What people build with Qai.

The bread-and-butter LLM use case. A user types a question, your app sends it to a model with some context, the model streams back an answer. Wire it into a Discord bot, a Slackbot, a customer support widget, a help overlay inside your app, or a standalone chat product.

const stream = await qai.chat.completions.create({
  model: 'qai-pro',
  messages: [
    { role: 'system', content: 'You are a helpful support agent for Acme.' },
    ...conversationHistory,
    { role: 'user', content: userMessage },
  ],
  stream: true,
});

for await (const chunk of stream) {
  const delta = chunk.choices[0]?.delta?.content;
  if (delta) {
    res.write(delta);  // stream to client
  }
}

Blog posts, email replies, social captions, product descriptions, ad variants, internal docs. The "stare at the blank page" problem solved with a prompt and a model. Common pattern: user fills in a short brief, you generate 3 variants, they pick one and edit.

Retrieval-augmented generation: take a knowledge base of documents, find the chunks most relevant to a user's question, stuff them into the prompt, ask the model to answer using only those chunks. The standard pattern for "AI that knows about my company / product / data."

Free-form input in, structured output out. Receipts to JSON. Resumes to candidate records. Customer notes to ticket fields. Emails to CRM entries. The pattern that quietly replaces a thousand regex rules.

response = client.chat.completions.create(
    model="qai-pro",
    messages=[
        {"role": "system", "content": """Extract candidate info as JSON:
{
  "name": string,
  "email": string,
  "years_experience": number,
  "skills": string[],
  "current_role": string
}
Output JSON only. No prose."""},
        {"role": "user", "content": resume_text},
    ],
)

# Always run through clean-json to handle markdown fences etc.
cleaned = requests.post(
    "https://llm.quickcasa.ai/v1/utilities/clean-json",
    headers={"Authorization": f"Bearer {api_key}"},
    json={"input": response.choices[0].message.content},
).json()

candidate = cleaned["json"]

Avatar generators, product mockups, AI art tools, social-media-card factories, marketing visual pipelines. With Qai-hosted media URLs, your app does not even need its own storage bucket - you call the API and embed the returned URL.

curl -X POST https://llm.quickcasa.ai/v1/images/generations \
  -H "Authorization: Bearer $QAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qai-imagine-turbo",
    "prompt": "a friendly robot mascot holding a coffee mug, flat illustration style",
    "hostMedia": true
  }'

# Response includes: { "data": [{ "url": "https://llm.quickcasa.ai/media/{id}" }] }
# That URL is permanent. Embed it anywhere.

Marketing clips, product demo b-roll, social hooks, intro animations, ad creative. Six seconds at a time, hosted on the Qai CDN, ready to embed. Stitch a few together for a longer piece.

Anywhere you need a model to think before it speaks. Scheduling assistants, financial decisioning, multi-constraint planners, code review bots, debate referees. qai-think spends extra compute working through a problem instead of guessing the first plausible answer.

n8n, Make, Zapier, custom cron jobs, GitHub Actions, internal data pipelines - anywhere a workflow needs "an LLM call" as one of its steps. Qai's OpenAI-compatible endpoint slots into every automation platform that already supports OpenAI.

Point Cursor, Continue, Aider, Cline, or any custom-base-URL coding tool at Qai and get a paid coding assistant for your team. Mix tiers across operations - cheap for autocomplete and quick refactors, qai-max for big architectural changes.