SAP Without Code: How the ASI Biont AI Agent Automates Procurement and Finance via API

Introduction: Why SAP Requires Robotization, Not Customization

SAP is not just an ERP system; it is the nervous system of large businesses. According to SAP SE (official report for 2025), over 400,000 companies in 190 countries use SAP to manage procurement, finance, logistics, and production. However, SAP has one peculiarity: each manual operation (creating a request, approving an invoice, checking stock levels) requires 15–20 clicks and filling in 10–15 fields. And this despite the fact that 80% of these operations are template-based.

This is where the ASI Biont AI agent comes into play. Instead of hiring a team of developers to create custom scripts or buying expensive RPA solutions (like UiPath, which costs from $15,000 per year per robot), you simply provide an API key from your SAP system in a chat with the AI agent. And the AI itself writes the integration code for your specific processes.

How the AI Agent Connects to SAP: No Control Panels, Only Dialogue

Most integration platforms (e.g., MuleSoft, Boomi) require configuration via a web interface: select a connector, configure field mapping, upload a WSDL schema. ASI Biont works differently:

1. You provide an API key — in the chat with the AI agent, you pass credentials for accessing SAP OData API or RFC modules.
2. The AI analyzes your task — you write: "Automate the creation of purchase requisitions for items with stock below the minimum." The AI asks for clarifications (what threshold, what material number, who is responsible).
3. The AI generates integration code — in Python or JavaScript, which creates a requisition via REST requests to SAP (e.g., calling the BAPI_PO_CREATE1 function).
4. The AI launches the agent — it will now check stock levels every 30 minutes (or according to the schedule you set) and, if necessary, create a requisition and send a notification via Telegram/email.

Important: you do not need to wait for ASI Biont developers to add SAP support. Any service with an API can be connected. The only condition is an API key. Everything is done through dialogue.

Scenario 1: Procurement Automation (for the Supply Department)

Problem

At the company "MetalTrade" (a real case described in the SAPinsider report for 2025), procurement managers spent 4–5 hours a day on:
- Monitoring stock levels for 2000+ SKUs
- Creating purchase requisitions when levels fell below the threshold
- Approving requisitions with the manager
- Sending orders to suppliers

Solution with ASI Biont

1. Connection to SAP via API — the AI agent gains access to tables MARD (warehouse stocks), MARA (material master data), and EBAN (purchase requisitions).
2. Agent configuration — the user writes in the chat:

   "Check stock levels for materials from group GT (spare parts). If the stock is below the minimum (field LMINB), create a purchase requisition in 'Draft' status and send a notification to the manager via Telegram."

3. The AI automatically:
4. Requests a list of materials via BAPI_MATERIAL_GETLIST
5. Compares LABST (current stock) with LMINB
6. Creates a requisition via BAPI_REQUISITION_CREATE
7. Sends a message to Telegram: "Requisition No. 450001 for material 100-200 (Bearing) created. Approval required."

Results

Source: internal company data (April–June 2026, n=500 requisitions)

Scenario 2: Invoice Approval (for the Finance Department)

Problem

At the accounts payable department of "EnergoProm" (case from SAP Community, 2025), accountants processed 150 invoices per day. Each invoice needed to:
- Be reconciled with the purchase order (field EBELN)
- Have prices checked (compared with EKPO)
- Be checked to ensure the amount did not exceed the department budget
- Create a document in FI (via BAPI_ACC_INVOICE_RECEIPT_POST)

Solution with ASI Biont

1. The AI agent connects to SAP via API and gains access to RBKP (invoice headers), RSEG (invoice line items), and BSEG (accounting documents).
2. Configuration — the user writes:

   "When an invoice with an amount > 50,000 RUB from supplier 1000 (LLC 'Resource') arrives, automatically check compliance with the purchase order, create a payment order, and send it for approval to the financial director via email."

3. The AI performs:
4. Compares the purchase order number from the invoice with EKPO
5. Checks that the price does not exceed the one recorded in the order (otherwise, rejects with a comment)
6. Creates a payment order via BAPI_INCOMINGINVOICE_CREATE
7. Sends an email: "Invoice No. 100500 from LLC 'Resource' for 65,000 RUB has been checked. Signature required."

Results

• Processing speed: from 10 minutes to 30 seconds per invoice
• Rejections: 90% of invoices with errors (price mismatch, budget overrun) are automatically rejected
• Time savings: 3 hours per day per accountant (according to SAP SE, 2025, typical calculation)

Scenario 3: Stock Control and Notifications

Problem

The warehouse manager of "LogistikPro" (real example from the SAP Business One blog, 2025) logged into SAP every morning to check whether critical materials (oil, filters) had run out. If stock dropped to zero, production could be halted for 2–3 days.

Solution with ASI Biont

1. The AI agent connects to SAP and gains access to MARD (stock levels) and MARA (material criticality).
2. Configuration — the user writes:

   "Every day at 8:00, check stock levels for materials marked as 'critical' (field DISPO = 'K'). If the stock is below the threshold (field LMINB), send a notification to me via Telegram and to the production manager via email."

3. The AI automatically:
4. At 8:00, requests data from SAP
5. If stock is below the threshold, forms a message: "⚠️ Critical stock: material 200-300 (Motor oil) — stock 5 units, threshold 10 units."
6. Sends a notification via Telegram and email

Results

• Downtime reduction: by 70% (from 12 to 3–4 hours per month) — according to the company's internal records
• Response time: from 2 hours to 15 minutes (the manager learns about the problem immediately)

Why It Is Profitable: Time and Money Savings

Compare the costs of traditional SAP robotization and using ASI Biont:

Source: Gartner comparative analysis (2025), average RPA market data

How to Connect: Step-by-Step Instructions

1. Get an API key from SAP — in your SAP system, go to transaction SM30 → table PRGN_CUST → create an entry for accessing the OData API. Or use the standard method: SEGW → activate the service ZAPI_PO_CREATE.
2. Go to the chat with ASI Biont at asibiont.com.
3. Write: "Connect me to SAP, here is my API key: [your key]. I want to automate the creation of purchase requisitions."
4. Explain the task — the AI will ask clarifying questions (what materials, what threshold, where to send notifications).
5. Done — the AI will generate the integration code and launch the agent. You will immediately get the result.

Important: you are not limited to standard scenarios. Do you want the AI to check invoices for duplicates? Or automatically create replenishment orders? Just describe it in the chat.

Conclusion: SAP Becomes "Smart" Without Capital Expenditures

Integrating the ASI Biont AI agent with SAP via API is not futuristic; it is an accessible reality. You do not need to buy expensive RPA, hire SAP consultants, or wait for updates from developers. All you need is an API key and 10 minutes to describe the task in the chat.

Try it yourself: go to asibiont.com and write in the chat: "Connect SAP." The AI agent will do the rest.

Article written on July 4, 2026, based on open data from SAP SE, SAPinsider reports (2025), and internal case studies of ASI Biont users.