Data and Tools

The AI Assistant does not have unrestricted access to your FieldOps database.

Instead, it uses a controlled set of FieldOps data tools to retrieve the information needed to answer your questions.

This approach allows the assistant to work with operational data while keeping access limited to the information and capabilities made available to it.

How the Assistant Accesses Data

When you ask a question, the assistant determines what information is needed and uses the appropriate FieldOps tools to retrieve it.

A typical request may involve several steps:

1. You ask a question in natural language.
2. The assistant determines what information is needed.
3. The assistant selects the relevant FieldOps data tools.
4. The tools retrieve information from your organization's FieldOps data.
5. The assistant analyzes the retrieved information.
6. The assistant provides the response.

For example, a question such as:

Why has Kapsabet Sub-County Hospital been underperforming for the last three months?

may require the assistant to retrieve information about:

• The site's performance
• Recent visits
• Findings
• Corrective actions
• Historical performance information

The assistant can combine the retrieved information to provide a more useful explanation.

Controlled Access to FieldOps Data

The AI Assistant does not receive unrestricted access to the FieldOps database.

It can only access information through the data tools made available to it.

This means:

• The assistant cannot freely query the database.
• The assistant cannot access arbitrary database tables.
• The assistant cannot bypass FieldOps access controls.
• The assistant only receives information returned by the tools it uses.
• The available information depends on the data and integrations configured in your organization.

This controlled approach helps keep the AI Assistant focused on the operational information it needs.

Available Data Areas

Depending on your FieldOps configuration, the AI Assistant can work with information from areas such as:

Projects

The assistant can retrieve information about projects, including project status, project summaries, and project-related sites.

Examples:

How many projects do we have?

Give me a summary of this project.

Which sites belong to this project?

Sites

The assistant can retrieve information about sites and their operational status.

Examples:

How many sites do we have?

Which sites require attention?

What is happening at Kapsabet Sub-County Hospital?

Visits

The assistant can retrieve information about monitoring visits and their results.

Examples:

What happened during the latest visit?

Show me the visit history for this site.

What findings were recorded during the visit?

Findings

The assistant can retrieve and analyze findings recorded during monitoring.

Examples:

Which findings are unresolved?

What are the main findings for this project?

Which issues have remained unresolved?

Corrective Actions

The assistant can retrieve information about corrective actions and their status.

Examples:

Which corrective actions are overdue?

Which sites have outstanding actions?

What actions are associated with this finding?

Performance

Where performance information is available, the assistant can analyze operational performance and identify areas requiring attention.

Examples:

Which sites are below target?

Which site is performing worst?

Why has this project been underperforming?

DHIS2 Performance

When DHIS2 is configured and the relevant data is available, the assistant can retrieve and analyze DHIS2 performance information.

Examples:

Which indicators are performing poorly?

Which sites are below target?

Which sites need performance follow-up?

The availability of DHIS2 information depends on the DHIS2 connection, synchronized metadata, site mappings, project indicators, and available performance data.

Combining Multiple Data Sources

One of the most useful capabilities of the AI Assistant is its ability to combine information from different areas of FieldOps.

For example:

Why has Kapsabet Sub-County Hospital been underperforming for the last three months?

The assistant may need to consider performance data together with recent visits, findings, and corrective actions.

Another example:

Which sites are underperforming and also have overdue corrective actions?

This requires information about both site performance and corrective actions.

The assistant can combine the relevant information and present it as a single response.

FieldOps Tools

FieldOps provides the AI Assistant with purpose-built tools for retrieving operational information.

These tools can cover areas such as:

• Projects
• Sites
• Project summaries
• Project sites
• Sites requiring attention
• Visits
• Visit details
• Visit history
• Findings
• Corrective actions
• Overdue actions
• Site operational summaries
• DHIS2 performance
• Performance follow-up

The exact tools available to the assistant may change as FieldOps evolves.

Tool Selection

You do not need to know which tool the assistant should use.

For example, you can simply ask:

Which sites require attention?

The assistant determines which FieldOps information it needs and selects the appropriate tools.

You do not need to provide database fields, table names, IDs, or technical commands.

Data Context

The assistant operates within your active FieldOps organization context.

This means that the information retrieved by the assistant is based on the data available to your organization and the access provided through the FieldOps application.

The response can therefore depend on:

• Your organization's projects
• Your organization's sites
• Your organization's visits
• Findings and corrective actions
• Configured integrations
• Available performance data
• The tools available to the assistant

External Integration Data

The AI Assistant can also work with information originating from configured integrations.

For example:

• KoboToolbox data can provide visit information.
• ODK Central data can provide visit information.
• DHIS2 can provide performance and indicator information.

The assistant does not automatically retrieve information directly from every external system.

The relevant integration must be configured and the required data must be available in FieldOps.

What the Tools Do Not Allow

The current AI Assistant is designed for read-only analysis and information retrieval.

Its tools do not currently allow it to:

• Create records
• Edit records
• Delete records
• Approve visits
• Reject visits
• Create corrective actions
• Change project configuration
• Modify monitoring templates
• Perform operational workflows

The assistant can identify an issue and explain it, but any operational change must still be performed through the normal FieldOps workflows.

For example, the assistant may tell you:

Kapsabet Sub-County Hospital has two overdue corrective actions related to vaccine stock management.

It cannot currently create or modify those corrective actions on your behalf.

Why Controlled Tools Matter

Using purpose-built tools provides an important boundary between the AI model and your operational data.

The AI model is responsible for:

Understanding → reasoning → explaining

FieldOps is responsible for:

Retrieving → controlling → providing the relevant operational data

This separation allows FieldOps to give the AI Assistant access to useful operational information without giving the model unrestricted access to the application database.

Data Availability Affects Answers

The assistant can only analyze information that is available through FieldOps and its configured integrations.

For example, if historical performance data is missing, the assistant may not be able to reliably answer:

How has this site's performance changed over the last six months?

Similarly, if visits have not been synchronized from an external integration, the assistant cannot use those visits in its analysis.

When an answer depends on missing information, check the relevant FieldOps records or integration configuration.

Source of Truth

The underlying FieldOps records remain the source of truth.

The AI Assistant is an analytical layer that retrieves and interprets available information.

For important operational decisions:

1. Review the AI response.
2. Check the relevant FieldOps records.
3. Verify important figures and conclusions.
4. Use the normal FieldOps workflows when action is required.

A useful way to think about the relationship is:

FieldOps data → controlled tools → AI analysis → natural-language answer