OpenAI
OpenAI is one of the core AI platforms we use to build production features around language, reasoning, extraction, classification, summarisation, and agents. We use it where the model clearly improves the product, not as a layer added for effect. OpenAI’s API supports text, vision, and agentic use cases, and its business offerings are designed with enterprise data controls and privacy commitments in mind.
Why we use it
- /Strong general-purpose models for reasoning, language, and multimodal workflows
- /Good fit for extraction, summarisation, classification, and conversational features
- /Production-ready API and developer tooling
- /Enterprise controls for privacy, security, and data handling
When we reach for it
- /AI assistants grounded in internal or product data
- /Document extraction and summarisation workflows
- /Classification, routing, and structured output
- /Agentic workflows that need reasoning across multiple steps
- /Products where quality, latency, and model capability matter more than using the cheapest possible model
We start with the task, the data, and an evaluation set before choosing the model. Prompting, retrieval, latency, cost, and failure cases are measured before the feature reaches production.
We keep the architecture model-aware but not model-dependent. That means clear abstraction around providers, observability, fallbacks, and evaluation so the implementation can evolve as models change.
For business workloads, we also design around access controls, data handling, retention, and regional requirements from the start. OpenAI states that business data is not used for model training by default and offers controls including data retention options and regional data residency for eligible customers.
Common questions
What is OpenAI used for?+
OpenAI models can be used for text generation, reasoning, document analysis, image understanding, extraction, classification, summarisation, and agentic workflows.
Can you build an AI assistant using our own company data?+
Yes. We can combine OpenAI models with retrieval, permissions, and your internal data sources to build grounded assistants and knowledge workflows.
Does OpenAI train on our business data?+
By default, OpenAI states that data from its API and business products is not used to train its models unless the customer explicitly opts in.
Can OpenAI be used in enterprise environments?+
Yes. OpenAI provides business and enterprise controls around access, encryption, retention, compliance, and data residency for eligible customers.
Do you only build with OpenAI?+
No. We are model-agnostic and also work with platforms such as Anthropic and AWS Bedrock. We choose based on quality, cost, latency, privacy, and the specific workload.
Is higroup part of the OpenAI Partner Network?+
Yes. higroup is part of the OpenAI Partner Network, which is designed to help partners build, deploy, and deliver AI solutions with OpenAI.