Mellea
build enterprise AI without guesswork

uv pip install mellea

Mellea

Build reliable generative programs with our library that merges the power of LLMs with the predictability of engineering.

Get started

Python not ProseWrite object-oriented code instead of free-form text.
Requirements DrivenDefine requirements that ensure good model responses.
Predictable and ResilientProvide structures for failure detection and recovery.
MCP and A2A CompatibleConnect Mellea programs with other agent frameworks.

Instruct, Validate, Repair

Clean interfaces for Gen AI output validation and sampling strategies.

Learn more

import mellea
from mellea.stdlib.sampling import RejectionSamplingStrategy


def write_email_with_strategy(m: mellea.MelleaSession, name: str, notes: str) -> str:
    email_candidate = m.instruct(
        f"Write an email to {name} using the notes following: {notes}.",
        requirements=[
            "The email should have a salutation.",
            "Use a formal tone.",
        ],
        strategy=RejectionSamplingStrategy(loop_budget=3),
        return_sampling_results=True,
    )

    if email_candidate.success:
        return str(email_candidate.result)

    # If sampling fails, use the first generation
    print("Expect sub-par result.")
    return email_candidate.sample_generations[0].value
Copy

Generative Slots

Build composable Gen AI using tried-and-true abstractions.

Learn more

@mellea.generative
def classify_sentiment(text: str) -> Literal["positive", "negative"]:
  """Classify the sentiment of the input text as 'positive' or 'negative'."""

sentiment = classify_sentiment(m, text=customer_review)

if sentiment == "positive":
    msg = m.instruct("Thank the customer for their post")
else:
    msg = m.instruct(
       description="Apologize for the customer's negative experience and offer a 5% discount for their next visit",
       grounding_context={"review": customer_review}
    )

post_response(msg)
Copy

MObjects

Seamlessly sprinkle GenAI into existing systems.

Learn more

import mellea
from mellea.stdlib.mify import mify, MifiedProtocol
import pandas
from io import StringIO


@mify(fields_include={"table"}, template="{{ table }}")
class MyCompanyDatabase:
  table: str = """| Store      | Sales   |
                    | ---------- | ------- |
                    | Northeast  | $250    |
                    | Southeast  | $80     |
                    | Midwest    | $420    |"""

  def transpose(self):
    pandas.read_csv(
      StringIO(self.table),
      sep='|',
      skipinitialspace=True,
      header=0,
      index_col=False
    )


m = mellea.start_session()
db = MyCompanyDatabase()
assert isinstance(db, MifiedProtocol)
answer = m.query(db, "What were sales for the Northeast branch this month?")
print(str(answer))
Copy

import mellea
from mellea.stdlib.sampling import RejectionSamplingStrategy


def write_email_with_strategy(m: mellea.MelleaSession, name: str, notes: str) -> str:
    email_candidate = m.instruct(
        f"Write an email to {name} using the notes following: {notes}.",
        requirements=[
            "The email should have a salutation.",
            "Use a formal tone.",
        ],
        strategy=RejectionSamplingStrategy(loop_budget=3),
        return_sampling_results=True,
    )

    if email_candidate.success:
        return str(email_candidate.result)

    # If sampling fails, use the first generation
    print("Expect sub-par result.")
    return email_candidate.sample_generations[0].value
Copy

Mellea build enterprise AI without guesswork

Mellea

Mellea
build enterprise AI without guesswork