27 January 2026

This AI Tool Works Like a $300,000 McKinsey Consultant

2Frameworks

9Insights

Frameworks in this episode

Marketing6 steps

Competitor Advertising Budget Estimation

Estimate competitor ad spend by combining traffic data with channel cost benchmarks.

Read →

Marketing6 steps

Contra-Seasonal Advertising

Shift spend into competitors' quiet months to capture lower-cost attention.

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Hot Take· 2

Hot Take03:00

A $10 Research Query Can Still Deliver Thousands in Value

The integration consumes substantially more credits than a routine model prompt, with some research requests costing several dollars or more. The hosts argue that this is expensive only within the low-cost world of AI prompts; compared with conventional research and consulting, the value-to-cost ratio remains compelling.

Research-intensive agent queries consume meaningful credits.
Users can see the projected credit cost before proceeding.
The relevant comparison is professional research, not a basic chat prompt.
A modest query cost can replace hours of manual analysis.

“I think it's you're getting thousands of dollars of value, but you need to know like one of these queries is probably like 5, 10,…”
03:00

“It's only expensive in AI research terms.”
09:00

#ai economics#research cost#consulting#credits

Hot Take19:00

The Best AI Is the One with the Most Relevant Context

The hosts argue that raw model intelligence matters less than access to the right domain context. Manas becomes especially useful for marketers when SimilarWeb's proprietary traffic information is available alongside general web research, reasoning, and content-generation capabilities.

Individual models remain specialized rather than universally best.
Relevant proprietary data can matter more than marginal model intelligence.
Aggregator agents can combine context with specialized models.
Domain context turns generic capability into a high-value workflow.

“the best AI is not the smartest AI, it's the AI that has the most context.”
19:30

“we've all this SimilarWeb context in Manas now and there's bunch of cool stuff we can do for marketing.”
19:30

#context engineering#ai agents#aggregators#marketing data

Explainer· 2

Explainer01:30

What SimilarWeb Data Manas Can Analyze

The hosts inventory the marketing intelligence available through the integration, including traffic, audience, source, geography, ranking, bounce-rate, and YouTube data. Combining those structured metrics with external web research gives the agent enough context to analyze competitors rather than merely summarize public pages.

Available metrics include total and unique visits.
Traffic can be broken down by geography and source.
Bounce rate and global ranking add comparative context.
The agent can blend SimilarWeb metrics with external web information.

“You can get traffic metrics, total visits, unique visitors, bounce rate.”
02:30

“you can actually blend in the SimilarWeb data.”
04:30

#similarweb#manas#market research#traffic data

Explainer20:00

Learn Growth Plays from the Fastest-Growing SaaS Websites

The agent identifies SaaS websites by traffic growth rate rather than total traffic, then investigates how the winners acquire users. The output highlights patterns such as AI-powered tools, consumer-facing SaaS, product-led growth, search acquisition, and paid channels while also supplying the underlying data as a CSV.

Rank companies by growth rate rather than absolute size.
Segment the search by industry when a global ranking is unavailable.
Investigate acquisition mechanisms behind the growth.
Export the data for further analysis.
Translate company-specific findings into an actionable playbook.

“just give me the top 10 SaaS websites by Traffic Graph.”
20:30

“this is not by total visits, this is by growth rate.”
20:30

#saas#growth research#traffic growth#acquisition

Story· 1

Story21:30

The Agent Turned Growth Research into Three Coded Homepages

After researching successful SaaS companies and their growth playbooks, the hosts ask Manas to apply the lessons directly to HubSpot. It produces three homepage concepts with copy, customer proof, feature callouts, graphics, and interactive behavior, demonstrating how research can move immediately into a shareable implementation.

The designs are based on observed growth lessons.
The agent generates both messaging and visual structure.
The output includes coded interactions such as hover behavior.
Generated pages can be published and shared immediately.

“I had Manas go and break down three different homepage designs for hubspot.com based on these lessons and it mocked them all up.”
22:00

“you can then go and publish this to share it with people immediately.”
23:00

#web design#prototyping#homepage#generative ai

Tool· 2

Tool06:00

Convert a Dense Market Report into a Visual Deck with NotebookLM

A detailed HubSpot competitive-intelligence PDF is uploaded to NotebookLM and converted into a presentation without manual slide production. The resulting deck organizes market position, traffic strategy, audience profile, competitive gaps, and recommendations into an executive-friendly narrative that can serve as an editable first draft.

Create a notebook containing the finished research report.
Ask NotebookLM to build a presentation from the source.
Use the deck as a first draft rather than unquestioned analysis.
The generated narrative can surface strategic recommendations and investor framing.

“I simply just create a notebook with this report and ask it to create a presentation.”
06:30

“So it's how quickly you can get from a single prompt to a first draft that you can start to like iterate and improve upon.”
09:00

#notebooklm#presentations#competitive intelligence#executive communication

Tool16:30

Generate a Website Competitive Analysis and 90-Day Plan

The hosts compare hubspot.com with salesforce.com using traffic scale, bounce rate, ranking, referrals, and paid advertising signals. The agent converts the findings into a slide deck and proposes a 90-day action plan, despite receiving little guidance about format or required sections.

State the business objective behind the comparison.
Compare scale and engagement instead of traffic alone.
Ask for strengths and areas for improvement.
Convert findings into a prioritized implementation plan.

“using data from SimilarWeb, do a competitive analysis hubspot.com and salesforce.com. Please highlight strengths and and areas for improvement.”
16:30

“It gives you a 90-day action plan, Q&A.”
18:30

#website optimization#competitive analysis#slide deck#action plan

Takeaway· 2

Takeaway11:30

Domain Expertise Determines the Quality of AI Follow-Ups

The hosts find the agent's default follow-up questions too generic to uncover meaningful marketing advantages. A knowledgeable practitioner can instead ask about channel-level bounce rates, competitive arbitrage, or another mechanism tied to an actual decision, making the same tool far more valuable.

Suggested follow-ups often produce superficial comparisons.
Experts know which variables can reveal an actionable advantage.
Clear practitioner-generated questions outperform agent-generated prompts.
AI appears to amplify strong marketers more than replace expertise.

“My biggest hot take here is that most of suggested follow-ups for marketing aren't that good.”
12:00

“it feels like it 10x's the best marketers is one of my big takeaways.”
12:30

#domain expertise#prompting#marketing strategy#ai copilots

Takeaway24:30

All-in-One Agents Reduce the Gap Between Research and Building

The episode closes by emphasizing agents that coordinate multiple specialized models inside one continuing thread. This lets a marketer move from research to visualization, design, and functional implementation without repeatedly exporting context or choosing a separate model for every stage.

Aggregator agents can select specialized models by task and cost.
A persistent thread preserves context between modalities.
Research can flow directly into design and implementation.
The best model can be used for each stage without fragmenting the workflow.

“it allows you to switch type modalities there where you can go from research to design to building all in one thread which is really,…”
24:30

“you're using the best model for the job.”
25:00

#multimodal ai#agent workflows#model routing#marketing tools