What are tokens in ERPure.AI?
Tokens represent the total amount of work the AI performs during a single interaction including both usage of LLMs and Tools.
This includes:
- Understanding your question
- Retrieving relevant data
- Reasoning through the request
- Generating a response
Both the input (your request) and output (the response) are included.
Do tokens affect what I can or can’t do?
No.
Tokens do not limit functionality, and customers are not expected to manage or optimise tokens.
You simply ask questions or run tasks — ERPure.AI handles the complexity behind the scenes.
What are “tools” in ERPure.AI?
Tools are the specialised skills ERPure.AI uses to do real work, such as:
- Retrieving SAP Business One data
- Validating business rules
- Processing documents
- Executing approved transactions
The AI automatically decides which tools are needed based on your intent.
Why does ERPure.AI sometimes use multiple tools in one request?
Some requests are simple.
Others are more complex and require multiple steps.
For example:
- Checking data
- Applying rules
- Validating results
- Preparing a response
Each step may involve a different tool working together to produce a correct result.
Why do tools use tokens as well?
Although tools return structured data, the AI still needs to:
- Interpret the results
- Combine them with other information
- Explain the outcome clearly
That reasoning and explanation process uses tokens.
What does a high token count actually mean?
A higher token count usually means:
- Your request was more complex
- More data needed to be retrieved
- More reasoning was required
- Multiple tools were involved
For example:
A request using 50,000–60,000 tokens and several tool calls indicates a complex, multi-step task, not inefficiency.
Is a high token count bad?
No.
In fact, it often means ERPure.AI is doing more valuable work for you.
Complex tasks naturally require:
- More validation
- More checks
- More reasoning
That’s expected in ERP environments.
Can customers control token usage directly?
Not directly — and that’s intentional.
Token usage is a technical detail, not a business control.
Instead, customers control:
- Which users can access the system
- Which workflows are enabled
- Which actions require approval
This keeps the focus on outcomes, not mechanics.
Why does ERPure.AI show token & tool usage at all?
Token & Tool usage is shown to provide:
- Transparency
- Insight into task complexity
- An indication of how much work the system performed
It is an informational metric, not a tuning requirement.
How should I interpret this as a customer?
Think of tokens as a “complexity indicator”, not a cost meter.
Low tokens = quick, simple task
High tokens = deeper analysis or multi-step processing
Either way, ERPure.AI is working to deliver the best possible result.
“Tokens and Tools show how hard the AI worked — not something you need to manage.”
How much is it going to cost for Token usage
Ultimately this will depend on the model and the usage. So while you are in Pilot mode we are picking up the cost of your token usage.
When you decide to move off Pilot into Production mode we will be able to estimate your utilisation based on your patterns established during the Pilot.