A vendor onboarding workflow: review the vendor, check the details, then approve to trade
SynaptaGrid8 min read
In short
A vendor onboarding workflow is the path a new supplier takes from the first request to a recorded decision: procurement checks that the business needs the vendor, compliance checks the tax registration, bank details and sanctions screening, and only then is the vendor marked as approved to trade. Most vendor problems come from skipped checks and unclear ownership. A vendor gets paid before anyone verified the bank details, or a rejected vendor is never marked as rejected and comes back through another door. A good workflow keeps the business question and the compliance question apart, gives each a different owner, lets either stop the vendor, and writes the status onto a record so the vendor list is always current. We built a small version in our own test workspace in which an AI tool, written for this job and used as an action in a normal step, reads the vendor’s submitted details, names what the vendor supplies and lists what is missing, and the workflow writes both onto the vendor record before procurement reviews it. The tool reads text, not attached documents, because reading an attached file is not available yet. We ran it three times, once approved, once rejected by procurement and once stopped by the compliance checks, and checked the vendor records afterwards.
What a vendor onboarding workflow is
Before a company pays a new supplier it wants to know that it needs them, that they are who they say they are, and that the bank account belongs to them. A vendor onboarding workflow writes down who checks each of these, in what order, and what happens when a check fails.
The workflow does not decide who is a good supplier. People do. It makes sure both checks happen, in the right order, that a vendor who fails one has a clear ending, and that every vendor ends in a status anyone can look up before paying them.
The version we built
The steps to include, and who owns each
- Record the vendor
- When a request arrives, a record is created with the vendor name, a status of onboarding started and a flag that the vendor is not yet approved to trade. Every later step reads from and writes to that record.
- Read the submitted details
- This is where an AI tool helps. It reads what the vendor sent, names what the vendor supplies, such as goods or services, and lists what is missing, such as an insurance certificate or bank details. The next step writes both onto the vendor record, so procurement sees the gaps with the request. The tool does not approve or reject anyone, and it reads text only: an attached certificate is not read.
- Procurement review
- Procurement checks that the business needs this vendor and that the details given are complete. Procurement knows the need and the alternatives, which compliance cannot see.
- Compliance checks
- Compliance verifies the tax registration, the bank details and the sanctions screening. This is a different question from procurement’s, so it is a different step with a different owner.
- Two places to stop
- A vendor can be stopped by procurement, who decide not to use them, or by compliance, whose checks fail. The two endings are recorded differently, so anyone reading the record can tell which it was.
- Mark the result on the record
- Each ending sets the status and the approved flag: approved to trade, rejected by procurement, or failed compliance checks. The workflow does this itself, so the vendor list does not depend on someone remembering to update it.
- Hand off to the payment system
- Setting the vendor up for payment usually happens in another system. Decide where an approved vendor goes next and who is told, and add that as its own step.
Where approvals and checks belong
Put a check where a mistake is expensive. Paying a vendor with fraudulent bank details is hard to undo, so the compliance checks come before the vendor is approved to trade, and nobody can mark a vendor approved without passing them. Procurement goes first because there is no point verifying a vendor the business will not use.
Name each approver by role, such as procurement and compliance, rather than by person. If you need a further sign-off for a high-risk country or a large contract, add it on your own policy; we did not build that.
The old way and the new way
The table compares vendor onboarding handled by email and a spreadsheet with the same requests run as a workflow.
- First read of a new vendor
- The old way: Procurement reads the email cold
- The new way: An AI tool names the category and lists what is missing first
- Where a request is
- The old way: In an inbox or a shared folder
- The new way: At a named step, with an owner
- Whether the checks happened
- The old way: Nobody can tell
- The new way: A separate compliance step that is recorded
- A vendor who fails a check
- The old way: Dropped quietly, or paid anyway
- The new way: Marked failed checks on the record
- Whether a vendor can be paid
- The old way: Ask whoever set them up
- The new way: The record says approved to trade or it does not
- Finding a vendor later
- The old way: Search inboxes and spreadsheets
- The new way: Open the record and read its status
- Judgement
- The old way: Still procurement’s and compliance’s
- The new way: Still theirs: whether to use the vendor, whether the checks pass
How we decided
We built the version in the diagram in our own test workspace in SynaptaGrid Automation, with a small record type in SynaptaGrid Data for the vendor, and ran it three times in November. The AI part is a custom tool we wrote for the job: a prompt that says how to review a vendor, an input that takes the submitted details as text, and an output with a category from a fixed list, a list of what is missing and a short summary. In the workflow it is an action in an ordinary step, set up like any action that calls another system, and the next step writes its result onto the record. In the first run procurement approved and compliance passed the checks. The workflow created the vendor record with the status Submitted and the approved flag off, then wrote the AI result onto it, then updated the same record to Approved with the flag on, and the run finished with the approved result. In the second run procurement rejected the vendor: the run skipped compliance, marked the record Rejected and ended with a rejected result. In the third run procurement approved and compliance failed the checks, so the record ended as Failed checks.
We checked the three records in Data afterwards. The approved one showed Approved with the flag on. The rejected one showed Rejected, and the one that failed the checks showed Failed checks, both with the flag off. All three carried the AI result: the category goods, the missing items insurance certificate, bank details and named contact, and a summary of two sentences restating what the vendor had sent. The three notes differed only in wording. The tool listed a named contact as missing although the text gave a sales desk as the contact, which is a judgement call and shows why procurement reads the list against what was sent and does not treat it as fact. The AI tool did not work first time: our earlier attempts to call it failed until we named the model in full in the tool’s settings, and our first version did not use its output in a later step. The version described here is the one whose three outcomes we checked.
We chose two separate checkers, two distinct endings and a record that carries the status because those are the places vendor onboarding usually goes wrong. The choice of steps is ours and a judgement; it is not a standard.
Six limits, so this guide does not claim more than the runs showed. The tool saw one vendor description, so we did not test other kinds of vendor, a long description or one that contradicts itself, and we did not compare its list of gaps with a person’s. The tool read text only: an insurance certificate or a bank letter attached to the request is not read. One test account held every human step, so the runs show the sequence and the branching, not hand-offs between different people. The compliance step was a person’s decision: the workflow does not look up a tax registration, verify a bank account or screen against a sanctions list, and we did not connect it to any such service or to a payment system, so the approved flag is a stand-in and nothing was paid. We did not build an extra sign-off for high-risk vendors, a check for a vendor already on the list, or a way to bring a rejected vendor back for a second review. We did not test reminders when a step waits too long or notifications to the vendor. And the template carries one test account as every approver and the actions tied to our test record type, so to reuse it you would change each step to a role and point the actions and the tool at your own, which we did not do.
Standards and accepted practice behind this
These are the practices the guide rests on. We align with them. This page does not claim a certification.
- Human oversight of AI
- Where AI helps with a decision, accepted practice is to keep a person accountable for it and to let them see the AI’s input and output. Here the tool lists what is missing and procurement and compliance decide.
- Source: EU AI Act (Regulation 2024/1689), Article 14: human oversight (opens in a new tab)
- Segregation of duties
- The person who asks for a vendor, the person who verifies them and the person who pays them should not all be the same. Separate steps with separate owners support that.
- Source: NIST glossary: separation of duty (opens in a new tab)
- Know your supplier
- Many companies verify who a supplier is before trading, including tax registration and bank ownership. The compliance step is where that verification is recorded.
- Sanctions screening
- Companies are generally expected to avoid trading with sanctioned parties, and the rules depend on your country and sector. Check yours before you decide what the compliance step must cover.
- Documented approval
- Auditors expect to see who approved a vendor and when. A step that records the decision and a record that carries the status are that evidence.
- BPMN
- The standard notation for drawing a process. The diagram above is a plain flowchart, and a BPMN diagram of the same steps would show the same decisions.
- Source: OMG: BPMN 2.0 specification (opens in a new tab)
Where SynaptaGrid fits
SynaptaGrid Automation lets you write the steps as a workflow, assign each decision to a person, use an AI tool you wrote as an action in a step, pass its result to the next step, create and update records in SynaptaGrid Data as part of the run, and see where every vendor stands. You can start from the version above, change the steps to match your own policy, and keep the result as a template.
Check the limits above before you rely on it: whether each step is assigned to a role rather than a person, whether the compliance step needs to reach a verification or screening service, whether approved vendors must reach a payment system, and who should be told when a vendor waits.
Frequently asked questions
- What are the steps of a vendor onboarding process?
- Record the vendor, a procurement review, compliance checks of tax registration, bank details and sanctions, a way to stop the vendor at either step, and a status written on the record.
- Why separate procurement from compliance?
- They answer different questions. Procurement decides whether the business needs the vendor, and compliance decides whether the vendor is who they say they are. Each needs different information.
- What should happen to a vendor who fails the checks?
- The record should be marked as failed checks, not left open, so the vendor cannot be set up for payment by mistake and the reason is visible later.
- Can AI review a new vendor?
- It can read the details the vendor sent as text, name the category and list what is missing. In our test run it listed an insurance certificate, bank details and a named contact and summarized the request in two sentences. It cannot read attached documents yet, and procurement and compliance still decide.
- Can a workflow update the vendor record for me?
- Yes. In our test run the workflow created the vendor record when the request arrived and set its status and approved flag at each ending. You need a record type for the vendor and each action pointed at it.
SynaptaGrid
Build your vendor onboarding steps
Start with the steps above in a test workspace and run one vendor through each outcome before you use it for real.