Classic Document Intelligence deprecation and steps to plan your move to Now Assist in Document Intelligence (NADI) - Support and Troubleshooting

Classic Document Intelligence deprecation and steps to plan your move to Now Assist in Document Intelligence (NADI) Summary .ns-kb-css-body-editor-container { p { font-size: 12pt; font-family: Lato; color: var(--now-color--text-primary, #000000); } span { font-size: 12pt; font-family: Lato; color: var(--now-color--text-primary, #000000); } h2 { font-size: 24pt; font-family: Lato; color: var(--now-color--text-primary, black); } h3 { font-size: 18pt; font-family: Lato; color: var(--now-color--text-primary, black); } h4 { font-size: 14pt; font-family: Lato; color: var(--now-color--text-primary, black); } a { font-size: 12pt; font-family: Lato; color: var(--now-color--link-primary, #00718F); } a:hover { font-size: 12pt; color: var(--now-color--link-primary, #024F69); } a:target { font-size: 12pt; color: var(--now-color--link-primary, #032D42); } a:visited { font-size: 12pt; color: var(--now-color--link-primary, #00718f); } ul { font-size: 12pt; font-family: Lato; } li { font-size: 12pt; font-family: Lato; } img { display: ; max-width: ; width: ; height: ; } } PRODUCT ANNOUNCEMENT

Document Intelligence is moving to Now Assist

Plan your move from Document Intelligence (Classic) to Now Assist in Document Intelligence (NADI).

AUDIENCE

Customers and ServiceNow admins

APPLIES TO

Zurich release and later

PRODUCTS

Document Intelligence & Now Assist Document Intelligence

ACTION REQUIRED

Yes. Set up new use cases in NADI

SUMMARY

The classic ML-based Document Intelligence has been deprecated and replaced by Now Assist in Document Intelligence (NADI), the GenAI-powered evolution of Document Intelligence. NADI fully supports document extraction, processing, and multimodal capabilities through GenAI and third-party frontier models, with no model training required. For more information, see Now Assist in Document Intelligence .

Bottom line: Existing Classic use cases cannot be migrated automatically. You will rebuild them in NADI and gain zero-shot extraction, Q&A, and summarization in the process.

Why NADI, and why now

Document Intelligence (Classic) was built on optical character recognition plus traditional machine learning. It works well on familiar, structured forms, but every new document layout meant new labels, new training, and another round of model tuning.

Now Assist in Document Intelligence takes a different approach. It uses a large language model to read documents with zero-shot, guided by the key field extraction descriptions you write in plain language. There are no templates to maintain, no annotations to gather, and the same setup can answer questions and summarize content alongside extracting fields.

The migration to NADI will remove the maintenance burden and unlocks document types that the Classic Document Intelligence could not handle reliably.

How the two compare

Dimension

Document Intelligence (Classic)

Now Assist in Document Intelligence (NADI)

Technology

OCR + traditional machine learning

Zero-shot extraction powered by an LLM

Setup effort

Label training data, then iterate on the model

No training, no templates. Describe the key information you want to extract

Capabilities

Extraction only

Key information extraction, Question & Answer, and summarization

Document fit

Structured and semi-structured forms

Structured, semi-structured, and unstructured documents

Ongoing maintenance

Continuous model retraining as document changes

Refine descriptions when behavior shifts

Required platform setup

OCR engine and model training pipeline

Gen AI Controller configured with an LLM provider

What changes for you

Annotation and training cycles go away. You no longer label sample documents and wait for a model to converge. Task definitions still exist, but the keys behind them no longer require custom trained models. Key descriptions become the most important thing you write. The LLM uses them to locate the right data, so clarity and specificity in the description directly drive accuracy. You need the Generative AI Controller installed with an LLM provider set-up before NADI can do its work. What stays the same

The core tables are unchanged: sys_di_task_definition , sys_di_task , sys_di_key , sys_di_key_group , and sys_di_extracted_value all continue to exist. The workspace experience is similar. Reviewers still open extracted documents and correct values where needed. Integration patterns with surrounding tables and workflows do not fundamentally change. Can I migrate my existing use cases?

Short answer: No, not automatically. Although Classic and NADI share many of the same tables, NADI relies on key descriptions to guide the LLM. Classic learned from labeled training data and never required these descriptions, so the metadata simply does not exist in the Classic keys.

You will rebuild each use case in NADI. The upside is that the build is dramatically faster: no labelling, no training runs, and the same use case can now support Q&A and summarization alongside extraction.

New use cases created in NADI can be captured and promoted between instances using Update Sets, the same way you move other configurations today.

Recommended next steps

Inventory your active Document Intelligence (Classic) use cases and the document types each one handles. Confirm that Generative AI Controller is installed and configured on the target instance and connected to an approved LLM provider using the Now Assist Admin console. Activate Now Assist in Document Intelligence and recreate one use case end-to-end as a pilot. Refine the key descriptions until extraction quality meets your bar, then expand to the remaining use cases. Plan the cutover with your business stakeholders. Classic continues to run during the transition, so there is no forced switch date. If you need help planning your migration, please contact your assigned ServiceNow representative.

Related Links .ns-kb-css-body-editor-container { p { font-size: 12pt; font-family: Lato; color: var(--now-color--text-primary, #000000); } span { font-size: 12pt; font-family: Lato; color: var(--now-color--text-primary, #000000); } h2 { font-size: 24pt; font-family: Lato; color: var(--now-color--text-primary, black); } h3 { font-size: 18pt; font-family: Lato; color: var(--now-color--text-primary, black); } h4 { font-size: 14pt; font-family: Lato; color: var(--now-color--text-primary, black); } a { font-size: 12pt; font-family: Lato; color: var(--now-color--link-primary, #00718F); } a:hover { font-size: 12pt; color: var(--now-color--link-primary, #024F69); } a:target { font-size: 12pt; color: var(--now-color--link-primary, #032D42); } a:visited { font-size: 12pt; color: var(--now-color--link-primary, #00718f); } ul { font-size: 12pt; font-family: Lato; } li { font-size: 12pt; font-family: Lato; } img { display: ; max-width: ; width: ; height: ; } } • Classic Document Intelligence product documentation (deprecated)

• Now Assist in Document Intelligence - Setup and configuration guide

• Generative AI Controller / Manage Model Providers (Required prerequisite for NADI)