APIs Foundry — Kirsantov


 █████╗ ██████╗ ██╗███████╗    ███████╗ ██████╗ ██╗   ██╗███╗   ██╗██████╗ ██████╗ ██╗   ██╗
██╔══██╗██╔══██╗██║██╔════╝    ██╔════╝██╔═══██╗██║   ██║████╗  ██║██╔══██╗██╔══██╗╚██╗ ██╔╝
███████║██████╔╝██║███████╗    █████╗  ██║   ██║██║   ██║██╔██╗ ██║██║  ██║██████╔╝ ╚████╔╝
██╔══██║██╔═══╝ ██║╚════██║    ██╔══╝  ██║   ██║██║   ██║██║╚██╗██║██║  ██║██╔══██╗  ╚██╔╝
██║  ██║██║     ██║███████║    ██║     ╚██████╔╝╚██████╔╝██║ ╚████║██████╔╝██║  ██║   ██║
╚═╝  ╚═╝╚═╝     ╚═╝╚══════╝    ╚═╝      ╚═════╝  ╚═════╝ ╚═╝  ╚═══╝╚═════╝ ╚═╝  ╚═╝   ╚═╝

GPT-based API Constructor

Turn prompts into production-ready APIs using Next.js on the frontend and FastAPI on the backend.

An experimental platform for building and shipping small, focused GPT-based APIs without writing backend code for every idea.

What it is

APIs Foundry is an experimental platform where I build and ship small, focused GPT-based APIs. Instead of writing backend code for every idea, you describe what the endpoint should do, define the JSON output, choose a model – and the system generates a FastAPI endpoint for you.

Think of it as a constructor for AI-powered microservices. You define the behavior in natural language, the platform handles the infrastructure, API generation, and documentation.

How it works

1. Define your endpoint

- Describe the task in natural language (for example: "summarize and classify support tickets into bug, feature, or billing")

- Optionally use a built-in assistant to help you refine the prompt

2. Design the JSON response

- Specify the response schema: fields, types, and possible values

- The platform enforces this schema on the FastAPI side for predictable outputs

3. Choose a model

- Select from GPT-based models (for example gpt-4.1, gpt-4.1-mini, or other compatible models)

- Configure parameters like temperature and max tokens

4. Configure visibility

- Private endpoint for your own usage

- Or shareable endpoint for the community catalog

5. Get an instant API

- APIs Foundry generates a FastAPI route such as:

POST /api/{user-or-space}/{endpoint_slug}

- OpenAPI docs and example requests are generated automatically

Example GPT-based APIs

Support Ticket Triage

Summarize a support ticket and classify it as bug, feature, or billing.

Input: ticket_text

Output: JSON with { summary, type }

Meeting Actions Extractor

Take a meeting transcript and extract action items per participant.

Input: transcript

Output: JSON with { summary, actions: [{ owner, task, due_date_hint }] }

Outbound Email Personalizer

Personalize a sales email based on a lead description and a base template.

Input: lead_context, base_template

Output: JSON with { subject_line, personalized_email }

Review Sentiment & Features

Analyze a user review, detect sentiment and product features mentioned.

Input: review_text

Output: JSON with { sentiment, mentioned_features: [{ name, sentiment }] }

Tech stack

The frontend is built with Next.js and React, leveraging server-side rendering and a modern component system. The backend is implemented with FastAPI, which exposes clean REST endpoints and generates OpenAPI documentation out of the box.

Each endpoint uses GPT-based models under the hood, but presents a stable, typed JSON interface to the caller.

┌──────────────────────────────────────────────────┐ │ Component Technology │ ├──────────────────────────────────────────────────┤ │ Frontend Next.js (React 19) │ │ Backend FastAPI │ │ AI Layer GPT-based models │ │ Docs Auto-generated OpenAPI │ │ State React hooks & server components │ └──────────────────────────────────────────────────┘

Pricing idea

This project is still experimental, but the long-term idea is to support three tiers:

Free Plan

Access to public community endpoints with basic limits

- Use public community endpoints

- Limited requests per day

- Read-only access to endpoint catalog

Pro Plan

Create your own endpoints and optionally publish them to the community catalog

- Create and manage your own endpoints

- Publish to community catalog (optional)

- Higher request limits

- Basic analytics and monitoring

Max Plan

Private-only endpoints with higher limits and advanced configuration

- Private endpoints only (not visible in catalog)

- Highest request limits

- Advanced model configuration

- Custom rate limiting

- Priority support

🚧 Experimental Project

APIs Foundry is currently in the experimental phase. The concept is being validated and the architecture is being refined. If you're interested in following the progress or have ideas for endpoints you'd like to see, I'd love to hear from you.

-> Back to Lab -> Get in Touch