Overview
Prowork integrates AI throughout its platform to help operations teams automate complex data workflows without writing code. Parabola does not train its own AI models. Instead, Parabola relies on third-party large language models (LLMs) under enterprise agreements that include Data Processing Addendums (DPAs), which contractually prohibit the model providers from using customer data to train their models. This means data processed through Prowork’s AI features is used solely to generate results for the customer and is never used for model improvement by the underlying providers.
AI use cases in Prowork
AI-assisted workflow building
Prowork’s in-app AI assistant helps users design and build automated workflows using plain language. Users describe what they want to automate conversationally, and the agent translates that into a working Prowork flow, adding steps, configuring settings, and troubleshooting errors along the way.
Prowork offers AI-powered steps that operate directly on data within a workflow, including the ability to categorize rows into user-defined groups, extract structured fields from unstructured text, standardize inconsistent values into canonical formats, and execute custom transformation logic described in plain language. These steps let users apply LLM-powered intelligence to their data at scale without writing code.
Intelligent data ingestion
Prowork uses AI with vision capabilities to extract structured data from documents like PDFs, invoices, packing lists, and purchase orders, even when formats vary across vendors or change over time.
AI-generated documentation
Prowork’s AI automatically generates and maintains documentation describing each workflow’s logic, supporting auditability, onboarding, and knowledge transfer as flows evolve.
Data privacy and security
Prowork takes a privacy-first approach to its AI features:
- No proprietary model training. Parabola does not train its own AI models on customer data or any other data.
- Enterprise agreements with DPAs. Parabola maintains enterprise-level agreements with its third-party AI model providers (incl. OpenAI and Anthropic) that include Data Processing Agreements. These agreements contractually prohibit the providers from using the contents of any data sent through Prowork’s AI steps as training data for their models.
- Data minimization. When using AI transformation steps, only the specific columns selected by the user are sent to the AI provider, not the entire dataset.
- Security standards. Parabola maintains SOC 2 Type II controls, encryption in transit and at rest, IAM-based least-privilege access, and regular third-party penetration testing.
In short, customer data processed through Prowork’s AI features is handled securely, used only to produce the requested output, and is never used to train or improve any AI model, whether Parabola’s or a third party’s.Last modified on August 29, 2026
