> For the complete documentation index, see [llms.txt](https://pumpad.gitbook.io/pumpad.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://pumpad.gitbook.io/pumpad.ai/technical-architecture.md).

# Technical Architecture

Pumpad’s platform is built around two core service layers:

* **AI Service Engine** — powers data, content, and model tasks
* **Community Tool Layer** — Telegram-based bot plugins for user & task interactions

These layers are connected via a unified **task dispatcher** and **PUAI billing system**, creating a seamless “function-as-a-service” experience.

***

#### 🧠 AI Service Engine

Pumpad’s AI engine is a hybrid middle-layer built to call top-tier AI APIs and host internal tools. Key capabilities:

* **Model Router:** Compatible with OpenAI, Anthropic, Stability AI
* **Data Engine:** On-chain parsing, user clustering, predictive analytics
* **Content Tools:** Tweet generator, article summarizer, multilingual content
* **Risk Modules:** Smart contract scanner, wallet behavior alert, scam detector

Core stack: Python + Node.js + queue-based parallel dispatch.

***

#### 🤖 Telegram Bot System

Pumpad’s Telegram automation tools are serverless (Cloudflare Workers + Supabase), with each plugin acting as an **AI micro-agent**.

Live plugins include:

| Plugin Name     | Function                                        |
| --------------- | ----------------------------------------------- |
| TaskBot         | Launch tasks, verify results, rank participants |
| RankBot         | Score and rank user activity                    |
| AirdropVerifier | Wallet check, task sync, airdrop eligibility    |
| InsightBot      | User behavior dashboard and segmentation        |
| AIWriter        | Tweet generator with custom keyword libraries   |

All plugins are unlocked via PUAI usage.

***

#### 🔌 Plugin System & Extensibility

Pumpad will offer a **Plugin SDK** for developers to build and submit their own AI modules. Future capabilities:

* Plugin-to-plugin interactions
* Cross-group point syncing
* Custom model deployment for BRC20/NFT projects