Build a Chatbot with Fikra API in 10 Minutes (OpenAI-Compatible, M-Pesa Payments)

#ai #programming #automation #api

Build a Chatbot with Fikra API in 10 Minutes (OpenAI-Compatible, M-Pesa Payments)

If you're an African developer trying to build AI apps, you've probably hit this wall: Stripe doesn't work in Kenya. You can't use global AI APIs like OpenAI or Anthropic because they require credit cards and payment rails that don't exist here.

I built Fikra API to fix this. It's a fully OpenAI-compatible AI inference gateway that works with M-Pesa, has a $1 minimum (2M tokens), and requires no credit card. You get 87% cheaper pricing than OpenAI.

In this tutorial, you'll build a working chatbot in Python using Fikra API. No code rewrites needed.

What You'll Build

A conversational chatbot that remembers previous messages in the conversation. By the end, you'll have:

A working Python chatbot
Understanding of OpenAI-compatible API calls
Your Fikra API key (with 2M free tokens)

Prerequisites

Python 3.8+ installed
Basic Python knowledge
Fikra API key (get one at fikraapi.co.ke)

Step 1: Install the OpenAI SDK

Fikra API is 100% OpenAI-compatible, so you use the same SDK:

pip install openai

Step 2: Set Up Your API Key

Create a file called chatbot.py:

import openai
import os

# Set your API key
os.environ["FIKRA_API_KEY"] = "fk_live_YOUR_KEY_HERE"

# Initialize Fikra client (change base_url only)
client = openai.OpenAI(
    base_url="https://api.fikraapi.co.ke/v1",
    api_key=os.getenv("FIKRA_API_KEY")
)

Security Tip: Use environment variables instead:

export FIKRA_API_KEY="fk_live_YOUR_KEY_HERE"

Step 3: Make Your First Chat Completion

Add this to chatbot.py:

response = client.chat.completions.create(
    model="fikra-fast-8b",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello! What can you do?"}
    ]
)

print(response.choices.message.content)

Run it:

python chatbot.py

Expected Output:
Hello! I'm a helpful AI assistant. I can help you with:

Answering questions

Writing code

Explaining concepts

And much more!

Step 4: Make It Interactive (Conversation Memory)

Now let's add conversation memory so the bot remembers previous messages:

conversation = [
    {"role": "system", "content": "You are a helpful assistant."}
]

while True:
    user_input = input("You: ")

    if user_input.lower() == "quit":
        break

    conversation.append({"role": "user", "content": user_input})

    response = client.chat.completions.create(
        model="fikra-fast-8b",
        messages=conversation
    )

    assistant_reply = response.choices.message.content
    conversation.append({"role": "assistant", "content": assistant_reply})

    print(f"Assistant: {assistant_reply}")

Example Conversation:
You: What's the capital of Kenya?
Assistant: The capital of Kenya is Nairobi.

You: What's special about Nairobi?
Assistant: Nairobi is Kenya's largest city and serves as the country's economic hub. It's known for Nairobi National Park (the only wildlife park within a city capital), vibrant culture, and as a major business center in East Africa.

You: quit

Step 5: Add Streaming (Real-Time Responses)

For faster, more natural responses, use streaming:

response = client.chat.completions.create(
    model="fikra-fast-8b",
    messages=[
        {"role": "user", "content": "Write a poem about Kenya."}
    ],
    stream=True
)

for chunk in response:
    if chunk.choices.delta.content:
        print(chunk.choices.delta.content, end="", flush=True)

Output:
In the heart of East Africa lies Kenya,
Where sun-kissed plains and mountains sway...

Tokens appear as they're generated, not waiting for the full response.

Why Fikra API? (The Pain Point)

Problem with Global APIs	Fikra Solution
Stripe doesn't support Kenya/East Africa	Native M-Pesa top-ups from your phone
$20+ monthly minimums	$1 minimum spend (2,000,000 tokens)
Credit card required	No credit card needed
High costs ($15/1M tokens)	87% cheaper ($2/1M tokens)
Slow global routing	Local Kenya infrastructure

The math: 2M tokens per dollar = 20,000 chatbot messages per $1.

Available Models

Model String ID	Context Window	Best For
`fikra-fast-8b`	8,192 tokens	Real-time chat, basic routing
`fikra-pro-20b`	32,768 tokens	Production workflows, agentic tasks
`fikra-pro-120b`	128,000 tokens	Complex coding, deep analysis

Troubleshooting

Error	Fix
`401 Unauthorized`	Check API key starts with `fk_live_`
`402 Payment Required`	Top up via M-Pesa at fikraapi.co.ke
`429 Too Many Requests`	Hit 30 RPM limit (unverified account). Wait 1 minute
`model not found`	Use exact strings: `fikra-fast-8b`, `fikra-pro-20b`