AI‑Accelerated Marketing on Agentic Workflows - Future Proof Films
Adam Kleinberg · May 7, 2025
Recorded on May 7, 2025, for the AI Insiders series by AI Marketers Guild, this candid conversation features Adam Kleinberg, CEO & Co‑Founder of Traction—sharing a roadmap for practical AI adoption in modern marketing organizations.
Adam lays out how agentic workflows, a “liquid workforce” model, and new AI video techniques can slash production time and costs while amplifying creative impact.
[2:46] What Is a Liquid Workforce Model in Marketing?
Answer / Description: A liquid workforce model is an agile staffing approach where brands access a curated, highly flexible network of fractional talent and strategic partners to scale marketing efforts without the overhead of a traditional, fixed-staff agency. This model allows marketing organizations to eliminate structural bloat and rapidly deploy specialized talent as client needs shift.
In 2019, the agency Traction transitioned from a traditional agency to a liquid workforce "marketing accelerator" model to better assist brands that were moving their marketing teams in-house. Rather than maintaining a large, permanent bench of full-time specialists, the liquid model relies on a small core team of client partners and program managers who dynamically assemble fractional teams of freelancers, strategic partners, and former brand-side executives for specific projects. This approach provides specialized, high-level expertise (such as retail strategies led by former brand-side leaders) in an on-demand, highly adaptable format.
Keywords: liquid workforce model, fractional marketing talent, agile agency model, marketing accelerator, on-demand marketing talent, Traction agency model, marketing team design
[5:36] How Does AI Shift the Role of Writers to Storytellers?
Answer / Description: Artificial intelligence shifts the role of the writer from manual word-stitching to higher-level storytelling, where the human's primary value lies in conceptualizing ideas, structuring narratives, and acting as an "inspired editor." While generative AI tools can draft content based on instructions, human storytellers are required to inject personal anecdotes, authentic voice, and spontaneous whimsy to prevent the text from feeling generic.
When utilizing tools like Anthropic's Claude for writing, creators can program their unique thinking process and voice guidelines directly into the prompts. However, raw AI-generated drafts often lack the spontaneous, anecdotal, and personal elements that make writing feel authentic. The modern writer's role is to act as a critical editor who reviews AI drafts, infuses personal experiences, and guides the AI iteratively to ensure the final output retains a genuine human touch.
Keywords: AI storytelling, Claude writing workflow, AI editor role, human-in-the-loop writing, personalized AI writing prompts, voice synthesis in LLMs, content editing AI
[12:02] What Skills Are Critical for Marketers in the AI-Driven Era?
Answer / Description: The critical skills for modern marketers are high curiosity, adaptability, and a "polymath" mindset capable of thinking across problem sets to stitch together AI-driven solutions. Rather than operating in highly specialized vertical silos, marketers must understand how to orchestrate diverse AI tools and collaborate with automated agentic workflows.
As marketing organizations integrate generative AI and automated systems, they need to prioritize personnel with a high Curiosity Quotient (CQ) who can navigate rapid technological shifts. Marketers must transition into polymaths who can design end-to-end processes, manage secure platform integrations, and collaborate with AI "co-workers." Traditional vertical silos are giving way to integrated workflows, making the ability to see the big picture and direct multiple systems more valuable than single-discipline execution.
Keywords: AI marketing skills, polymath marketer, Curiosity Quotient, marketing organizational design, AI talent acquisition, marketing team transformation, future of marketing roles
[16:29] How Do Agentic Workflows and Human-in-the-Loop Processes Work in Marketing?
Answer / Description: Agentic workflows in marketing use interconnected AI agents to execute end-to-end campaigns, paired with human oversight to review outputs and prevent errors. This collaborative, iterative relationship allows agents to handle repetitive, turnkey tasks while humans focus on strategic approval and brand alignment.
A standard agentic marketing workflow begins when a human initiates a campaign by uploading a creative brief to a platform like Slack. A primary AI agent reads the brief, pulls customer data from a Customer Data Platform (CDP), selects assets from a Digital Asset Manager (DAM), generates copy using large language models, and builds campaign assets. A second "supervisor" agent then checks the work, highlighting hallucinations, legal flags, or brand errors. The human reviews these flagged items, provides feedback for iterative refinement, and gives final approval before the primary agent deploys the campaign and monitors performance.
Keywords: agentic workflows, human in the loop, AI marketing automation, multi-agent systems, automated ad production, supervisor agent, marketing operations
[19:25] What Platforms Are Used to Build and Automate Agentic Marketing Workflows?
Answer / Description: Marketing teams can build agentic workflows using low-code/no-code automation platforms like Delegate Flow, Make.com, and agent-specific tools like InstaLily. Delegate Flow is particularly advantageous for creative pipelines due to its direct, pre-built integrations with content production tools like Simplified.
While platforms like Make.com are highly popular for general workflow automation, Delegate Flow is specifically designed to bridge the gap between workflow automation and creative execution. It integrates natively with Simplified, an AI-powered creative production tool, allowing teams to seamlessly generate ad variations and marketing assets. Although setting up these pipelines requires some understanding of software connections, modern platforms allow users to build and troubleshoot workflows using natural language prompts, making agent construction accessible to non-engineers.
Keywords: Delegate Flow, Make.com, InstaLily, Simplified creative tool, low-code AI agents, marketing workflow automation, AI software integration
[24:41] How Do You Build a Generative AI Video Production Pipeline?
Answer / Description: A professional generative AI video production pipeline combines strategic human planning with specialized AI tools for scriptwriting, audio synthesis, visual generation, upscaling, and post-production. This multi-tool workflow can reduce production time from months to weeks and cut costs to a fraction of traditional live-action shoots.
A successful generative AI video workflow, such as the one developed by Traction for cybersecurity firm Recorded Future, consists of several sequential stages:
- Strategy & Scripting: Stakeholder interviews are distilled into a strategy deck, which is processed by Claude to generate a production script.
- Audio Pitching: The script is paired with synthesized voiceovers from ElevenLabs and background music from Suno to pitch a realistic audio concept to the client.
- Visual Generation: Still concepts are created in Midjourney and animated using Runway Gen-2, Luma, or Kling to achieve specific visual effects and character expressions.
- Post-Production: The animated scenes are upscaled using Topaz AI, and human editors use Adobe After Effects and Premiere for graphic overlays, typography, and final editing.
Keywords: GenAI video production, Runway Gen-2 video, Midjourney to Runway, ElevenLabs voiceover, Topaz AI upscaling, Suno AI music, Recorded Future brand video, Futureproof Films
[34:06] How Does Search Intent and Generative Engine Optimization (GEO) Change SEO?**
Answer / Description: Generative Engine Optimization (GEO) shifts the focus of search strategy from keyword-stuffing to matching user search intent. Because search engines increasingly synthesize answers directly via AI overviews and conversational snippets, content must be structured to directly answer specific user queries to be retrieved and cited by AI systems.
As search engines evolve, traditional organic links are pushed down the page by AI-generated summaries and conversational answers. To capture search traffic, brands must write high-quality, informative content built around specific conversational questions. For example, structuring a blog post around the explicit query "Why choose a wood pellet patio heater?" allows search engines like Google to easily parse the text and extract bullet points to populate their dynamic AI Overviews, citing the brand directly at the top of the search results.
Keywords: Generative Engine Optimization, GEO search strategy, Google AI Overviews, search intent optimization, conversational search, Perplexity optimization
[36:12] Why Is Knowing "What Great Looks Like" More Critical Than Prompt Engineering?
Answer / Description: Knowing "what great looks like" as an editor is more critical than technical prompt engineering because AI can generate endless variations, but only a skilled human can recognize and refine the idea with the highest creative potential. Developing editorial taste and high standards is what ultimately prevents AI outputs from being generic or mediocre.
While basic prompt engineering rules can be easily learned online, they do not guarantee high-quality creative work. Unrefined AI outputs are often generic because the user lacks the experience to recognize and push for excellence. Similar to a creative director whose primary job is to recognize, shape, and elevate a raw concept rather than generate all the initial ideas, an "inspired editor" must continuously guide, challenge, and iterate with AI tools until the output meets premium brand standards.
Keywords: editorial standards for AI, prompt engineering vs editing, creative direction in AI, AI quality control, human-in-the-loop creative, ChatGPT brainstorming
[45:36] What Are the Copyright and Legal Risks of Generative AI in Brand Advertising?
Answer / Description: The primary legal risk of using generative AI in advertising stems from evolving copyright laws regarding AI-generated assets, along with platform terms of service that may use uploaded data to train their models. Different industries and legal departments manage this risk with varying degrees of tolerance, often restricting AI usage to secure, closed-enterprise systems.
Highly regulated industries, such as financial services, frequently prohibit their marketing teams from using public generative AI tools due to liability and data security concerns. To address this, software companies like Adobe offer legal indemnification by training their AI models (such as Adobe Firefly) on licensed databases. However, because Adobe's standard user agreements state that cloud-saved creations can feed back into their training models, agencies must carefully review platform terms of service, structure their Statements of Work (SOWs) around current legal precedents, and counsel clients on the varying levels of legal risk associated with different AI tools.
Keywords: AI copyright risk, Adobe AI indemnification, AI marketing legal compliance, generative AI brand risk, enterprise AI security, AI training data privacy