Slides Presentation Builder

Creates a complete multi-slide presentation from a natural language brief.

presentation_builder.py
"""
Slides Presentation Builder
============================
Creates a complete multi-slide presentation from a natural language brief.

The agent plans slide structure, creates the presentation, adds slides with
appropriate layouts, inserts tables with data, and adds text annotations.

Key concepts:
- create_presentation: creates a new blank presentation
- add_slide: supports TITLE, TITLE_AND_BODY, TITLE_AND_TWO_COLUMNS, BLANK, etc.
- add_table: pre-populates tables with structured data
- add_text_box: positions text annotations on slides
- get_presentation_metadata: returns page size, slide IDs, and element positions in inches

Setup:
1. Create OAuth credentials at https://console.cloud.google.com (enable Slides API + Drive API)
2. Export GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET, GOOGLE_PROJECT_ID env vars
3. pip install openai google-api-python-client google-auth-httplib2 google-auth-oauthlib
4. First run opens browser for OAuth consent, saves token.json for reuse
"""

from typing import List, Optional

from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.google.slides import GoogleSlidesTools
from pydantic import BaseModel, Field


class SlideSpec(BaseModel):
    layout: str = Field(
        ..., description="Slide layout: TITLE, TITLE_AND_BODY, BLANK, etc."
    )
    title: Optional[str] = Field(None, description="Slide title text")
    body: Optional[str] = Field(None, description="Slide body content")


class PresentationPlan(BaseModel):
    title: str = Field(..., description="Presentation title")
    slides: List[SlideSpec] = Field(..., description="Ordered list of slides to create")


agent = Agent(
    name="Presentation Builder",
    model=OpenAIResponses(id="gpt-5.5"),
    tools=[GoogleSlidesTools()],
    instructions=[
        "Create well-structured presentations with logical slide flow.",
        "Use TITLE layout for the cover slide with a subtitle.",
        "Use TITLE_AND_BODY for content slides with bullet points.",
        "Use SECTION_HEADER to divide major topics.",
        "Use BLANK slides for tables and custom layouts.",
        "Always call get_presentation_metadata to get page dimensions and element positions.",
        "Use page_width_inches and page_height_inches to position elements correctly.",
        "Check existing element positions to avoid overlaps when adding text boxes.",
        "After creating, return the presentation URL.",
    ],
    markdown=True,
)


if __name__ == "__main__":
    agent.print_response(
        "Create a presentation titled 'Engineering Team Q3 Review'. Include: "
        "1. A title slide with subtitle 'Performance, Goals, and Roadmap'. "
        "2. An agenda slide listing Revenue, Key Metrics, Product Updates, Q4 Goals. "
        "3. A two-column slide comparing Q2 vs Q3 metrics. "
        "4. A blank slide with a 4x3 table of KPIs (MRR, Churn Rate, NPS, DAU with Q2 and Q3 values). "
        "5. A section header for 'Q4 Goals'.",
        stream=True,
    )

    # Smart layout: use metadata to position annotations without overlaps
    # agent.print_response(
    #     "Using the presentation you just created, call get_presentation_metadata "
    #     "to check slide dimensions and element positions. Then add a text box "
    #     "annotation in the bottom-right corner of the KPI table slide that says "
    #     "'Source: Finance Dashboard, Sept 2026'. Make sure it does not overlap "
    #     "with the table.",
    #     stream=True,
    # )

Supply actual Q2/Q3 values, or explicitly ask for labeled sample data: the script has no source for company metrics. Use five rows if the KPI table needs a header plus the four named metrics. PresentationPlan and SlideSpec are defined but not used as an output_schema; the agent creates slides through tools and returns free-form text.

For subsequent prompts, include the returned presentation ID. The agent has no database/history configuration to recall it in a new run. Check the created presentation visually after the tools complete; metadata-based positioning does not itself verify the rendered layout.

Run the Example

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Install dependencies

uv pip install -U agno google-api-python-client google-auth google-auth-httplib2 google-auth-oauthlib openai

Configure Google OAuth

Enable the Google API used by this example, configure the consent screen, and create a Desktop OAuth client in your Cloud project. Export its GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET, and GOOGLE_PROJECT_ID, or place the downloaded client JSON at credentials.json in the directory where you run Python. Google's Python quickstart shows the Desktop client setup.

The first tool call opens a browser for consent and caches credentials in token.json. Use a separate token_path when switching accounts. Passing an Agno user_id does not switch the authenticated Google account.

Export environment variables

export GOOGLE_CLIENT_ID="your_google_client_id_here"
export GOOGLE_CLIENT_SECRET="your_google_client_secret_here"
export GOOGLE_PROJECT_ID="your_google_project_id_here"
export OPENAI_API_KEY="your_openai_api_key_here"

Run the example

Save the code above as presentation_builder.py, then run:

python presentation_builder.py

Full source: cookbook/91_tools/google/slides/presentation_builder.py