What is automated metadata enhancement for ArcGIS?

Automated metadata enhancement uses FME workflows combined with AI to generate meaningful descriptions, tags, field aliases, column names and field-level documentation for feature services published in ArcGIS Enterprise, ArcGIS Portal or ArcGIS Online. It removes the need for manual metadata entry by analysing the actual data content and producing contextually appropriate metadata at scale.

Why is metadata important for ArcGIS feature services?

Metadata makes your feature services searchable, understandable and properly governed. Without it, datasets are difficult to find, their purpose is unclear and your organisation risks duplicating effort or making decisions based on misunderstood data. Good metadata also supports regulatory compliance and audit requirements.

How does AI improve ArcGIS metadata quality?

The workflow samples your feature data and generates statistical summaries for each field, including value distributions, ranges and common patterns. These summaries are passed to an AI model which produces descriptions, tags and field documentation that reflect the actual content of the dataset rather than relying on generic or template-based text.

Do we need to use a cloud AI like OpenAI or Gemini?

Not necessarily. While we can integrate with Google Gemini or OpenAI, the workflow also connects to a local LLM deployment. If your organisation has strict data security or data residency requirements, your data never needs to leave your network.

Can we see the changes before they are applied to our live services?

Yes. The solution includes a preview mode where improved metadata is written to a staging output for your team to review before updating production feature services. Nothing is changed in your live environment until you approve the results. Human in the loop review is an important step in the process to ensure we have validation and sense checking of the created or updated metadata before any changes are implemented.

Will this work with our existing metadata?

The workflow adapts to your current state. Services with complete metadata can be skipped, while those with partial or missing metadata are improved. You control the scope through configurable filters including folder, tag and service name patterns.

How long does processing take?

For 100 to 500 feature services, processing typically takes between 30 and 120 minutes depending on sampling depth and AI model speed. The workflow includes checkpointing, so if it is interrupted, it can resume from the last successfully processed service.

Can we run the workflow ourselves after the engagement?

Yes. We hand over the FME workspaces with full documentation and provide mentoring so your team can run, modify and extend the workflow independently. This is not a black box delivery.

What version of FME is required?

The workflow requires FME Form 2024 or above. If you want to schedule automated runs or provide metadata enhancement as an internal service, FME Flow is recommended but not required for the core workflow.

Is our data sent to external services?

Only the statistical summaries and field schema information are sent to the AI provider, not your raw feature data. If this is still a concern, the workflow supports local LLM deployments where nothing leaves your infrastructure.

How does this support data governance and compliance?

Consistent, well-structured metadata creates a clear record of what each dataset contains, who it is for and how it should be used. This supports data cataloguing, audit trails and regulatory requirements. The full audit log produced by each run provides a before-and-after record of every change.

Can this be used across multiple ArcGIS Portal or ArcGIS Online environments?

Yes. The workflow is configured through published parameters, so it can be pointed at different portal URLs or ArcGIS Online organisations. Separate runs can be configured with different AI prompts, scope filters and metadata rules to suit each environment.

Quality metadata is a challenge

We get it - keeping ArcGIS Enterprise or ArcGIS Online metadata up to date and to standard is a challenge. Many organisations struggle with poor metadata quality across their feature services. Datasets lack meaningful descriptions, tags are inconsistent or missing entirely, field names use cryptic abbreviations without aliases, and column descriptions are absent altogether.

Manual metadata creation is time-consuming, inconsistent and often neglected. This impacts data search and discovery, the understanding of a dataset's purpose and content and limits data governance.

The result is data assets that are technically accessible but practically unusable, unless you already know the data intimately, or can track down whoever created it.

Metadata Enhancement for ArcGIS Enterprise/Online powered by FME and AI

Our technical solution offers a short engagement to prepare an FME workflow to provide an automated, FME + AI driven solution that systematically discovers and enhances metadata across ArcGIS Enterprise or ArcGIS Online published feature services.

Key benefits

Find your data faster

Properly tagged and described feature services are easy to search and discover. Your team spends less time hunting for datasets and more time putting them to work.

Understand data at a glance

AI augmented descriptions, column aliases and field-level documentation give every dataset clear context. New starters and wider business users can understand what a dataset contains without specialist knowledge or chasing the original creator.

Strengthen your data governance

Consistent, well-structured metadata supports audit trails, regulatory compliance and data stewardship. It brings order to environments where naming conventions have drifted over time.

Reduce repetitive manual effort

Rather than asking GIS analysts to write metadata by hand for hundreds of services, the workflow handles this in a single automated run. Your specialists are freed up for higher-value work.

Dynamic and schema-agnostic

This solution will adapt to whatever fields and data structure it encounters. All data is pulled dynamically from the ArcGIS environment via REST API calls.

Any AI support

Any Generative AI tooling with an API is supported - this includes OpenAI, Gemini, Claude alongside locally running open source models such as Qwen.

Human-In-The-Loop

Proposed metadata changes are generated for review and tweaking before any metadata is updated.

What going on under the bonnet

Feature Service Discovery

All accessible Feature Services are iterated through, for each service existing metadata is extracted and samples of the data are taken.

Statistics and context

For each Feature Service comprehensive context is compiled and sent to the AI of choice for metadata generation.

Review and Update

Enhanced metadata descriptions, relevant tags, user-friendly field aliases and detailed column descriptions are populated into an Excel report for human-in-the loop-review before updates are run against the Feature Service.

Engagement

Avineon Tensing deliver this as a short engagement, depending on your setup this typically ranges from 5-8 days. We configure the workflow for your environment, run and validate it against your feature services and hand over the FME workspaces alongside full documentation.

This is not a black box. We will have updated metadata across your feature services: descriptions, tags, aliases and field documentation. Provided a set of FME workspaces that you own and build upon internally going forward and audit capabilities to track changes

We will spend time with your team, explain how the workspaces work and step through the process. When the engagement ends, your team have the knowledge and confidence to run, modify and extend the workflow independently.

Get in touch