Technology
OpenAI
OpenAI makes the GPT family of language models and offers them through its API. I have built with GPT since early 2023, starting with a chatbot that reproduced how a specific person wrote
- Type
- Technology
- Official site
- openai.com
- In my work
- Services 3
- Published
Where I use it
From idea to a working MVP
MVP development for startups: scope, UX, design, frontend, backend, AI and launch, by one senior engineer who has been a founder and a CTO
ServicesA fractional CTO for your startup
Fractional CTO for startups: architecture, team, code review and delivery, from a CTO who also builds. Hours a week, part-time or full-time
ServicesAI agents and LLM apps that work in production
AI agent and LLM development: agents, chatbots on your data, AI features, voice and video pipelines built into a working product, not a demo
What is the OpenAI API and what have I built with it?
My first AI product was built on GPT in early 2023, when most teams were still showing demos. Since then I use OpenAI models next to Claude, chosen per task
Key facts
How I use OpenAI
- Imit, 2023. As CTO I built a chatbot on GPT that reproduced how a specific person wrote, based on their message history.
- AI features in products. Generation, search, classification, summaries and recommendations as ordinary product features. See AI development.
- One layer, several providers. OpenAI and Claude models sit behind one layer in my code, so a product can switch the model for a task without a rebuild.
When I choose OpenAI and its limits
OpenAI models are a strong choice for general assistants, structured output, speech and images. The best model differs per task, so I test on real data before choosing, and I compare price and speed, not only quality.
Data protection is agreed before the build: business APIs, only the data the task needs and clear limits on what the model may do. A model's answer is checked before it changes anything important.
Used together with
Sources
- OpenAI API Documentation — OpenAI ·
