RIP, BI (The Death and Rebirth of Thinking in the AI Age)
Traditional Business Intelligence is Dead, Thanks to AI. Analysts Who Evolve Will Thrive.
The work of 300 million1 knowledge workers has undergone a seismic shift, a fundamental reordering of how we interact with information, thanks to the latest AI reasoning models. Human-designed, handcrafted reports and data visualizations now feel like relics from a bygone era, akin to a farmer plowing a massive field with bare hands and a hoe instead of a John Deere tractor.
Making such a dramatic claim about one of the most important jobs on the planet —thinking —is heartbreaking. It’s also exciting, because thinking isn’t going away, it’s just changing — shaken to its very core.
For context, I’m passionate about analyst work for several reasons: I was the general manager of a pioneering business intelligence (BI) tool, Spotfire (below), the kind of tool millions of analysts use every day. I serve as an advisor for the Data Visualization Society (DVS), one of the largest networks of visual storytellers. Finally, I’m an analyst, so I get paid to think.
I’ll be speaking at this year’s DVS Outlier conference — my presentation is titled “RIP, BI: The Death and Rebirth of Thinking.” Although that sounds like a dark take, I’m decidedly optimistic about the future of analysts.
Here’s what I’m going to say.
What is an Analyst, Anyway?
Before exploring the demise and rebirth of a billion-dollar industry, millions of jobs, and thinking itself, let’s start with what analysts do and what's at stake.
Analysts are truth sleuths. They formulate questions, set context, then collect, process, and analyze data to uncover insights.2 Then, they craft stories about what they found. If they did a good job, they inspire others to act, or, at least, view the world through a lens with a new tint.
AI fundamentally improves what analysts do, and not just by a little bit — it can help you think between 480 and 6,000 times faster and better.
Those numbers sound hyperbolic? Let me explain my math with a story.
Deep AI Analysis in the TSA Precheck Line at LaGuardia Airport
I was flying out of LaGuardia Airport. During my cab ride from Manhattan, I developed a 700-word prompt for a product idea. I got into the TSA Precheck line and let my prompt fly.
As I shuffled through the serpentine security line, AI went to work. Before going to work, the AI reasoning model I used (Manus) asked questions to clarify my intent. After some back-and-forth, it dove deep. It “read” five books, found and analyzed patents, grabbed quotes from related blog posts, and formulated a detailed report, all before I reached my gate (which, in the new LaGuardia, takes just a few minutes!)
I didn’t have much time before my flight, so I asked AI to flip its findings into six new forms:
A long, detailed summary, including a SWOT3 analysis, with links and patent references,
A short 1-page executive summary, written in the style of William Zinsser (one of my favorite writers who advocates concise, direct language),
A list of 20 non-obvious, counterintuitive ideas AI uncovered,
20 Ernest Hemingway six-word stories that summarized the big ideas,
A haiku,
A NotebookLM-generated podcast, based on the detailed report.
I downloaded these to my laptop and boarded my two-hour flight.
The results, as anyone who uses the latest AI reasoning models knows, were stunning. The multiple transmission modes of ideas sparked me to find new insights, aha moments, and new questions.
I sketched some ideas in my paper journal (yup—good old-fashioned paper and a multi-color erasable Frixion pen), wrote a few additional prompts for AI, and closed my computer.
I took a nap.
Insights After a Nap: Thinking, 480X Faster and Better
I woke up with a sudden insight: in three hours (including the nap), AI and I had already produced research and seeds of ideas that would have taken months to accomplish on my own, including three truly novel ideas. I did the rough math: my AI-aided thinking process was 480 times faster (three hours versus two months) and, qualitatively, better than I could have done on my own.
The most novel findings, as it turned out, were hallucinations. AI had “synthesized” three ideas from two patents and a blog post — it made them up.
Some fear AI hallucinations; I devour them. I read the sources and found one idea to be fascinating, including one PhD paper that I struggled to understand — I had to ask AI to reprocess and reinterpret it. Ironically, AI’s most valuable ideas come from creative "mistakes"—connections that AI synthesized. Just like humans.
Again, it struck me: AI far outstripped my ability to connect these hidden dots to paint a new, novel picture.
Telling the Story
The final work of an analyst is storytelling, which is how we make an impact on the world, by changing minds.
Again, I partnered with AI. I prompted AI to convey my final ideas to three audiences: a CEO, an engineer, and a board member. For the CEO, that meant a concise executive summary. For the engineer, the technical details. For the board, a strategic view of how the ideas fit into the broader market, impact competitive differentiation, and associated risks. I used Midjourney to create imagery that explained key ideas with visual analogies. I generated over 200 headline ideas and landed on one that I liked.
After about five hours of editing, the result was professional and share-worthy. The final reports were entirely my own, supercharged by AI.
That’s when the analogy of an AI-powered analyst leaping over an old-school BI user came to mind. I asked Midjourney to generate an image of an analyst jumping over a rival analyst, sitting at her desk, crafting data dashboards by hand like a farmer with a hoe in the 1800s:
Why AI is a John Deere Tractor for Analysts
My LaGuardia experience wasn't just a personal anecdote—it revealed three fundamental principles about the transformative power of AI. First, AI demolishes the barrier between the question-asker and the answer. As Einstein said, questions are often more important than answers, and AI transfers the power of question-asking to domain experts.
Second, AI is overwhelmingly more powerful than we are in “dot connection.” For example, the latest AI models are 6,000 times more potent than using Google for conducting manual research. (6,000:1 is the ratio of responses to redirects performed by Anthropic. Read this4, this5, and listen to this6 for an explanation of this idea.) And AI doesn’t tire, take vacations, is cheap, and can simultaneously approach a problem from many angles.
Finally, AI supercharges the art of storytelling—reports targeting multiple personas, using new artwork, and poems. Even hearing your ideas, though jaw-droppingly awesome AI-generated podcasts via NotebookML (Want to check it out? Listen to this AI-generated podcast here or anywhere you listen to podcasts. Literally, all I did was load the text of this article, give it a simple 30-word prompt, and publish it to Substack.)
But what makes AI so valuable creates an existential paradox. The insight gatekeeper role, however well-intentioned and skillfully executed, is fundamentally reordered. This isn't a threat to intelligence itself, but rather a profound shift in where, how, and when that intelligence is revealed, accessed, and applied, and to whom.
Six Steps to Evolve as an AI Age Thinker
Analysts who fail to adopt a new critical thinking approach put their passion to make a difference at risk. This disruption, however, isn't merely a threat to be survived—it's a once-in-a-generation opportunity for transformation.
Here are six ideas to help you not just survive, but thrive as an AI age thinker.
Ask Yourself: “What Makes Me a Special Snowflake?”
Self-knowledge first: Know thyself. Knowing what makes you special forms the foundation for all other steps. Without understanding your unique value, you can’t select tools, craft prompts, judge AI outputs, or create an AI style. Just as being able to play a trumpet doesn’t make you Miles Davis, AI is of little use if you don’t use it to express what makes you special.
So the first step to becoming an AI-savvy thinker has nothing to do with AI. Simply take time to reflect on what makes you truly unique. What wrongs do you want to right in the world? What do you have to say?
What makes me unique? I’m obsessed with how technology meets what it means to be human— how it augments (and, at times, replaces) human cognitive abilities. I’m a tech geek steeped in neuroscience and behavioral economics.
My geek-meets-tech POV helps me pose novel questions to AI and filter its output in ways others might miss.
Got what makes you special in mind? Now let’s get to the AI.
Create a Personal AI Backpack of Tools
Understanding your snowflakeness is essential, but it's not enough—many AI tools obscure rather than amplify what makes you special. The good (and daunting) news is that there are thousands to choose from.
Industry analysts7 8 9 10 estimate that over 10,000 rounds of financing went to AI companies in the last two years. That means there are thousands of AI trumpets for you to play (Miles Davis used just three11!)
Finding AI tools that fit you is a task in and of itself. Test new tools regularly. Curate the ones you like best. As you do, you fill your backpack and build AI muscle. I try a few a month. One of my favorite ways to test AI tools is to try the same prompt on multiple LLMs. You’ll see vastly different results and learn how they work.
Eventually, you’ll cultivate a personal portfolio of tools that align with your unique perspective, workflow, and style of thinking. Currently, my list includes Manus, Perplexity, Midjourney, Otter AI, RunwayML, Grammarly, Dust, and Cursor, but it changes frequently.
Find Your AI Style (Your $20,000 Prompt)
Knowing your tools isn't enough—there's a qualitative leap from playing notes to playing them with artistry. As you work with tools, your AI style will emerge.
After three years, my AI style includes rewriting stories in the style of Haruki Murakami to uncover new analogies and metaphors; using the “But Therefore” rule used by Matt Stone and Trey Parker, the creators of South Park, to check story arc transitions; reimagining visual metaphors in the style of Joan Miró, Paul Rand, David Hockney, Da Vinci, or Jean-Michel Basquiat (The Midjourney-generated image above came from a simple prompt, “/imagine a trumpet in the style of Jean-Michel Basquiat.”)
On LinkedIn, Hamna Aslam Kahn shared this12 prompt from a business consultant that she calls “The $20,000 Prompt” because it embeds the wisdom of a $20,000 consultant, with 41 parts:
If you continually push AI with new angles and prompts, you’ll eventually develop a signature style that produces results only you could have envisioned. Maybe your prompts will be “worth $20,000.”
Edit, Viciously (Don’t Be an Idea Mule)
Kahn has it just half right: a $20,000, 41-part prompt is only valuable if you know how to handle its output. So, AI’s overwhelming “thinking” power creates a paradox: You must edit. Viciously.
Unfortunately, AI is so good that most people skip the best part — the thinking part. Research at MIT found that 68% of participants submitted AI’s output without modification in response to a complicated critical thinking task. That’s not thinking, that’s just carrying AI on your back like a mule (apologies to mules—they’re actually very smart13), adding no value.
Never, ever use or share the output of AI without transforming it into original thinking, adding your unique elements, and taking pride in your work, not the machine’s. I lost count, but I’d hazard a guess that I prompted five different AI tools over 500 times while writing this article, but every word is mine. Every. Single. Word.
So, as William Zinsser said, edit like your readers have better things to do. They can give ChatGPT a banal and boring prompt, too. Put in the work.
Employ Multimodal Thinking
How can you use AI to edit more “viciously?” Think multimodal.
In The Creative Act, Rick Rubin (he produced Johnny Cash, Adele, Tom Petty, The Beastie Boys, and many, many more great artists) wrote: "To vary your inspiration, vary your inputs. Turn the sound off to watch a film, listen to the same song on repeat, read only the first word of each sentence in a short story, or arrange stones by size or color. Break habits. Look for differences. Notice connections." AI helps you work like Rubin in new, fun, analytical, and poetic ways.
I call this “multimodal thinking:” reconsidering ideas through multiple forms, formats, and even sensory channels. My LaGuardia experience showcased six modes of AI-generated input: a deep research report, a poem, six-word stories, my physical paper journal, and a podcast (auditory).
Then, as Rubin suggests, be an antenna: “We are all antennae for creative thought. Some transmissions come on strong, others are more faint. If your antenna isn't sensitively tuned, you're likely to lose the data in the noise. Particularly since the signals coming through are often more subtle than the content we collect through sensory awareness. They are energetic more than tactile, intuitively perceived more than consciously recorded.”
So, think, feel, and listen in a multimodal fashion.
Become an AI Ethnographer
Multimodal thinking alone can lead to scattered insights without a deeper framework for understanding. So, finally, become an AI ethnographer. An ethnographer is a social scientist who immerses themselves in a community to study and understand its culture, behaviors, and social interactions. Do this with AI.
In a way, AI is like a camera14. The better you understand your lenses, lighting, aperture, shutter speed, and their numerous settings, the more effectively you can express yourself.
But AI has infinite settings. So think deeply about how they work and build a second brain with this new, voluminous input. When AI generates a novel idea, have it cite its sources and read them. Journal about them — writing, as Isaac Asimov said, is thinking with your fingers. Using AI’s output without digestion isn’t cheating the world; it’s cheating yourself.
Developing a nuanced, ground-level understanding of AI helps you not only think better yourself, but also become more valuable as an analyst: a better team member, design thinker, and advocate for approaches that leverage machine intelligence and human ingenuity. This isn't just about using AI; it's about shaping its ethical and practical integration and use as an analytical tool by all.
It’s Not RIP, BI: It’s the Evolution of Thought
AI hasn't just arrived; it's a revolution that’s dismantling old ways of accessing and understanding information. Yes, it’s an extinction event for banal analyst work, much like the replacement of the garden hoe for large-scale farming.
But despite this existential threat, something unexpected emerges: Those who embrace what makes them a special snowflake, employ multimodal thinking, and supercharge their signature style will discover that AI doesn’t just alter their jobs—it elevates them, removing routine drudgery and unlocking space for more creative, strategic, and fulfilling work that matters.
So, RIP doesn’t stand for “Rest in Peace,” it’s a new call to evolve and “Rise Into a New Place” — a transformative thinking force that can make you 480 to 6,000 times better, faster, and, ultimately, have more fun.
Come to Miami from June 11th to the 13th for the Outlier conference — I’ll be there, talking about the death and rebirth of analysts, and thinking.
The global analyst workforce comprises approximately 2.9-3.1 million professionals across business, financial, and data analysis roles within a total global workforce of 3.627 billion workers. These analysts are distributed unevenly, with higher concentrations in North America and Europe, while the Asia-Pacific regions show the fastest growth. Based on a 25-26% adoption rate of BI tools among knowledge workers and an estimated 70-85% adoption rate among analysts, we can extrapolate that approximately 2.0-2.6 million analysts worldwide regularly use BI tools. Extending this to the broader global workforce, with knowledge workers representing roughly 35-40% in various capacities, an estimated 300-350 million workers globally interact with business intelligence tools at some level.
U.S. Bureau of Labor Statistics. (2024). *Market Research Analysts: Occupational Outlook Handbook*. Retrieved May 7, 2025, from https://www.bls.gov/ooh/business-and-financial/market-research-analysts.htm
SWOT: Strengths, Weaknesses, Opportunities and Threats
According to Matthew Prince, CEO of Cloudflare, there is a dramatic shift in how AI systems interact with web content compared to traditional search engines. While Google previously crawled 6 pages for every 1 visitor it sent to websites (as of six months ago), that ratio has now increased to 15:1. More strikingly, OpenAI (creator of ChatGPT) has a ratio of 250:1, and Anthropic has an astonishing ratio of 6,000:1. This means Anthropic's AI reads or processes 6,000 pieces of web content for every single time it directs a user to an original source, representing a fundamental shift in how value is exchanged on the internet.
Podcast: "Hard Fork" episode titled "Is ChatGPT the Last Website? Grok's System Prompt, Meta's AI Glasses". The hosts discuss statements made by Matthew Prince about the changing nature of web traffic and AI's impact on publishers.
Fortune article: "Can this tech billionaire save the media from an AI apocalypse?" by Alyson Shontell, published February 4, 2025 (https://fortune.com/2025/02/04/matthew-prince-ai-audit-block-media/ ). This article contains the verified direct quote from Matthew Prince about Anthropic's 6,000:1 crawl-to-referral ratio and explains Cloudflare's methodology for calculating these figures.
CB Insights - State of Venture Q1'25 Report: https://www.cbinsights.com/research/report/venture-trends-q1-2025/
Crunchbase - North American Startup Funding Report: https://news.crunchbase.com/venture/north-american-startup-funding-ai-eoy-2024/
TechCrunch - AI Investments Report https://techcrunch.com/2025/02/11/ai-investments-surged-62-to-110-billion-in-2024-while-startup-funding-overall-declined-12-says-dealroom/
Miles Davis played several types of trumpets throughout his career, most famously the Martin Committee and his custom “Moon and Stars” models, which became iconic in jazz history: https://www.trumpetwarmup.com/post/what-makes-miles-davis-trumpet-playing-so-outstanding?utm_source=perplexity
Here is the full “$20,000 Prompt of a company growth expert.":
<instructions> You are a top-tier strategy consultant with deep expertise in competitive analysis, growth loops, pricing, and unit-economics-driven product strategy. If information is unavailable, state that explicitly. </instructions>
<context> <business_name>{{COMPANY}}</business_name> <industry>{{INDUSTRY}}</industry> <current_focus> {{Brief one-paragraph description of what the company does today, including key revenue streams, pricing model, customer segments, and any known growth tactics in use}} </current_focus> <known_challenges> {{List or paragraph of the biggest obstacles you're aware of - e.g., slowing user growth, rising CAC, regulatory pressure}} </known_challenges> </context>
<task> 1. Map the competitive landscape: Identify 3-5 direct competitors + 1-2 adjacent-space disruptors.. Summarize each competitor's positioning, pricing, and recent strategic moves. 2. Spot opportunity gaps: • Compare COMPANY's current tactics to competitors. Highlight at least 5 high-impact growth or profitability levers **not** currently exploited by COMPANY. 3. Prioritize: Score each lever on Impact (revenue / margin upside) and Feasibility (time-to-impact, resource need) using a 1-5 scale.. Recommend the top 3 actions with the strongest Impact x Feasibility. </task>
<approach> - Go VERY deep. Research far more than you normally would. Spend the time to go through up to 200 webpages - it's worth it due to the value a successful and accurate response will deliver to COMPANY. - Don't just look at articles, forums, etc. - anything is fair game... COMPANY/competitor websites, analytics platforms, etc. </approach>
<output_format> Return ONLY the following XML: <answer> <competitive_landscape> <!-- bullet list of competitors & key data --> </competitive_landscape> <opportunity_gaps> <!-- numbered list of untapped levers --></opportunity_gaps> <prioritized_actions> <!-- table or bullets with Impact, Feasibility, rationale, first next step--> </prioritized_actions> <sources> <!-- numbered list of URLS or publication titles --> </source> </answer> </output_format>"
My apologies to mules, who are actually really smart: they demonstrate exceptional problem-solving abilities that surpass both horses and donkeys, with research from finding their cognitive skills comparable to dogs. Their superior problem-solving capability is supported by research from the Donkey Sanctuary in the UK, where Kristin Hayday suggests mules should be recognized as "smart instead of stubborn," challenging traditional perceptions of these highly intelligent hybrid animals.
Palmer, M. (n.d.). *Techno Sapien*. Substack. Retrieved May 7, 2025, from https://technosapien.substack.com
Inspirational. Fantastic. 👏
Agree totally - though I don't think this is just limited to BI or Analysts. I think we must all "re-program" ourselves and evolve the way we think to be more effective and productive. I'm joining Allie Miller's AI session on this on Thursday and am looking forward to it. Most professionals are using AI as a search engine or productivity tool, but there's much more on how to truly be "AI-First". The future's gonna be wild!