Transcriptly

This Founder is Making 1B+ Excel Workers 20x Faster | Meridian, John Ling

EO11:58Added Aug 31, 2026

Why did Silicon Valley’s top VC invest $17M in this startup founder?John Ling, co-founder & CEO of Meridian, is building AI for one of the most overlooked bu...

Watch on YouTube →
Contributed by 刘嘉琪

Transcript

Transcript format

  1. Intro

  2. 00:00I think I just really enjoyed learning
  3. 00:01new things, doing more work was just
  4. 00:03like more opportunities to learn. Oh,
  5. 00:05there's like 50 problems. Like each
  6. 00:06problem you probably like learn a little
  7. 00:08bit more about something completely
  8. 00:09different. The way I would just think
  9. 00:10about it is I would just go try it. And
  10. 00:12if you fail, that's okay. I don't
  11. 00:13believe any person on the planet spent a
  12. 00:17thousand hours trying to build financial
  13. 00:19models with AI. Okay, I'm going to do
  14. 00:21nothing except for like construct the AI
  15. 00:23and I'm going to try to build this like
  16. 00:25LBO [music] model that I would otherwise
  17. 00:27have to do for work. If you think about
  18. 00:28it, the bankers are just like, "We're
  19. 00:29just going to do it by hand." And then
  20. 00:30if you don't know how to do it, you
  21. 00:32probably just don't know how to do it. I
  22. 00:34think that there's probably some kind of
  23. 00:35big decomposition that you can do where
  24. 00:37like models can do different parts of
  25. 00:39this workflow very very well, but you
  26. 00:41just don't know because you haven't
  27. 00:42really like spent the effort to do like
  28. 00:44the investigation. And I was like, we
  29. 00:46should go solve this problem. My name is
  30. 00:49John, co-founder and CEO of Meridian.
  31. 00:51We're essentially building AI for
  32. 00:52spreadsheets. We think about like
  33. 00:53Microsoft Excel as most distributed
  34. 00:56programming language in the world. And
  35. 00:57our goal really is to say, "Hey, how can
  36. 00:59we help all of the people that spend a
  37. 01:01lot of time in spreadsheet software
  38. 01:03today, just moved 20 times faster."
  39. 01:05Prior to that, I spent about a year and
  40. 01:07a half at Scalei. Before that, I started
  41. 01:10a couple companies. We've raised
  42. 01:12slightly more than $15 million. Our CE
  43. 01:14brand was led by Andre Horowitz and the
  44. 01:16general partnership. And that's kind of
  45. 01:18where we are relatively early, but
  46. 01:20hopefully we can continue to grow.
  47. 01:23[music]

  48. How he became a top 1% performer at ScaleAI

  49. 01:33I think I just really enjoyed learning
  50. 01:34new things. I think more than anything
  51. 01:36else, I felt like doing more work was
  52. 01:38just like more opportunities [music]
  53. 01:39to learn. Oh, there's like 50 problems
  54. 01:41and each problem you probably like learn
  55. 01:43a little bit more about something
  56. 01:44completely [music] different. And I
  57. 01:45think skill was one of those places
  58. 01:47where if you wanted to learn about a
  59. 01:49different side of the business, you
  60. 01:51could go [music] do that. It wasn't
  61. 01:53like, hey, your job is like X. You can
  62. 01:55only do X. It was like, your job is X,
  63. 01:58but like if you do X and you realize
  64. 02:01that like Y and Z and ABC could also be
  65. 02:04done. There was the opportunity to
  66. 02:08essentially say, hey, I'm going to go
  67. 02:10learn and like [music] expand my
  68. 02:12personal sort of like knowledge space
  69. 02:14and like go do these things. Being
  70. 02:16willing to sit down and like dig into
  71. 02:19[music] research, for example, is
  72. 02:21extremely valuable. I think especially
  73. 02:23in AI, it it becomes relatively [music]
  74. 02:25easy to get lost in like the execution,
  75. 02:27meaning like, oh, okay, we're just going
  76. 02:29to do do this because [music] like we
  77. 02:30need to get this thing done. It's
  78. 02:32actually really valuable to take a step
  79. 02:33back. It's like, why are we [music]
  80. 02:34doing this? And then the way you learn
  81. 02:37is like you probably go read all these
  82. 02:39research papers. Let's just for example
  83. 02:40take like quality of data. Like what
  84. 02:42does it mean for data to be high quality
  85. 02:43versus low quality? What do researchers
  86. 02:45care about? What specifically makes this
  87. 02:48data point valuable? like I sat down and
  88. 02:50I read like I went through like so much
  89. 02:53of our data across so many domains and I
  90. 02:56think that's that's one way to learn.
  91. 02:58I met John through mutual friends at

  92. Why I Bet on This Founder - a16z, Kimberly Tan

  93. 02:59scale where I had consistently heard
  94. 03:01[music] that he was really a top 1%
  95. 03:03performer at scale. I heard this across
  96. 03:05the board from many many people. He
  97. 03:07didn't allow the confines of [music]
  98. 03:08scale which was already a growth stage
  99. 03:11larger startup at that point in time
  100. 03:13confine like what he thought was [music]
  101. 03:16right or not right to do in the
  102. 03:17business. And so he really took a very
  103. 03:20first principles approach in thinking
  104. 03:21about [music] what would the right thing
  105. 03:24for scale be and he was unafraid to
  106. 03:27voice those opinions to people and then
  107. 03:29actually move mountains [music] to make
  108. 03:30them happen.
  109. 03:30Why go over to scale but I do think like
  110. 03:32the biggest reason was definitely like I
  111. 03:35felt like it was a very unique place to
  112. 03:39[music] observe AI develop. I think they
  113. 03:42were very convinced obviously that the
  114. 03:44next wave of like [music] types of like
  115. 03:46large language models are going to very
  116. 03:47dramatically change trajectory of what
  117. 03:50the world looks like. For myself, I
  118. 03:53think selfishly I've always wanted to
  119. 03:54start another company.
  120. 03:55[music]
  121. 03:55I think that not knowing what LLMs can
  122. 04:00do or like not really immersing yourself
  123. 04:04[music] in sort of like this rapidly
  124. 04:07developing ecosystem or technology or
  125. 04:10however you want to think about it. It's
  126. 04:11like a mistake. [music] I would be much
  127. 04:12better off spending like the next four
  128. 04:14years at least at that time I thought I
  129. 04:16was going to be at scale for four years
  130. 04:18really like learning as much as I can
  131. 04:21about how large language models worked
  132. 04:24and how it was developing what was
  133. 04:26trajectory of technology and like how
  134. 04:28people are like implementing it etc. A
  135. 04:30lot of my job was making sure that like
  136. 04:32hey the data that scale ultimately
  137. 04:34produced was valuable. Spent a lot of
  138. 04:36time thinking about like benchmarks and
  139. 04:39evaluations. also spent a lot of time
  140. 04:41thinking about like hey how can we
  141. 04:42internally like leverage LLMs [music] to
  142. 04:46make our internal processes more
  143. 04:48efficient. Um so I think that for me was
  144. 04:50like really really really interesting. I
  145. 04:53started using cursor a lot um over the
  146. 04:55last [music] couple months or like you
  147. 04:56know the last generation of models where
  148. 04:58like hey coding like really felt very
  149. 05:00[music] real 0 to one actually went from
  150. 05:042 weeks to like 30 minutes or [music]
  151. 05:06like half a day. I had a moment where I
  152. 05:09was just like, "Wow, this thing is like
  153. 05:10magical." And I want like everyone
  154. 05:11[music] to like go use it, you know? I
  155. 05:13was just like, "Everyone on this team
  156. 05:14must vibe code." And if you don't know
  157. 05:16how to vibe code, I feel like you're
  158. 05:17[music] just going to be lost or you be
  159. 05:19left behind. But like ultimately, I
  160. 05:21think it was just, hey, there's like a
  161. 05:23new calculator, [music] but it's like
  162. 05:26not it's like a super super powerful
  163. 05:27calculator. But I think like more
  164. 05:29tangibly cuz I live in New York, a lot
  165. 05:32of my friends work in finance. And I
  166. 05:33think that like the energy is just like
  167. 05:35completely not the same, right? where
  168. 05:36like you're in San Francisco, everyone
  169. 05:39is like super super excited about like
  170. 05:40okay here's like the latest vibe coding
  171. 05:43like unlock right where like oh you have
  172. 05:46all these like skills that you can
  173. 05:47leverage for like [music] claude for
  174. 05:49example or like here's how you can do
  175. 05:51these like crazy architectures it feels
  176. 05:53[music] like the ground or the the the
  177. 05:56number of tools sort of like is
  178. 05:58increasing [music] like exponentially
  179. 05:59and then like you come back to New York
  180. 06:00and like that's just like not true when
  181. 06:02I talk to like our team when I talk to
  182. 06:05like candidates hits or even like
  183. 06:07investors. I think I [music] think a lot
  184. 06:08about the idea that I don't believe any
  185. 06:12person on the planet spent a thousand
  186. 06:15hours trying to build financial models
  187. 06:17with AI. I don't think anyone has been s
  188. 06:19sitting down and be like, "Okay, I'm
  189. 06:20going to do nothing except for like
  190. 06:21construct the AI and I'm going to try to
  191. 06:23build this like LBO model that I would
  192. 06:26otherwise have to do for work." If you
  193. 06:28think about the bankers, they're just
  194. 06:28like, "We're just going to do it by
  195. 06:29hand." And then if you don't know how to
  196. 06:31do it, you probably just don't know how
  197. 06:33to do it. But I think that there's
  198. 06:35probably some kind of like decomposition
  199. 06:36that you can do [music] where like
  200. 06:37models can do different parts of this
  201. 06:39workflow very very well, but you just
  202. 06:42don't [music] know because you haven't
  203. 06:43really like spent the effort to do like
  204. 06:45the investigation. In contrast to that,
  205. 06:47when you think about code, I think that
  206. 06:50a lot of these coding tools are built by
  207. 06:53the people who use them. So they have a
  208. 06:55much clearer idea of like what the
  209. 06:57success look like, what are the
  210. 06:58different use cases that I care about. I
  211. 07:01can very clearly articulate where the
  212. 07:02model is failing. But I do think when
  213. 07:04you take that and you apply it to a
  214. 07:06domain where you're like not really an
  215. 07:08expert, it's it's pretty easy to say
  216. 07:09like this model is wrong, but it's
  217. 07:11pretty difficult to really identify
  218. 07:14exactly why [music] like the number is
  219. 07:17not the number that you would expect it
  220. 07:18to be. But yeah, that's kind of how I
  221. 07:21thought about it and I was like we
  222. 07:23should go solve this problem.

  223. Bias Towards Action

  224. 07:27I think if I look back my first job out
  225. 07:30of college, I think [music] that most
  226. 07:32sales people can probably also tell you
  227. 07:34this, right? Is like if you don't try to
  228. 07:36talk to someone like you will never
  229. 07:37know. And I think that's something that
  230. 07:39I've like always done. I would say like
  231. 07:40don't be scared to reach out to people.
  232. 07:42Don't [music] think that like hey Satya
  233. 07:44Nadella will never respond to your
  234. 07:45email. I mean if you think that way he's
  235. 07:47obviously never going to respond to your
  236. 07:49email [music] but if you reach out you
  237. 07:51might be surprised. Maybe he'll respond.
  238. 07:53That's like something that you know that
  239. 07:55I thought was really really interesting.
  240. 07:56[music] It's really easy to fall into
  241. 08:00this narrative that [music] like oh
  242. 08:02these things are like impossible but you
  243. 08:04actually don't know and [music] I think
  244. 08:06like you know most entrepreneurs sort of
  245. 08:08just have that belief. I think it
  246. 08:10requires like an enormous [music] amount
  247. 08:12of like suspension of disbelief right
  248. 08:14where you can where most people would
  249. 08:16just be like you're crazy but you can
  250. 08:18actually go in and just be like I don't
  251. 08:20know what they're talking about. sounds
  252. 08:21totally doable, right? And then you
  253. 08:23would go try to do it. You also learn by
  254. 08:25like trying things that you've never
  255. 08:27tried before. And like if you up,
  256. 08:29you up. It's okay. Nothing wrong
  257. 08:30with that. But at least you know, right?
  258. 08:33And you can build reps internally. You
  259. 08:35know, our [music] company as a whole
  260. 08:36actually promotes and allows people to
  261. 08:39like try to solve things their own way.
  262. 08:40And if you fail, it's okay. You just go
  263. 08:42support them, right? You're like, "Hey,
  264. 08:44you tried this thing. maybe we need to
  265. 08:46push back the deadline by a few days and
  266. 08:48then we'll like find other people to
  267. 08:50support you, right? Everyone in the
  268. 08:52company will come support you. And I
  269. 08:53think you have to build this like
  270. 08:54environment where it's okay for people
  271. 08:56to like experiment and not succeed. I
  272. 08:57mean I think like obviously you always
  273. 08:59want to build something that is like
  274. 09:01like a masterpiece, right? Like I think
  275. 09:02our goal for like starting a company
  276. 09:04obviously is to like build something
  277. 09:05that we can be really really proud of
  278. 09:07that we think is going to transform a
  279. 09:09lot of people's lives that is going to
  280. 09:11be like hey here's a company that we can
  281. 09:13look back on in like 5 years and it has
  282. 09:15like dramatically impacted [music] the
  283. 09:17lives of like a lot of people as we
  284. 09:19think about how knowledge work is going
  285. 09:21to change with AI. [music] There's
  286. 09:23almost no bigger category of knowledge
  287. 09:25work than the spreadsheet and Excel
  288. 09:27worker. And as someone who worked in
  289. 09:29spreadsheets [music] and Excel as a
  290. 09:31banker for a brief period of time and
  291. 09:32then as a consultant, um I could just
  292. 09:34viscerally understand one [music] like
  293. 09:37why this was an enormous market um and
  294. 09:40probably in some sense like one of the
  295. 09:42largest uh software markets out there
  296. 09:44and two why AI was going to
  297. 09:47fundamentally change how we did work on
  298. 09:49spreadsheets. And so I think that uh
  299. 09:51Meridian's vision to really augment this
  300. 09:54form of knowledge workers similar to how
  301. 09:56a lot of the the coding companies have
  302. 09:58augmented the work of the developer. I
  303. 10:00think there's just so much potential
  304. 10:01here to actually be able to infuse the
  305. 10:03work done in spreadsheets with
  306. 10:05meaningful intelligent and automation.

  307. Spend 10,000 hours with AI - Own your unfair advantage

  308. 10:10The more time you spend with the
  309. 10:13technology, the easier it is for you to
  310. 10:16[music] have an intuition around like
  311. 10:18what is possible today. And if you do
  312. 10:20this over like a very sustained period
  313. 10:22of time, you also build an intuition of
  314. 10:25what is going to be possible in like 3
  315. 10:26months or what is going to be possible
  316. 10:28in like 6 [music] months or a year,
  317. 10:29right? And I think like that in of
  318. 10:31itself is extremely valuable. I would
  319. 10:33just spend as much time as you can
  320. 10:35playing with it, right? like I think it
  321. 10:37will be advantageous to be one of the
  322. 10:40people that have spent let's say you're
  323. 10:42interested in finance right that have
  324. 10:44spent you know like 10,000 hours trying
  325. 10:48to do finance with AI I think that
  326. 10:51prompting is still a very very valuable
  327. 10:53skill like when you apply to like Y
  328. 10:56combinator they actually tell you that
  329. 10:58like doing the application in and of
  330. 11:00itself is super valuable because it
  331. 11:03[music] forces you to sit down and think
  332. 11:06through these aspects of your business
  333. 11:08that maybe is [music]
  334. 11:11not as well articulated in your head as
  335. 11:14it is until you write it down. I think
  336. 11:16that when you explain a task to a large
  337. 11:21language model in a similar vein where
  338. 11:24you learn how to be relatively specific
  339. 11:27[music]
  340. 11:27about your ask, you learn a lot from
  341. 11:30that process, right? like trying to
  342. 11:31explain to NLM like what you really
  343. 11:33wanted to do actually gives yourself a
  344. 11:36lot of clarity around what you really
  345. 11:38want to do and I think that part of it
  346. 11:40is actually very valuable by itself.
This Founder is Making 1B+ Excel Workers 20x Faster | Meridian, John Ling — Transcriptly