Limitations

The AI Assistant is designed to help you understand and analyze FieldOps data. It is not a replacement for the underlying FieldOps records, operational workflows, or human decision-making.

AI-Generated Responses

AI responses are generated from the information available to the assistant and may occasionally be incomplete, inaccurate, or incorrectly interpreted.

The assistant may:

• Misinterpret a question
• Miss relevant information
• Provide an incomplete explanation
• Make an incorrect inference
• Present information with more confidence than the available data warrants

For important operational decisions, verify the response against the relevant FieldOps records.

Data Availability

The quality of an answer depends on the quality and completeness of the underlying data.

The assistant cannot reliably answer questions when:

• Required records have not been entered
• Visits have not been completed or synchronized
• Findings or corrective actions are missing
• Historical data is incomplete
• Integration data has not been synchronized
• Project indicators have not been configured
• The requested information is outside the data available to the assistant

For example, if only one month of performance data is available, the assistant cannot reliably explain a three-month performance trend.

Available Data and Tools

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

It uses controlled data tools to retrieve information relevant to a question.

This means that some questions may not be answerable even when related information exists somewhere in the FieldOps application.

The information available to the assistant depends on:

• The data available in your organization
• Configured integrations
• Synchronized data
• Configured project indicators
• The tools available to the AI Assistant

See Data and Tools for more information about the information available to the assistant.

No Autonomous Actions

The current AI Assistant is focused on reading, analyzing, explaining, and summarizing information.

It does not independently:

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

Use the normal FieldOps workflows when you need to make changes.

For example, the assistant may identify an overdue corrective action, but it cannot currently create, update, or close that action on your behalf.

Integration Limitations

Information from external systems depends on the relevant integration being configured and synchronized correctly.

For example, DHIS2-related questions depend on:

• A working DHIS2 connection
• Synchronized DHIS2 metadata
• Correct site mappings
• Configured project indicators
• Available performance data

Similar limitations apply to data received through KoboToolbox, ODK Central, and other integrations.

If information has not yet been synchronized into FieldOps, the AI Assistant may not be able to use it.

Historical and Trend Analysis

Trend questions require sufficient historical data.

For example:

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

The assistant needs enough historical performance information to make a meaningful comparison.

If data is missing, inconsistent, or covers only a short period, the resulting analysis may be limited.

The assistant should not be expected to infer historical trends from insufficient data.

Ambiguous Questions

The assistant works best when questions clearly identify the information or analysis you need.

For example:

Which sites have overdue corrective actions?

is more specific than:

What is happening?

Natural-language questions are supported, but ambiguous questions may produce a less useful response.

You can ask a follow-up question to clarify or narrow the analysis.

Model Differences

Different AI providers and models can produce different responses to the same question.

One model may provide a short summary while another may provide a more detailed explanation.

Models may also differ in:

• Reasoning ability
• Response quality
• Response speed
• Handling of complex questions
• Interpretation of ambiguous information
• Ability to work with larger amounts of context

The model guidance displayed in FieldOps can help you choose a model based on the type of AI work you expect to perform.

Even when two models receive the same FieldOps data, their responses may not be identical.

The underlying FieldOps data remains the source of truth.

Provider Availability

The AI Assistant depends on a configured and functioning AI provider connection.

Responses may be unavailable when:

• No AI provider connection is configured
• The connection is inactive
• Provider credentials are invalid
• Provider credits or quotas have been exhausted
• A provider imposes a rate limit
• A selected model is unavailable
• A provider experiences an outage
• The provider connection encounters an error

These issues may require attention to the FieldOps AI connection or the provider's own account.

AI Usage and Billing

FieldOps records AI usage information so organizations can monitor how their AI connections are being used.

However, FieldOps usage information is not the same as provider billing information.

Your organization pays the selected AI provider directly, and the provider remains the authoritative source for charges, credits, and billing.

For billing-related questions, always refer to the AI provider's own account and billing records.

AI Does Not Replace Operational Judgment

The AI Assistant can help identify patterns, summarize information, explain findings, and support investigation.

It should not be treated as the final authority for important operational decisions.

For example, if the assistant reports that a site is underperforming, review the underlying performance data, visits, findings, and corrective actions before taking action.

A useful workflow is:

Ask → investigate → verify → decide → act

Getting Better Results

You can often improve an answer by making your question more specific.

Instead of:

Tell me about this site.

Try:

Why has this site been underperforming for the last three months?

You can then continue with follow-up questions such as:

What findings contributed to the problem?

Which corrective actions are still overdue?

What happened during the latest visit?

You can also specify:

• A project
• A site
• A time period
• A type of record
• A performance measure
• A particular issue

This gives the assistant a clearer analytical objective.

Verify Important Information

For important operational decisions, verify AI-generated information against the underlying FieldOps records.

This is particularly important when the response includes:

• Performance figures
• Trends or comparisons
• Explanations of underperformance
• Findings
• Corrective action status
• Information used to make operational decisions

The AI Assistant is an analytical tool. The underlying FieldOps records remain the source of truth.

Summary

The AI Assistant is a tool for operational intelligence and analysis, not an autonomous decision-maker.

Its usefulness depends on:

• The quality of your FieldOps data
• The availability of relevant historical information
• Configured integrations
• Available AI tools
• The selected AI model
• The reliability of the configured AI provider
• The clarity of your question

For important decisions, always verify AI-generated insights against the underlying FieldOps records.