CHAPTERS
- 0:00 – 1:33
Meridian’s mission: AI for spreadsheets to make Excel users 20x faster
John Ling opens by describing his curiosity-driven approach to learning and why spreadsheets are a massive, under-optimized workflow for AI. He introduces Meridian and frames Excel as the world’s most widely used “programming language,” where AI can meaningfully accelerate knowledge work.
- •Learning-by-doing mindset: more work equals more learning opportunities
- •Spreadsheets as a universal interface for business logic and decision-making
- •Meridian’s core goal: help spreadsheet users move ~20x faster with AI
- •Observation: few people have seriously invested time using AI to build real finance models
- •Problem framing: AI can handle parts of spreadsheet workflows, but requires investigation and decomposition
- 1:33 – 2:04
Becoming a top performer at ScaleAI through curiosity and cross-functional ownership
John explains how ScaleAI rewarded people who expanded beyond their formal role and actively sought new problems. He emphasizes deliberate learning, especially stepping back to understand “why” before executing.
- •Scale’s culture enabled people to tackle problems beyond their job description
- •Personal growth came from constantly expanding into adjacent domains
- •Execution is important, but understanding first principles prevents wasted work
- •“Do more work” reframed as accelerating learning loops
- •High agency: choosing problems that increase knowledge breadth
- 2:04 – 2:58
Learning AI the hard way: reading research, interrogating data quality, and building intuition
He describes how deep research—especially around data quality—helped him develop stronger judgment in AI. The chapter highlights methodical learning: reading papers, analyzing datasets, and understanding what makes data valuable.
- •Research-driven learning: read papers to clarify what matters and why
- •Data quality as a foundational concept for AI performance
- •Hands-on review across domains to understand what makes examples useful
- •Building judgment by connecting theory (papers) with practice (real data)
- •Avoiding getting lost in execution without understanding objectives
- 2:58 – 3:30
Why investors bet on John: first-principles thinking and the courage to “move mountains”
A voice from John’s network explains why he stood out at Scale: he wasn’t limited by organizational boundaries and consistently pushed for what he believed was right for the business. His edge was pairing conviction with follow-through.
- •Reputation as a top 1% performer validated by multiple colleagues
- •First-principles approach rather than accepting company constraints
- •Willingness to voice opinions to leadership and peers
- •Operational intensity: turning ideas into reality despite friction
- •Signal of founder potential: conviction plus execution
- 3:30 – 4:31
Joining ScaleAI to immerse in LLMs before founding again
John explains his strategic motivation for joining Scale: it was a front-row seat to LLM development and adoption. He viewed deep immersion in the ecosystem as necessary preparation for starting his next company.
- •Scale as an “observation post” for the trajectory of LLM capabilities
- •Belief: ignoring rapid LLM progress is a strategic mistake
- •Intentional career move to learn how models work and how they’re implemented
- •Desire to found again guided decisions about where to learn fastest
- •Focus on understanding where the technology is heading, not just where it is
- 4:31 – 5:02
Operational lessons at Scale: evaluations, benchmark thinking, and internal automation
He details his work ensuring Scale’s data outputs were valuable, including benchmarks and evaluations. He also explored using LLMs to improve internal workflows, reinforcing the idea that AI should directly boost productivity.
- •Quality assurance: ensuring produced data was truly valuable
- •Benchmarking and evaluation as core mechanisms for progress
- •Exploring internal LLM adoption to improve efficiency
- •Learning how organizations integrate LLMs into real processes
- •Seeing productivity gains as a practical path to adoption
- 5:02 – 5:32
The “vibe coding” wake-up call: tools that collapse build times from weeks to minutes
John describes experiencing modern coding copilots (e.g., Cursor) as a step-change in speed and empowerment. This shifted his expectations for what knowledge workers should demand from tools—and helped shape his view of spreadsheets as the next frontier.
- •Recent coding tools made software creation feel dramatically more accessible
- •Zero-to-one prototyping time compressed from weeks to hours/minutes
- •Cultural push: teams that don’t adopt these tools risk falling behind
- •AI framed as a “super calculator” that changes capability boundaries
- •Desire to bring similar leverage to non-developer workflows
- 5:32 – 6:02
SF vs. NYC AI adoption gap: why finance workflows lag behind coding workflows
John contrasts the excitement and rapid tool adoption in San Francisco with slower uptake among New York finance circles. He argues that coding tools succeed partly because builders are power users; finance/spreadsheet AI lacks that same tight feedback loop.
- •Perceived cultural/market difference in AI enthusiasm and experimentation
- •Finance communities still rely on manual spreadsheet work patterns
- •Coding tool builders are often end users, clarifying what “good” looks like
- •Spreadsheet/finance AI struggles because failure modes are harder to diagnose
- •Opportunity: build tools for users who can’t easily articulate model errors
- 6:02 – 7:33
The Meridian insight: nobody has spent 1,000 hours doing finance modeling with AI
He introduces a key belief: serious, repeated experimentation with AI in finance/spreadsheets is rare. Because bankers build by hand and many users can’t pinpoint why numbers are wrong, there’s massive room to decompose spreadsheet workflows and apply models effectively.
- •Claim: almost no one has done sustained, intensive AI-driven financial modeling practice
- •Manual-by-default mindset in banking creates inertia
- •AI can likely excel at components of modeling workflows if decomposed correctly
- •Non-experts struggle to debug outputs, slowing adoption and trust
- •This gap became the catalyst: “we should go solve this problem”
- 7:33 – 8:33
Bias toward action: reach out, test assumptions, and suspend disbelief
John describes how sales taught him that doors don’t open unless you knock. He encourages proactive outreach and a mindset of treating ambitious goals as doable until proven otherwise.
- •Sales lesson: if you don’t ask, the answer is always no
- •Outreach isn’t futile—surprising responses happen when you try
- •Most “impossible” narratives are untested assumptions
- •Entrepreneurship requires suspension of disbelief and experimentation
- •Learning comes from attempting unfamiliar tasks and iterating on failures
- 8:33 – 9:19
Building a high-agency culture: safe-to-fail experimentation with team support
He explains how Meridian aims to create an environment where people can try solving problems their own way, fail, and receive support. The focus is on building something meaningful and enduring rather than optimizing for short-term perfection.
- •Cultural principle: experimentation is encouraged; failure is acceptable
- •When someone fails, the team supports them and adjusts timelines pragmatically
- •Empowerment: letting people own problem-solving approaches builds capability
- •Long-term ambition: build a “masterpiece” company with real impact
- •Craft and pride as motivators beyond speed and growth
- 9:19 – 10:10
Market thesis: spreadsheets as the biggest knowledge-work category ripe for AI augmentation
An investor/operator perspective argues that spreadsheet work is one of the largest segments of knowledge work and software spend. Meridian is positioned to do for spreadsheet users what coding copilots did for developers—injecting intelligence and automation into everyday work.
- •Excel/spreadsheets represent an enormous market and time sink
- •Firsthand pain: banking/consulting spreadsheet workflows are intensely manual
- •AI can fundamentally change how spreadsheet work gets done
- •Analogy: developer augmentation is happening; spreadsheet augmentation is next
- •Meridian’s vision: meaningful automation embedded into spreadsheet workflows
- 10:10 – 11:58
Spend 10,000 hours with AI: prompting skill, task clarity, and building future intuition
John closes with advice: sustained time with AI builds intuition for what’s possible now and soon. He highlights prompting as a durable skill because explaining tasks precisely forces clearer thinking—similar to how writing a YC application clarifies a startup.
- •Time-with-tech creates intuition about current capability limits
- •Sustained practice also predicts near-future capability improvements
- •“10,000 hours with AI” as a personal unfair advantage, especially within a domain
- •Prompting remains valuable because specificity improves outcomes
- •Explaining tasks to an LLM clarifies your own thinking and intent
