Weissr Chatbot: Interacting with Capex Data Using Natural Language

The Weissr chatbot allows you to interact with Capex data using natural language.

You can ask questions, filter requests, and perform actions without using filters or navigating through menus.

This makes it faster and easier to find information and work with requests.


What you can do

You can use the chatbot to:

  • Find requests using natural language

  • Filter requests based on properties

  • Retrieve request property values

  • Create new requests (if enabled)

  • Navigate directly to requests from chat results


How it works

The chatbot understands your input and translates it into actions in Weissr.

Behind the scenes:

  • Your request is analyzed and matched to an intent

  • The system routes it to the correct function (e.g. search, data retrieval, action)

  • Results are returned directly in the chat

If your request is unclear or incomplete, the chatbot will ask follow-up questions.


Example requests

You can type requests like:

  • “Show me capex requests for Site A”

  • “Show approved IT projects”

  • “Create a new capex request for Site A”

  • “Show requests with expense type opex”

  • “Show currency and category for these requests”

You do not need to use exact system terms - the chatbot understands variations and synonyms.


Follow-up and confirmation

The chatbot may ask follow-up questions when:

  • Required information is missing

  • A value is unclear or incorrect

  • Multiple interpretations are possible

Example:

“Show me capex requests”
→ “Which site or category do you want to filter by?”


Confirmation before actions

For actions such as creating a request:

  • The chatbot summarizes the action

  • Asks for confirmation

  • Waits for your approval before proceeding

Example:

“You’re creating a request for Site A for IT, do you want to proceed?”

This prevents accidental changes.


Results and navigation

  • Results are displayed directly in the chat

  • Request names are clickable links

You can open requests directly from the chatbot.


Working with request data

You can ask the chatbot to return property values.

Example:

  • “Show me requests where Category is IT and return Site and Currency”

Results are displayed as a table.


Limits

  • Up to 25 requests are returned by default

  • You can extend results up to 100 requests

If too many results are found, the chatbot will ask you to refine your search.


Property limits

  • Maximum of 10 properties can be requested at once

If too many properties are requested, the chatbot will suggest reducing them.


Filtering behavior

  • Filtering works on request properties

  • Also supports properties from related expenditures

You can create complex queries using natural language.


Permissions and data access

The chatbot respects user permissions.

  • Only accessible requests are returned

  • Restricted properties are not shown

If you request data you do not have access to, it will not be returned.


Understanding requests vs expenditures

The chatbot distinguishes between:

  • Capex requests

  • Expenditures (expenses, outlays)

Expenditures are only included when explicitly requested.


When the chatbot cannot understand

If the chatbot cannot determine your request:

  • It will ask follow-up questions

  • Suggest rephrasing

If the intent remains unclear after multiple attempts, you may need to refine your request.


Tool availability

Available chatbot capabilities may vary.

  • Some features can be enabled or disabled per client

  • Not all actions may be available in your environment

Contact your administrator if a feature is not available.


Current limitations

This is the first version of the chatbot.

  • Not all actions are supported

  • Complex queries may require refinement

  • Some features depend on system configuration

The chatbot will continue to improve in future releases.