Marketing Against The Grain

Episode: Perplexity Computer: The Super Agent Playbook (5 Real Workflows)
Date: March 5, 2026
Hosts: Kipp Bodnar (HubSpot CMO), Kieran Flanagan (SVP of Marketing)

Episode Overview

In this episode, Kipp and Kieran take a deep dive into the emerging world of “super agents” in AI, focusing on Perplexity Computer. They break down how this new tool enables marketers (and anyone) to automate complex workflows, build skills, and connect multiple tools and data sources in creative ways. The conversation pivots around the automation of marketing processes using advanced AI agents, the trend toward model-agnostic platforms, and how to structure the adoption of these tools for maximal impact. Real-world use cases are provided throughout, including actual workflows the hosts built and tested.

Key Discussion Points & Insights

The Age of Super Agents in AI

[01:45]

• Convergence of Tools:
  Most leading AI companies are converging on the “super agent” model—AI agents with skills (what they can do) and connectors (what tools and data they can access).
• Main Differentiator:
  With so many similar tools, the real differentiator is where you run your workflows and how you integrate them into your daily habits.
• Key Players:
  • Claude Code/Cowork
  • Perplexity Computer
  • Manus
  • OpenClaw (noting its community momentum and quirky, fun branding)
• Quote:

  "Every single thing is converging on, hey, we’re going to be a super agent and you give us tools. And tools are things like, hey, you can access my email, you can access YouTube...they all kind of are converging on, hey, we’re a super agent." – Kieran [01:45]

What Makes Perplexity Computer Stand Out

[05:27, 08:22]

• Model Agnostic:
  Can leverage models from Google, OpenAI, Anthropic (Claude), etc.
• Agentic “Teamwork”:
  Launches parallel sub-agents for batch processing (e.g., analyzing 100 book covers at once).
• Disposable Web:
  Rapid creation of single-use, live interactive websites for tasks like data visualization or feedback review.
• Ease vs. Control:
  Easier than technical tools like Claude Code, but still offers significant depth and control for customization.
• Quote:

  "Let me launch four parallel sub-agents, each analyzing 25 book covers in depth—which is pretty wild." – Kipp [08:15]

Real-World Marketing Workflows & Use Cases

1. Designing a Book Cover – Reverse Engineering Success
[05:27–09:29]

• Workflow:
  Tasked Perplexity Computer to study top 100 business book covers, extract design principles, and generate concepts for a new marketing AI book.
  • Crawling bestseller lists
  • Extracting and analyzing images in parallel
  • Delivering a reusable skill MD file for internal iteration
  • Quickly building a showcase website of the results for team feedback
• Memorable Moment:

  "These are really good mocks. I was really impressed." – Kieran [09:26]

2. Live Political Betting Market Dashboard
[11:20–12:41]

• Workflow:
  Used Perplexity to build a live interactive website with real-time US political prediction data from Polymarket.
• Expansion Potential:
  Use for sports, industry news, or any domain with live public betting odds.
• Quote:

  "...You can basically create a website all around the bets being made in polymarket on that topic." – Kieran [12:11]
  "I would do a separate rev of this where you put a confidence score on each of these." – Kipp [12:28]

3. AI-Powered Salary & Job Offer Negotiation
[12:41–13:43]

• Workflow:
  Perplexity Computer crafts a detailed counter-offer for a Stripe product manager candidate, including market research on salaries and RSU packages.
• Quote:

  "...It's really hard to deal with humans anymore. Like negotiation because every human is so well informed, right? Like negotiation." – Kieran [13:14]

4. Marketing Turing Test: Reverse Engineering Growth Tactics
[13:43–16:20]

• Workflow:
  Gathers data on fastest-growing, non-obvious companies and infers their most potent marketing strategies.
  • Patterns identified:
    • PLG (Product-Led Growth)
    • Programmatic SEO
    • User-generated content
    • Adjacent category SEO
    • Product loops
    • Community-driven distribution
    • Low ad spend
• Quote:

  "I'm trying to create a mechanism to discover new hacks to grow, right?" – Kipp [14:16]

5. Auditing HubSpot’s Product Marketing
[16:20–17:47]

• Workflow:
  • Perplexity Computer builds a skill to evaluate product marketing across HubSpot’s site.
  • Crawls all product pages, automatically audits by set criteria
  • Pinpoints underperforming features and areas for improvement
• Quote:

  "You can have incredible insight into the work you're doing and your team's doing without having to spend like literally your entire weekend going through stuff manually." – Kipp [17:47]

Best Practices: Integrating Super Agent Workflows

[19:12–23:53]

• Key Advice:
  • Don’t just build—use and iterate. Integration into actual workflows is the major challenge now, not access to powerful tools.
  • Build one workflow at a time. Avoid having too many unused or redundant skills. Focus on refining and integrating new skills deeply into your routine.
  • Iterative improvement: Make skills great through repeated use and feedback. It’s not ready until a task can be completed with minimal/no edits.
• Quote:

  "You have to use those things too, Kieran, because you have to iterate on them and make them better. ...That skill's not good enough until you can run it on a new product and basically have no changes." – Kipp [22:09]
  "AI slop was one thing. AI clutter is the new thing." – Kipp [23:48]

The State of AI Tools in 2026

[18:14–19:12]

• Tool Overload:
  We now have more powerful software than most people can use. The hard part is changing habits and processes to make full use of this power.
• Quote:

  "We have way more product than the world could use at this point." – Kipp [18:43]

Notable Quotes & Moments

• "Marketing is going to be the defacto skill to have because you're going to have all of these super agents, they do very similar things... OpenClaw... is the one that has caught the imagination of people." – Kieran [05:02]
• "It's never been a better time to be a creative person with taste, right? Never." – Kipp [05:27]
• "I still think a lot of it comes back to… the real value will become: I'll use this repeatedly and iterate and make it better with the AI." – Kieran [23:05]

Useful Timestamps for Key Segments

• 01:45 – Trend recap: Rise of super agents, what distinguishes them
• 05:27 – Perplexity Computer’s unique position and model-agnostic power
• 08:12 – Book cover workflow example and parallel sub-agent action
• 11:20 – Use cases: Politics dashboard, job offer negotiation
• 13:43 – “Marketing Turing Test” to discover emerging growth hacks
• 16:20 – Product marketing audit automation for HubSpot
• 19:12 – Workflow integration challenges vs. tool access
• 21:46 – Best practices for adopting AI into work habits

Final Thoughts & Recommendations

• Choose One Platform: Focus deep learning/iteration on one super agent platform rather than dabbling in many.
• Workflow Discipline: Document and narrate your actual work to create repeatable skills and workflows.
• Continuous Refinement: The first version is never best—aim for repeated use and consistent improvement.
• Avoid AI Clutter: Don’t let “dopamine hit” of building new skills replace the discipline of actually using and evolving them in meaningful ways.

"AI clutter is the new thing... It's a dopamine hit. I made that. No one's actually using it, and I'm not using it." – Kieran [23:53]

Resource Mention:
The episode’s product marketing audit skill is available free—find the download link in the episode description.

Overall Tone:
Energetic, curious, at times irreverent—but grounded in practical advice for real marketing teams navigating the rapid evolution of AI. The hosts strongly advocate for thoughtful, hands-on adoption and stress the importance of quality over quantity.