unavailable, the system also notifies the user with a clear message.

The automated workflow does not block the process if integration is unavailable. For carriers other than DHL, FedEx, or DB Schenker, tracking information can be added manually. Documents can also be uploaded manually or edited in DGAssistant, and the updated version returned to the CRM.

An operator can open the relevant shipment in DGAssistant, make changes, and return the updated document to the CRM. Generated documents are automatically attached to the order. The Proforma invoice is generated immediately in both PDF and CSV formats, for human review and for subsequent machine processing.

The CRM also receives real-time status updates from carriers via API, and every change is timestamped in the delivery timeline. If an issue with the shipment is resolved by the carrier, the system can automatically return the order to the “Shipped Out” status.

All integrations adhere to actual logistics standards – HS codes, ISO country codes, Incoterms, UN numbers, and packing groups.

How TRIARE automated shipping, invoicing, payments, and customer notifications

After we automated the core processes connected with document processing, we integrated Sephyre’s key back-office operations into a single workflow. Orders no longer had to be manually transferred between systems: information flowed from the CRM to shipping, documentation, payment, and customer notifications.

Orders from the CRM are forwarded to the specific carrier, and the delivery status is automatically updated in the system. All necessary documents are also stored along with the order: shipping documentation for dangerous goods, MSDS, and additional files. This provides the team with a single location to manage documents related to a specific shipment.

A Proforma invoice is generated automatically based on order data, including products, cost, shipping terms, and insurance.

The system also takes into account the invoice payment status and terms of business with a specific customer. For customers who pay in advance, order fulfillment can continue after payment is confirmed, while for customers with deferred payment, the system monitors unpaid invoices and alerts the team to them.

From the customers’ side, they receive automatic email notifications at key stages: after the order is created, confirmed, and shipped. The shipment notification includes tracking information, and the link can be resent if necessary.

All emails are formatted in accordance with Sephyre’s style to ensure that automated communication remains part of a cohesive customer experience.

Sephyre results: How TRIARE reduced order processing time to one minute

The most notable result Sephyre received from the TRIARE team’s implemented solutions is significantly less time spent on manual order processing.

After the solution was implemented, the process changed: data is entered once via a single form and is then automatically transferred to the necessary systems. Based on this data, orders and company information are created, documentation for dangerous goods and pro forma invoices is generated, a shipment is created, and the customer receives a corresponding notification.

It was precisely this change in the process mechanics that made it possible to reduce order processing time to about one minute: instead of manually entering data sequentially into multiple systems, there is now a single form and automatic data transfer.

In addition to saving time, automation has delivered several other important benefits to Sephyre:

– reduced risk of errors when re-entering data on hazardous cargo;

– automatic statuses and tracking, eliminating the need to manually check shipments;

– single source of up-to-date information on orders, documents, delivery, and invoicing;

– less manual work for the team at every stage of order processing.

Thus, the key result was not simply speeding up a single operation, but restructuring the workflow itself: most of the repetitive work is performed by the system, while the operator focuses on process control.

Conclusion

The Sephyre case shows that effective automation starts from a deep understanding of how the entire business workflows are structured and work together. The TRIARE team brought together a workflow analysis, well-designed UX architecture, and integrations to transform a complex logistics process into a manageable digital workflow.