Is AI back end complexity a pain point? Seeking feedback on an open source idea

4 months ago 6

Backend-as-a-Service for AI Applications

Build AI apps without backend headaches.
Community-driven, open, and built for developers.

Flexible Billing & Usage Tracking: Manage usage-based subscriptions, one-time credits, and real-time usage metrics automatically.
Unified LLM API: Streamlined prompt handling with support for multiple models, streaming session resuming, conversation history and AI credits deduction.
Real-Time Analytics: Visual dashboards to monitor user behavior, model popularity, revenue and more.
Open Source & Community-Driven: Transparent, collaborative, and shaped by contributors.

AiBase condenses the entire backend workflow for AI apps into three simple steps:

1. Configure Billing Plans

Define subscriptions or credit packages in minutes. AiBase automates payments, tracks usage, and deducts credits—no manual work required.

Set up credit packages and pricing tiers.

POST /api/pricing-plans Authorization: Bearer sk_test_abc123456789 Content-Type: application/json [ { "name": "Free", "description": "Basic Free Plan", "monthly_price": 0, "yearly_price": 0, "ai_credits": 100 }, { "name": "Pro", "description": "For teams and growing businesses", "monthly_price": 50, "yearly_price": 500, "ai_credits": 10000 } ]

A seamless credit package purchasing experience for your users.

const { purchasePlan } = useAiBase(); <Button onClick={async () => { const result = await purchasePlan({ planId: "pro", customerId: "cus_abc123456789" }); window.location.href = result.url; }} />

Send prompts and parameters to your preferred LLMs through AiBase's API. Streaming, session resuming, history management, and credits deduction are all handled for you.

See LLM responses stream live and credits update instantly.

const response = await fetch("/api/llm/complete", { method: "POST", headers: { "Content-Type": "application/json", Authorization: "Bearer sk_test_abc123456789" }, body: JSON.stringify({ model: "anthropic-claude-4-sonnet", messages: [ { role: "system", content: "You are an AI software engineer." }, { role: "user", content: "Plan and create a data pipeline for me." }, ], model_kwargs: { temperature: 0.4, max_tokens: 2048 } streaming: true, input_token_per_credit: 10000, output_token_per_credit: 5000, customer_id: "cus_abc123", tracing_tags: ["agent1"] }) })

Seamlessly continue interrupted sessions for a better user experience.

const response = await fetch("/api/llm/stream/resume", { method: "POST", headers: { "Content-Type": "application/json", Authorization: "Bearer sk_test_abc123456789" }, body: JSON.stringify({ session_id: "sess_1234567890" }) });

Access and manage previous conversations easily.

Monitor your AI app's performance in real time. Get actionable insights into user behavior, model popularity, and revenue. All from your dashboard.

Track how users interact with your AI features.

Ask our AI assistant anything about user performance and behavior.

We're building AiBase in the open, with the community. Here's what's coming up:

Full Auth Support: Comprehensive authentication and authorization for users and teams.
Long-Term Memory Store: Persistent storage for conversations and user data.
RAG/Knowledge Base Integrations: Support for Retrieval-Augmented Generation and knowledge base.
Serverless Function Runtime: Run custom backend logic and webhooks with ease.
Community Feature Requests: Top requests from users prioritized for future releases.

Want something? Open an issue or upvote your favorites!

Coming soon.

We welcome contributions! Whether it's bug fixes, feature suggestions, or documentation, your input helps shape AiBase. Fork the repo and submit a pull request, or start a discussion in the issues.

AiBase is already powering production AI applications: