Generative AI
Easily transform your data to optimize for your specific use cases. Use your documents as context for LLMs by fine-tuning within DataRobot, building your own vector databases, or integrating your existing ones.
Build with fine-tuned LLMs or custom vector databases for RAG
Transform your organization’s unstructured documents into unique generative AI applications with streamlined processing to extract data.
Harness the power of your organization’s unique data to develop vector databases with customizable chunking and embedding strategies.
Manage these databases seamlessly with role-based access controls (RBAC), lineage tracking, and versioning. Maintain quality by filtering metadata, and quickly adding or removing specific documents.
Vector Database Builder to easily generate and manage custom VDBs
Build generative AI applications your way — whether deep-code, low-code, or a blended approach. Start with customizable, code-first application templates that include built-in business logic, application interfaces, and robust generative AI security. Or, create from scratch using enterprise-grade AI pipelines with notebooks or GUI workflows to streamline development and delivery. Tailor everything to fit your organization’s unique needs.
GUI and notebook-based workflows, Application Templates to accelerate delivery.
Quickly deliver comprehensive generative AI applications. Get your generative AI applications into users’ hands in hours, not weeks. Develop a repository of code-first templates for your teams, starting with our customizable and editable generative AI Application Templates. These templates offer seamless GitHub integration and built-in best practices for business logic, application interfaces, and generative AI security. Or, create your own shareable custom templates for a repeatable, scalable AI development process.
Accelerate delivery from idea to application. Empower your business users with high-quality generative AI solutions through our Declarative API Framework. With the ability to easily replicate work and visualize and save AI pipelines, you can scale AI initiatives rapidly and develop innovative, comprehensive generative AI applications faster—driving informed decisions that impact your bottom line.
Enable iterative experimentation. Elevate the quality of your generative AI applications by creating agentic flows in our Playground, using structured output from your queries. Visually compare different experiments—whether developing with a code-first notebook or through our GUI workflow—to assess output quality side-by-side and track changes effortlessly.
Generate question and answer pairs for rapid training
Launch an interactive application for business stakeholders to provide early feedback
Benchmark and evaluate your experiments. Gain confidence in your generative AI experiments by applying statistical evaluation and CI/CD testing metrics. Easily compare performance to ensure you choose the best model for production.
Evaluation and CI/CD testing metrics like response correctness, tokens, latency, and more
Use and compare best-of-breed components. Optimize your generative AI experiments by tailoring them to your specific use cases with the tools you prefer. Experiment with popular proprietary LLMs directly from the platform—keys included—or host open-source LLMs to pinpoint the best foundation model for your needs, all seamlessly supported out-of-the-box.
Bring your own LLM: fine-tuned, open-source, custom built, third-party, small language (SLM)
Bring your own embedding model or vector database
Access popular LLMs out-of-the-box
Build your own vector database
Red-team applications before deployment. Gain confidence in the security and reliability of your generative AI applications by red-teaming your projects before deployment. Use synthetic data to test for jailbreaks, bias, response correctness, toxicity, and other compliance metrics, proactively identifying and addressing vulnerabilities. Once in production, benefit from a wide library of guards for ongoing protection.
Red-team project prior to deployment: jailbreaks, bias, response correctness, toxicity, and more
Gain deep insight into every prompt and response. Dive into the underlying documents and lineage linked to every prompt and generated response. Validate alerts and intervention decisions, keep your vector databases current, and know when it’s time to retrain your generative AI models to ensure they remain effective and reliable.
Generative AI forensics for deep insight into every prompt and generated response.
Continuously improve application quality. Gain deep insight into your generative AI applications’ performance to drive continuous improvement, ensuring your business stakeholders can make informed decisions based on accurate data.
Track the most important metrics for your use case or organization
Improve your vector databases by assessing alignment with user prompts
Dive deep into the underlying documents leveraged for each generated output
Incorporate human feedback on the quality of generated outputs
Easily stay in compliance. Stay compliant with rapidly changing generative AI regulations while simplifying your workload with automated documentation. Run compliance tests for PII protection, toxicity, or prompt injection, generating reports with a single click. Customize these reports to align with your business or industry standards, and use flexible deployment options to ensure your AI infrastructure meets regulations like BAA and HIPAA.
Flexible platform options help comply with country or industry standards (e.g., BAA, HIPAA)
Support for Nutanix, on-premises, and STS deployments
Enforcement of your MRM policies and approval workflows
Real-time LLM guards and compliance monitoring
Detect and prevent unwanted behaviors and hallucinations with a robust library of out-of-the-box, custom and open-source guard models from NVIDIA, Microsoft, and more.
Real-time intervention: Automatically moderate generated content to prevent PII leaks, jailbreak attempts and toxic prompts.
Align stakeholders on model methodology. Overcome the challenge of explaining complex mathematical concepts to business and compliance stakeholders by automatically generating clear, plain-language documentation and presentations.
prompt token count, response token count, total tokens, total cost
toxicity, relevancy, prompt injection, PII or PHI detection
sentence count, word count, sentiment, topic, Flesch reading ease, Dale Chall readability
ROGUE, BLEU, METEOR, SelfCheck GPT, Faithfulness, Relevance
Maintain consistent standards. Establish scalable security and governance standards with customizable, shareable guard models, intervention rules, and data security guardrails. Ensure your team consistently applies best practices unique to your business to meet regulatory and safety standards.
Repeatable and scalable security and governance standards to maintain best practices.
Create intuitive generative AI application interfaces. Empower business stakeholders to interact seamlessly with your generative AI applications using the Application Workshop. Create front-end application interfaces through an intuitive, no-code workflow that supports custom options like Streamlit, Dash, and Shiny.
Application Workshop: Easy GUI to create application interfaces.
Integrate with existing business applications. Seamlessly integrate generative AI applications with existing chat applications to meet business stakeholders where they already work. Drive usage and track adoption to showcase the business impact of your generative AI solutions.
Stay agile, even as things change. Seamlessly swap underlying generative AI components and URLs without disrupting production, allowing you to adapt to changing needs or market trends as newer, more suitable components emerge.
Update LLMs, embedding models, chunking strategies and URLs without breaking production.