book

Multimodal Generative AI in the Enterprise: From Pixels to Profit

Name: Multimodal Generative AI in the Enterprise: From Pixels to Profit
ISBN: 9781806111671

by Maximilian Tschochohei, Fabian Schenker

June 2026

Intermediate

248 pages

6h 25m

English

Packt Publishing

Read now

Unlock full access

Preface
Free benefits with your book
Chapter 1: Introduction to Generative AI and Multimodality
The new frontier of content and creationFrom analysis to creation: a paradigm shiftThe power of perception: defining multimodal AIThe intersection: multimodal generative AIHow multimodal generative AI worksThe building blocks: neural networks and deep learningFusing modalities: creating a unified understandingCore technical challenges for enterprise deploymentRevolutionizing marketing and customer engagementCase studies in action: from global brands to niche retailersExpanding applications across the enterpriseA practical guide to enterprise implementationThe future is agenticSummaryReferences
Chapter 2: Enterprise Context and Case Study
The 2025 and 2026 enterprise environmentsOur case study: Urban-Wearables and the Aura Smart JacketExploring goals as Urban-Wearables' new Head of Marketing TechnologySummaryReferences
Chapter 3: Preparing the Environment
Installing Python for your operating systemOpening and navigating the terminalSetting up the IDE and Jupyter NotebooksCloning the Course RepositoryLaunching your first Notebook in VS CodeInstalling and configuring the Frontier AI SDKsInstalling the Python client librariesConfiguring Google Cloud Gemini Enterprise Agent PlatformOption A: Setting environment variables (recommended)Option B: Explicit project and key argumentsVerifying your Gemini connectionConfiguring the OpenAI SDKSummaryReferences
Chapter 4: Text Generation
Technical requirementsTechnology foundations of large language modelsThe self-attention mechanismFrom text to math with embeddings and positional encodingsRetrieving information with feedforward networksAI provides a competitive edge to marketersOur framework for applying AI to marketing text generationMarket intelligence with generative AISynthetic persona generationBuilding a hyper-personalized content generatorAdvanced prompt engineeringSummaryReferences
Chapter 5: Image Generation
Technical requirementsTechnology foundations of AI image generationUnderstanding the core technologiesVAEsGANsDiffusion modelsA comparative analysis of generative architecturesImage generation for marketingMastering the art of the promptImage editing for marketingEditing visuals with precision – the concept of maskingBackground swaps – setting the stage for your productInpainting – removing and replacing elements seamlesslyOutpainting – expanding your creative canvasMask-free image editingUsing LLMs for image creation and editingPractical considerations for enterprise deploymentBrand style guide for AIIntegrate into existing creative workflowsReview and approval processesBuild a centralized asset management systemTraining creatives in prompt engineering and AI literacySummaryReferences
Chapter 6: Audio Generation
Technical requirementsThe foundations of AI audio generationExploring early approaches – concatenative and parametric TTSThe waveform revolution – WaveNetModern state-of-the-art – transformers and diffusionAudio understanding with automatic speech recognitionHow Whisper changed the world of ASREnterprise applications of ASRPractical example – transcribing an audio file with TransformersGenerating speech with TTSFrom text to lifelike voiceEnterprise applications for TTSPractical examples – generating voice with cloud-based and local TTS modelsGenerating music and sound effectsMastering the prompt for musicEnterprise applications for generating music and audioPractical example – creating a jingle with MusicGenPractical considerations for enterprise deploymentCreating an AI audio brand guideCopyright and commercial use in AI musicTraining creatives in audio prompt engineeringSummaryReferences
Chapter 7: Video Generation
Technical requirementsUnderstanding video generation technology foundationsEarly attempts with Generative Adversarial NetworksDiffusion models increase coherenceTransformers for video as a sequence of patchesModern video generation modelsPractical video generation for enterprisesPreparationUse case 1 – Creating marketing shorts from text (text-to-video)Use case 2 – Animating product images (image to video)Use case 3 – Generative video with soundEnterprise deployment – From experiment to productionThe next wave of generative videoLong-form and story-driven contentInteractive and controllable synthesisThe convergence with 3D and immersive worldsSummaryReferences
Chapter 8: Responsible AI: Ethics, Bias, and Safety
Introduction to RAIModality-specific challengesImage modality – navigating bias, copyright, and deception in image generationThe persistence of stereotypes and representational harmsTraining data and output ownershipHarmful and deceptive imageryAudio modality – addressing bias and misuse in audio generationBias in accents, gender, and inclusivityVoice cloning for fraud and defamationVideo modality – confronting the challenges of synthetic realityTechnical hurdles and enterprise viabilityDeepfake videosAn actionable framework for responsible multimodal AIStrategy 1 – Establish robust AI governanceStrategy 2 – Curate, augment, and document your dataStrategy 3 – Engineer for fairness and safetyStrategy 4 – Transparency and authenticityYour multimodal RAI checklistSummaryReferences
Chapter 9: Security in Multimodal Enterprise AI
Technical requirementsHow attackers break multimodal AIMultimodal prompt injectionSteganographySide-channel attacksReal-world example – the Samsung leakSecuring your multimodal AI applicationTailor your architecture to your data protection obligationsProtecting AI systems from prompt injectionSummaryReferences

Chapter 10: Measuring ROI and Driving Business Value
Why financial metrics matterBusiness value from multimodal AI applicationsProduct photo generationShort-form video production and localizationHyper-personalization at scaleAI-driven product tagging and catalog managementThe economics of AI asset generationDeconstructing traditional photography costsComparing generative AI costsDefining the total cost of ownership (TCO)Infrastructure and API computeData engineering and implementation costsSystem maintenance and drift detectionHuman capital and operational inefficienciesEstablishing ROI and KPIKey Performance Indicators (KPI)The human-in-the-loop (HITL) factorSummaryReferences
Chapter 11: Project Definition: Automated Marketing Asset Generation
Introduction to the Agentic Developer WorkflowFunctional and Non-Functional RequirementsFunctional RequirementsNon-Functional RequirementsMapping Out the Automated PipelineStep 0: The Input TriggerStep 1: Market Analysis AgentStep 2: Persona Definition AgentStep 3: Campaign Formulation AgentStep 4: Asset Creation EngineCreating the Technical Specification (spec.md)Summary
Chapter 12: Testing, Deployment, and Iteration
Technical requirementsTesting the multimodal pipelineTesting against reference imagesUsing an LLM as a judgeHuman-in-the-Loop (HITL) for quality assurancePackaging and deploying the pipelinePackaging with Uvicorn and FastAPIServerless containerizationDeployment architectureHandling production realitiesIteration: deploying new versions to productionThe AI iteration cycle and version controlAdvanced deployment strategiesMonitoring, drift detection, and feedback loopsThe performance feedback loopSummary
Chapter 13: Building the Pipeline
Setting up the projectUsing Pydantic schemas to manage outputImplementing workflows and tool callingPrompt rewriting with the AI creative directorMultimodal synthesis and asset generationAssembling and running the ApplicationSummary
Index

Content preview from Multimodal Generative AI in the Enterprise: From Pixels to Profit

9 Security in Multimodal Enterprise AI

Bringing "multimodal" AI—models that can see, hear, and speak—into your company is a huge upgrade from the text-only chatbots of a few years ago. But it also opens up a whole new can of worms for security.

Back when AI just handled text, security was relatively straightforward. You could set up filters to catch things like credit card numbers or bad words. But now that we have models like GPT-5.2 and Gemini 3 Pro that can process images, audio, and video all at once, defenses based on filtering text are not enough anymore.

This gap is dangerous. If an AI can see a diagram on a whiteboard, it can also accidentally leak the trade secrets drawn on it. If it can hear a meeting, it can leak your confidential ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Three Approaches to Measuring and Managing AI ROI

Publisher Resources

ISBN: 9781806111671

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Multimodal Generative AI in the Enterprise: From Pixels to Profit

by Maximilian Tschochohei, Fabian Schenker

9

Security in Multimodal Enterprise AI

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.