Techno

Techno is an AI-powered API agent built for RSRoute.

It uses tool calling to understand natural-language requests, interact with APIs, work with files, search the web when necessary, and return structured results.

Techno is designed as a lightweight, self-hosted alternative for experimenting with Agentic AI + API workflows without depending on a paid AI agent platform.

CLI

python3 main.py

Streamlit UI

streamlit run agent_with_ui.py

Features

• Natural-language API tasks
• GET and POST API requests
• AI-powered tool calling
• Multiple tools in a single task
• API response inspection
• File creation and reading
• Web search
• Structured APIResult responses
• Configurable maximum tool calls
• Persistent conversation during a Streamlit session
• Streamlit chat interface
• Postman-style API result view
• Table and raw JSON result tabs
• Visible tool execution logs
• Google Gemini model support
• Local development through Python

How It Works

Techno uses a manual agent loop rather than a high-level agent executor.

User
 │
 ▼
Techno
 │
 ├── Tool call
 │      │
 │      ▼
 │   Tool result
 │      │
 │      ▼
 │   Techno
 │
 ├── Another tool call
 │      │
 │      ▼
 │   Tool result
 │
 ▼
Structured APIResult

For example, a request such as:

Send a POST request to
http://0.0.0.0:8000/v1/gemini/chat
with the required parameters.

can cause Techno to:

Natural language request
        ↓
    call_api
        ↓
   API response
        ↓
    Techno reads result
        ↓
    APIResult

For more complex tasks, Techno can chain tools:

call_api
   ↓
inspect response
   ↓
create_file
   ↓
final result

Available Tools

call_api

Makes HTTP requests to API endpoints.

Currently supports:

• GET
• POST

Example request:

Get http://0.0.0.0:8000/

Techno can determine the required tool arguments and execute the request.

create_file

Creates a file with supplied content.

Example:

Create info.csv with age and height data.

search_wikipedia

search specific topic on Wikipedia

Example:

search on Wikipedia about Andrew NG.

read_file

Reads an existing file from the agent workspace.

find_file

Finds files available in the workspace.

web_search

Performs a web search when additional information is required.

Techno is instructed to avoid unnecessary tool calls and prefer the shortest valid path to completing a task.

Structured Output

Techno uses a Pydantic model for its final response:

class APIResult(BaseModel):
    status: str
    summary: str
    api_result: Dict[Any, Any]

The final response contains:

• status — whether the operation succeeded or failed
• summary — a concise description of what happened
• api_result — the actual API/tool result

The final model response is parsed using PydanticOutputParser.

Streamlit Interface

Techno includes a Streamlit interface designed around API-agent workflows.

The interface provides:

Tool execution

Every tool call is displayed:

⚙ Tool: call_api

url
http://0.0.0.0:8000/v1/gemini/chat

method
post

✓ Tool result received

This makes Techno's agentic behavior observable instead of hiding tool execution.

Status

The final status is displayed separately in a compact box.

STATUS

✓ success

Successfully completed the API request.

API Result

Results are displayed using two tabs:

Table | JSON

Table provides a human-readable key/value representation.

JSON provides the raw structured result.

Installation

Clone the repository:

git clone <repository-url>
cd rsroute-agent

Create a virtual environment:

python3 -m venv .venv

Activate it:

source .venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Environment Variables

Create a .env file:

MODEL=gemini-3.5-flash-lite
GOOGLE_API_KEY=your_google_api_key
MAX_TOOL_USE=10

MODEL determines which Gemini model Techno uses.

GOOGLE_API_KEY is required by the Google Generative AI integration.

MAX_TOOL_USE limits the number of tool calls Techno can make for a single request.

Running Techno

Terminal version

The original agent can be run with:

python3 main.py

Streamlit version

Run:

streamlit run app.py

Streamlit will start a local web application where you can interact with Techno through a chat interface.

Example

Create a CSV

You:
Make a file named info.csv and create two columns:
age and height. Add 10 lines where height generally
increases with age.

Techno can call:

create_file

and produce the requested file.

API → File workflow

You:
Make a GET request to http://0.0.0.0:8000/
and write the output to root_output.csv.

Techno can perform:

call_api
    ↓
inspect response
    ↓
create_file

API request

You:
Send a POST request to
http://0.0.0.0:8000/v1/gemini/chat
with my API parameters.

Techno can construct the tool call, execute the request, inspect the response, and return an APIResult.