Skip to main content

AssistantBuilder

A fluent builder for assembling an assistant step by step before deploying. Each method returns self, so calls can be chained.

When to use this​

Use AssistantBuilder when you want a readable, declarative configuration — especially when adding multiple data sources or examples. For a one-liner, use create_assistant(). For full programmatic control, use the Assistant class.

Chained example​

from custodian_labs import AssistantBuilder

app = (
    AssistantBuilder()
    .model("gpt-4o")
    .prompt("You are a knowledgeable HR assistant.")
    .examples([
        {
            "user": "How many vacation days do I get?",
            "assistant": "Full-time employees receive 20 days of PTO per year.",
        }
    ])
    .data_source_file("./docs/employee-handbook.pdf")
    .data_source_file("./docs/benefits-guide.pdf")
    .deploy()
)

response = app.chat("What's the parental leave policy?")
print(response.response)

Methods​

.model(model: str)​

Set the LLM model.

builder.model("gpt-4o-mini")

.prompt(system_prompt: str)​

Set the system prompt.

builder.prompt("You are a helpful assistant.")

.examples(examples: list[dict])​

Add few-shot examples. Each dict must have "user" and "assistant" keys. Calling this method multiple times appends to the existing list.

builder.examples([
    {"user": "Hello", "assistant": "Hi! How can I help you today?"},
])

.data_source_file(path: str)​

Upload a file as a knowledge source for RAG retrieval. Supported formats: TXT, PDF, DOCX, XLS, XLSX. Calling this method multiple times adds each file as a separate data source.

builder.data_source_file("./knowledge-base.pdf")

.deploy()​

Create the assistant, upload any data sources, and deploy it as a live App. Returns an App object.

app = builder.deploy()
note

.deploy() makes network calls. Uploading large files may take a few seconds while the content is parsed and indexed.