Fourth Paradigm 4Paradigm

What is the fourth normal form?

Fourth Normal Form is a technology company that provides enterprises with artificial intelligence platforms, industry-specific solutions, and computing power products.

Its products cover AI for prediction, industry-specific large models, generative software development, model management, and production deployment.

Core products

Prophet AIOS 5.0

Prophet AIOS is the core platform of the Fourth Paradigm, used for building, deploying, and managing large models designed for specific industry applications.

• Natural language is used to assist business professionals in building industry models.
• Scenario models are created by utilizing data from various industries.
• It provides unified management of existing and newly created models for enterprises.
• It supports model deployment, inference, and lifecycle management.
• Adapted to domestic computing power and self-developed IT environments.
• Connects algorithm experts with task announcement resources.

How to implement industry-wide large models

The platform emphasizes building models around clear business metrics, rather than simply deploying a generic chatbot.

1. Identify the business objectives that need to be predicted, generated, or optimized.
2. Inventories data such as structured data, text, images, and devices.
3. Define data permissions, quality standards, and usage boundaries.
4. Select the industry-based model and computing environment.
5. Complete the training, fine-tuning, or retrieval enhancement configuration.
6. Use historical data for offline evaluation.
7. Set up manual review and business rules.
8. Perform grayscale verification at a low production volume.
9. Monitor accuracy, latency, and resource costs.
10. The data and models are continuously updated based on feedback.

4Paradigm AIGS

AIGS leverages the capabilities of large language models to use generative AI to assist companies in developing and modifying software.

• Generate code snippets based on requirements.
• Develop an enterprise knowledge base application.
• Assists with code review and issue identification.
• Integrate Copilot into existing business software.
• Tasks are executed automatically in accordance with the company’s rules.
• Split and process multi-step complex tasks.
• It supports application deployment and continuous iteration.

The three application phases of AIGS

Corporate knowledge base

AIGS can connect internal corporate knowledge and utilize knowledge graphs for cross-verification and tracing of results.

• First, organize the scope of knowledge by department and permissions.
• Remove duplicate, expired, and conflicting content.
• Configure version and responsible person for the document.
• Keep the evidence and citation locations in the response.
• Access to sensitive content is controlled based on job roles.
• High-risk answers must be confirmed by business staff.

Enterprise Copilot

Copilot can access the data and functions available within the software, and it assists employees in carrying out their tasks through conversations.

• Sales staff can analyze customer needs.
• The customer service agent can search for rules and generate draft responses.
• The product team can assist with simulation and design optimization.
• Operators can create and adjust marketing content.
• Developers can get assistance with code and reviews.
• Audit records must be kept for all automated operations.

4Paradigm SHIFT

SHIFT combines the underlying AI capabilities of the fourth paradigm with industry expertise to create deliverable intelligent solutions.

• Financial risk control and transaction anti-fraud.
• Retail recommendations and targeted marketing.
• Manufacturing quality inspection and equipment maintenance.
• Prediction of the status of energy equipment.
• Traffic inspection and anomaly detection.
• Medical and health risk management.
• Water supply forecasting and emergency support.

SageOne computing platform

SageOne provides the computational power needed for model development, training, deployment, and maintenance, through the integration of software and hardware as well as unified delivery.

AutoML capabilities

AutoML is used to reduce the complexity associated with feature processing, algorithm selection, parameter tuning, and model deployment.

• Suitable for data tasks with clear labels and business objectives.
• It cannot replace data quality and causal inference.
• An automatically optimal metric does not equate to the best business outcome.
• Fairness and stability must be checked before going live.
• After model drift, it needs to be re-evaluated and retrained.

Private deployment

AIGS supports private deployment in enterprises; the developers emphasize that data can remain local and there is no need to rely on external large models.

• It can be deployed in the customer’s own data center.
• It can be adapted to domestic software and hardware as required.
• The internal knowledge and permission system can be managed locally.
• The deployment topology and components depend on the project requirements.
• The methods for upgrading and authorization still need to be determined in an offline environment.
• Safety and performance tests must be completed prior to acceptance.

Tutorial on Using the Fourth Normal Form

1. Define the project objectives, target users, and success metrics.
2. Schedule a product discussion or demonstration through pre-sales channels.
3. Provide descriptions of the anonymized scenarios and data.
4. Choose a combination of Prophet, AIGS, SHIFT, or SageOne.
5. Assess the data volume, concurrency, model requirements, and computing power needs.
6. Decide on whether to deploy it in the cloud, locally, or as an integrated hardware and software solution.
7. Complete the concept validation and business metric assessment.
8. The contract specifies the prices, services, and data-related responsibilities.
9. Deploy the platform and integrate it with the identity and access management system.
10. Develop the first batch of models, knowledge bases, or intelligent applications.
11. Launch in grayscale and monitor quality and costs.
12. Continuously expand use cases through training and operation and maintenance mechanisms.

Product versions and prices

The Fourth Paradigm is designed for enterprise sales on a project basis; its official website does not offer any standardized monthly or annual subscription plans.

As of August 30, 2026, the official website offers only booking for demonstrations and pre-sales consultations; the final cost will be determined based on the proposed plan and the contract.

What should be verified during procurement?

• Software licensing is billed based on users, nodes, or computing power.
• Are there additional fees for model invocation and third-party components?
• Whether implementation, migration, training, and maintenance are included.
• How are upgrades, scale expansions, and the addition of new use cases priced?
• Hardware warranty and replacement timeline.
• Project acceptance criteria and procedures for handling cases that do not meet the standards.

Data security and privacy

In addition to the website’s privacy policy, enterprises should use contracts to clearly define who is responsible for handling business data and model data.

• Define the data controller, processors, and access personnel.
• Sensitive data is classified and anonymized before it is entered into the platform.
• Limit the scope of use for training, retrieval, and logging.
• Configure transmission encryption, storage encryption, and key management.
• Set the retention periods for logs, cache, and backups.
• Clarify the return and deletion of data after the project is completed.

Risk control in generative AI

• Answers from knowledge bases may contain hallucinations or incorrect references.
• The results of code generation must be tested and reviewed.
• Agents should use the minimum permissions when accessing business systems.
• High-risk decisions cannot be left entirely to models to make automatically.
• The training data must have its copyright and usage permissions verified.
• The model output should record the version, time, and evidence.

Official open-source projects

The official GitHub organization of Fourth Normal Form has made projects such as OpenMLDB and OpenAIOS Platform available.

In what scenarios is OpenMLDB suitable?

OpenMLDB is a machine learning database that focuses on ensuring consistency in feature computation, both online and offline.

• Real-time recommendations and targeted marketing.
• Detection of ad click fraud.
• Financial risk control and real-time decision-making.
• Stream-based data feature engineering.
• Reuse of training and online inference features.

Deployment requirements for the OpenAIOS Platform

This open-source platform is built on Kubernetes and requires capabilities in container management as well as storage operations.

• Kubernetes cluster.
• NGINX Ingress Controller.
• Harbor image repository.
• Keycloak identity system.
• CephFS shared storage.
• MongoDB database.

The boundary between open-source and commercial products

The fact that some of the underlying components are open source does not mean that all of Fourth Normal Form’s enterprise products and large-scale models can be deployed for free.

• Open-source repositories are used in accordance with their respective licenses.
• Commercial products require separate licensing to be purchased.
• Model weights and industry data are usually not made available along with the code.
• Enterprise support, implementation, and maintenance fall under business services.
• The costs associated with developing open-source solutions include computing power, security, and maintenance.

Which companies are suitable?

• Companies that need to build machine learning platforms on a large scale.
• Organizations that wish to deploy large industry models locally.
• Groups that possess a large amount of internal knowledge and business systems.
• Government and enterprise clients that require computing power tailored for domestic innovation.
• Industry clients that need anti-fraud, recommendation, or predictive applications.
• A team with capabilities in data governance and continuous operation.

Product advantages

• It covers the entire process, from model development to production and operation maintenance.
• Take into account both predictive AI and generative AI.
• It supports privatized and domestically developed computing environments.
• We offer industry-specific solutions and enterprise implementation services.
• It offers AutoML, model management, and computing power-related products.
• Some of the underlying technologies come with official open-source versions.

Usage restrictions and precautions

• It is designed for business purchases, not as a free chatting tool for individuals.
• The official website does not provide a unified, publicly listed standard price.
• The effectiveness of deployment depends on data quality and business cooperation.
• Privatization still requires ongoing operation and maintenance as well as model governance.
• The generation of results and automated decision-making require human oversight.
• Open-source components do not mean that the entire commercial platform is open source.

Frequently Asked Questions

What products does the Fourth Normal Form mainly offer?

These mainly include the Prophet AIOS, the AIGS generative software platform, the SHIFT industry solutions, and the SageOne computing platform.

What is AIOS Prophet suitable for?

It is used for building, deploying, and managing large industry-scale models, and it provides support for innovative computing resources as well as enterprise-level model management.

Can Fourth Normal Form AIGS generate code?

Yes, it supports code snippet generation, knowledge base applications, code review, deployment, and Enterprise Copilot.

Does the fourth paradigm support private deployment?

Yes, AIGS offers dedicated privatization solutions that allow corporate data to be processed within local environments.

How are Fourth Normal Form products priced?

The official website does not list any standard packages; quotes are usually determined based on the product modules, deployment method, computing power, and scope of implementation.

What are the official open-source projects related to the Fourth Normal Form?

The authorities have made projects such as OpenMLDB and OpenAIOS Platform available to the public; both of these projects are licensed under the Apache-2.0 license.

Is the Fourth Normal Form an open-source platform as a whole?

No, open-source components do not mean that AIOS, AIGS, SHIFT, SageOne, as well as the entire model, are open source.