This Founder is Making 1B+ Excel Workers 20x Faster | Meridian, John Ling
In a Nutshell
John Ling, co-founder and CEO of Meridian, is building AI tools for spreadsheets to make Excel users 20x faster, viewing Excel as the world's most distributed programming language. After excelling at Scale AI by deeply studying LLMs and data quality, he identified a gap in finance: no one has invested 1,000 hours using AI to automate manual tasks like LBO models, which bankers build by hand. Meridian, backed by a $15M Series A from a16z, fosters a culture of bold experimentation, rapid outreach, and hands-on AI intuition-building to transform knowledge work.
These notes were generated by AI and may contain inaccuracies.
Enjoyed learning new things; more work meant more opportunities to learn. Faced 50 problems, each teaching something different. Approach: try it, and if you fail, that's okay. No one on the planet has spent a thousand hours trying to build financial models with AI. Experiment: construct AI to build an LBO model that would otherwise be done manually for work. Bankers do it by hand; if you don't know how, you just don't do it.
Think there's a big decomposition where models can handle different parts of the workflow very well, but it requires investigation.
My name is John, co-founder and CEO of Meridian. Building AI for spreadsheets. View Microsoft Excel as the most distributed programming language in the world. Goal: help people who spend a lot of time in spreadsheet software move 20 times faster.
Prior: 1.5 years at Scale AI. Before that, started a couple companies. Raised slightly more than $15 million; Series A led by Andreessen Horowitz (a16z general partnership). Early stage, aiming to grow.
Enjoyed learning; more work = more learning opportunities. 50 problems, each teaching something different. Scale AI allowed learning different sides of the business; job wasn't confined to X, could expand to Y, Z, etc.
Being willing to dig into research is valuable, especially in AI. Easy to get lost in execution; valuable to step back and ask why. Learn by reading research papers. Example: quality of data—what makes data high quality vs. low? What do researchers care about? What makes a data point valuable? Went through much data across domains.
Met John through mutual friends at Scale; heard he was top 1% performer across the board. Didn't let confines of growth-stage Scale limit thinking; took first-principles approach to what was right for the business. Unafraid to voice opinions and move mountains to implement.
Went to Scale to observe AI development. They believed large language models (LLMs) would dramatically change the world. Selfishly wanted to start another company. Not immersing in rapidly developing LLM ecosystem is a mistake. Planned to spend next four years learning how LLMs work, their trajectory, implementation.
Job at Scale: ensure data produced was valuable. Thought about benchmarks, evaluations. Leveraged LLMs internally for efficiency.
Started using Cursor a lot with recent models; coding felt real. 0 to 1 went from 2 weeks to 30 minutes or half a day. Moment: "Wow, this thing is magical." Wanted everyone to use it: "Everyone on this team must vibe code." If you don't know how to vibe code, you'll be lost. It's a super powerful calculator.
In New York (lives there), friends in finance lack the same energy as San Francisco's excitement about latest tools/skills (e.g., Claude skills, crazy architectures). Tools increasing exponentially in SF, not in NY. Talks to team, candidates, investors—same observation.
No one has spent 1,000 hours building financial models with AI, specifically constructing AI for an LBO model. Bankers do it by hand; if you don't know, you don't. Possible decomposition where models excel at workflow parts, but requires investigation.
Coding tools built by users who know success metrics, use cases, failure points clearly. In non-expert domains, easy to say model is wrong, hard to identify why (e.g., number not expected).
Decided: we should solve this problem.
First job out of college: if you don't talk to someone, you'll never know. Advice: don't be scared to reach out. Don't think Satya Nadella won't respond—reach out, he might. Easy to think things impossible, but entrepreneurs suspend disbelief: "sounds totally doable," then try.
Learn by trying new things; if you fail, it's okay—build reps. Company promotes trying own way; if fail, support them, push deadlines, get company help. Build environment okay to experiment and not succeed. Goal: build masterpiece to transform lives, impact knowledge work in 5 years.
Worked in spreadsheets/Excel as banker briefly and consultant; viscerally understood enormous market—one of largest software markets. AI will fundamentally change spreadsheet work. Meridian augments spreadsheet workers like coding companies augment developers; huge potential for intelligent automation.
More time with technology builds intuition on what's possible today, in 3/6 months, 1 year—extremely valuable. Spend as much time playing with it. If interested in finance, spend 10,000 hours doing finance with AI.
Prompting is valuable skill. Y Combinator app forces clarity on business. Explaining task to LLM clarifies what you want; learn to be specific.
Keep EO in your library
Save the videos and channels worth coming back to, and find them again in one place.
