By the editorial team airassvet.ru. This guide was prepared on September 29, 2026 for managers of transportation and freight forwarding companies, dispatch teams, and shippers’ logistics departments. The rules are quoted from Article 8 of Federal Law No. 259-FZ dated November 8, 2007, “Charter of Motor Transport and Urban Ground Electric Transport” as amended by Federal Law No. 140-FZ dated June 7, 2025 (ConsultantPlus). Information about the no-penalty period is based on a publication on nalog-nalog.ru dated August 27, 2026. The legal sections are for informational purposes only and do not replace legal advice. The editorial team has no commercial relationship with the services mentioned, and there are no affiliate links.
Short answer. Starting September 1, 2026, under Part 1 of Article 8 of the Charter of Motor Transport, the transport waybill is issued as an electronic transport waybill (ETW), and under Part 5 of the same article, the order and shipping request are issued electronically; the Ministry of Transport determines the cases where paper documents are allowed. For a transportation company, this means the data for each shipment must be structured and accurate before the truck leaves. AI is useful at the intake stage: it turns requests from emails, messengers, and phone calls into structured data, checks them for completeness and inconsistencies, and prepares a draft for the dispatcher. Signing the ETW with an electronic signature and making transportation decisions remain human responsibilities.
Below is what the law actually requires, where the carrier’s bottleneck is, what AI can handle, and how to test it in a pilot.
Contents
- What changed on September 1, 2026
- Where the carrier’s bottleneck is
- What AI does in a transportation company
- What to leave to dispatch and accounting
- Data checks before the ETW
- Integrations: TMS, 1C, and the EPD operator
- How to launch a pilot
- Launch checklist
- Frequently asked questions
What changed on September 1, 2026
Federal Law No. 140-FZ dated June 7, 2025 amended Article 8 of the Charter of Motor Transport. In the current version:
| Rule | What it says | What it means for the carrier |
|---|---|---|
| Part 1 of Article 8 | The conclusion of a freight transportation contract is confirmed by a transport waybill; the transport waybill is issued as an electronic transport waybill; the Ministry of Transport determines the cases where paper form is allowed | ETW is the main format, paper is the exception |
| Part 3 of Article 8 | Cargo without a transport waybill is not accepted by the carrier for transportation (except as provided in Part 1 of Article 18) | Without ready data, the truck will not leave |
| Part 4 of Article 8 | The shipper provides documents required by sanitary, customs, quarantine, and other rules, or indicates their registration numbers in the waybill if the information is available in government systems | Supporting document details are required |
| Part 5 of Article 8 | A contract may be concluded by accepting an order or shipping request for performance; the order and request are issued as an electronic order or electronic request | Requests from emails and messengers must be converted into electronic format |
The right-hand column is the editorial interpretation. The ETW format is approved by the Federal Tax Service in coordination with the Ministry of Transport (Part 2 of Article 8), and the filing procedure is governed by cargo transportation rules.
According to nalog-nalog.ru from August 27, 2026, the transition deadline has not been postponed, and until March 1, 2027, if a road inspection finds a paper waybill instead of an electronic one, the inspector limits the response to an oral warning. This is a no-penalty period on the road, not a postponement of the obligation; check the current status of the moratorium with official sources.
The main practical takeaway: after September 1, 2026, the quality of the request data determines whether the truck leaves on time. An error in the address, weight, or details can no longer be corrected by hand on a paper form — it has to be corrected in the electronic document and coordinated with the other parties to the shipment.
Where the carrier’s bottleneck is
Editorial observation on the process structure: requests arrive in different formats — by email with an attachment, in a messenger, by phone, sometimes as a photo of a paper form. The dispatcher manually enters addresses, dates, cargo characteristics, and contacts into the transportation system or 1C, clarifies missing information, and only then can prepare the documents.
Typical errors at this stage:
- Incomplete request. Missing weight, dimensions, number of pieces, unloading contact person, or receiving window.
- Inconsistencies. The cargo weight exceeds the assigned vehicle’s payload, the loading date is later than the unloading date, or the address does not match the contract address.
- Discrepancies in company details of counterparties: TIN, KPP, company name.
- Lost clarifications. The client changed the loading time in the message thread, but the old time remained in the system.
After the move to ETW, each such error becomes a delay or an electronic document correction.
What AI does in a transportation company
An AI agent for a transportation company is a software tool built on a large language model and connected to email, messengers, the transportation system, and the accounting system. It extracts shipment data from unstructured requests, checks it, and prepares drafts for the dispatcher. Typical feature set:
- Incoming request parsing. From an email, attachment, or message, the agent extracts the route, dates and time windows, cargo characteristics, transport requirements, and contacts.
- Document recognition. Scans and photos of forms, certificates, and powers of attorney are converted into data.
- Completeness check. If anything needed for the electronic request and ETW is missing, the agent immediately sends the client a follow-up question.
- Consistency check. Weight against payload capacity, dates, addresses, and company details against reference data.
- Request draft in the transportation system or 1C for dispatcher approval.
- Customer status updates. Answers to “where is the truck” and “when is unloading” based on transportation system data.
We described in detail how AI parses requests from email in the article automating email requests with AI, and document scan recognition in the article OCR and document recognition for business.
What to leave to dispatch and accounting
The breakdown below is the editorial recommendation based on risk level.
| Task | AI | Human |
|---|---|---|
| Extracting data from emails, messages, and scans | Executes | Performs spot checks |
| Clarifying questions to the client about missing data | Sends using a template | Handles nonstandard cases |
| Checking for completeness and inconsistencies | Executes according to rules | Decides how to fix it |
| Confirming the request, assigning the truck and driver | Prepares a draft and options | The dispatcher makes the decision |
| Signing the electronic transport invoice (ETn) and other electronic transport documents (ETD) | Does not perform | Authorized persons using an electronic signature |
| Changes and adjustments after departure | Prepares a draft correction | Responsible employee |
| Rate and shipping terms | Does not quote without an approved price list | Sales manager |
The main rule: AI prepares and checks the data, but does not sign documents or confirm transportation. Signing the ETn is a legally significant action, and the company is responsible for it.
Data verification before the ETn
The most valuable part of the project is a set of verification rules that stops a request before it becomes a document. Example of a minimum set:
- all required fields in the electronic request are completed;
- the shipper and consignee details match the counterparty master data;
- the cargo weight and volume do not exceed the assigned vehicle's capacity;
- loading and unloading dates and time windows are logically consistent;
- for cargo that requires supporting documents under Part 4, Article 8, their details are provided or the documents are attached;
- changes from correspondence after the request is created are reflected in the system.
It is better to store verification rules separately from the model, in a reference guide that the dispatch team can update without a programmer. The model is responsible for understanding the text, and the rules determine what counts as an error.
Integrations: TMS, 1C, and ETD operator
The ETn is signed by the transportation participants and transmitted through the electronic transport document operator service that the company uses. AI does not replace either the operator or the accounting system — it sits in front of them:
- Email and messaging apps — the source of requests.
- Transportation system or 1C — where the agent creates a draft request and pulls master data from.
- ETD operator service — where verified data is sent for document creation and signing.
We covered how to connect an AI agent to 1C in the article how to connect AI agents to 1C. Route planning and fleet utilization forecasting is a separate task for ML models; an overview of those approaches is in the article AI in global logistics.
How to launch a pilot
A process that can be repeated in a transportation company of any size:
- Choose a workflow. For example, requests from one or two major clients, or only requests coming in by email.
- Record baseline metrics for 2–4 weeks. Time from request receipt to confirmation, the share of requests requiring clarification, the number of document corrections after departure, and the share of requests with errors in details or cargo characteristics.
- Build a test set. 100–200 past requests in their original form, including incomplete, inconsistent, scanned, and correspondence-based changes.
- Test the agent on the set before launch. For each request, compare the extracted data with what ultimately went into the documents: accuracy by field, inconsistencies found and missed, and the correctness of clarifying questions.
- Launch with oversight. For the first few weeks, the dispatcher approves every draft. Expand the workflow when extraction errors are isolated and understandable, and the verification rules catch known error types.
The pilot scope is our recommendation. For metric selection, see the article AI agent performance metrics.
Pre-launch checklist
- The required fields for the electronic request and ETn have been defined for your transportation types.
- Verification rules for completeness and inconsistencies have been created and an owner assigned.
- Master data for counterparties, vehicles, and addresses is up to date and accessible to the agent.
- The agent does not sign documents or confirm transportation without a dispatcher.
- It has been defined how the agent records changes from correspondence after the request is created.
- Transfer of verified data to the transportation system or 1C and then to the ETD operator has been configured.
- It has been decided where customer data is processed and which model is used.
- Baseline metrics and a test request set have been recorded.
- A person responsible for reviewing drafts in the first few weeks has been assigned.
Frequently asked questions
Will the electronic transport invoice be mandatory starting September 1, 2026?
According to Part 1 of Article 8 of the Charter of Road Transport, as amended by Federal Law No. 140-FZ, the transport invoice is issued as an electronic transport invoice, and the cases when paper is allowed are determined by the Ministry of Transport. According to nalog-nalog.ru, the transition deadline has not been postponed.
Will there be fines for a paper invoice?
According to nalog-nalog.ru as of August 27, 2026, until March 1, 2027, during roadside inspections a paper invoice instead of an electronic one results only in a verbal warning. This is not a deferral of the obligation; check the status of any moratorium through official sources.
What can AI do in a transportation company?
Process requests from emails, messengers, and scans; check data completeness and inconsistencies; ask the client clarifying questions; prepare a draft request in the transportation system or 1C; and respond to clients about shipment status.
Can AI sign the ETn?
No. Signing the ETn and other electronic transport documents is a legally significant action performed by authorized persons using an electronic signature. AI prepares and checks the data before signing.
Where should a carrier start implementing AI?
From a single flow of requests, baseline metrics for confirmation time and the number of document corrections, a test set of past requests, and data validation rules that dispatch can change without a programmer.
How the airassvet.ru team helps a transportation company move to e-waybills without manual data entry
If dispatchers manually copy requests from emails and messengers, and data errors turn into corrections to electronic documents, the team airassvet.ru can:
- audit the process from request intake to e-waybill creation: channels, fields, common errors, and manual entry points;
- set up recognition of requests and supporting documents, including scans and photos;
- develop an AI agent that extracts and checks request data, and integrate it with email, a transportation management system, or 1C, with testing on a set of real requests.
The first step is to choose one process, for example requests from email from one major client, document how it works now, what data and constraints it has, and what metric will determine whether the pilot is successful. If the process is selected, discuss the project.
Conclusion
Starting September 1, 2026, the waybill under Part 1 of Article 8 of the Road Transport Charter will be prepared electronically, and the order and request under Part 5 will be created as an electronic order or electronic request. For a carrier, this shifts the main risk to data quality before the truck leaves. AI for a transportation company closes exactly this gap: it processes requests from emails, messengers, and scans, checks for completeness and inconsistencies, and prepares a draft for the dispatcher, while signatures and decisions remain with people. Start with one request flow, validation rules, and a pilot that confirms each draft.