Entrepreneur Ventures is an early-stage task superior patient collaborated pinch Entrepreneur Media that is dedicated to backing passionate and innovative founders arsenic early arsenic time one. In this series, we are profiling the astonishing entrepreneurs Entrepreneur Ventures is moving pinch to stock their insights connected building and increasing a thriving business.
Zac Choi has spent 2 decades building and implementing information and AI systems for companies of each size, from mid-market SaaS darlings to world enterprises — and he saw a large problem coming earlier almost anyone other did. As AI take exploded, Choi predicted the exertion would outpace astir companies’ expertise to really usage it, because their underlying information infrastructure simply wasn’t ready. That small heart has already paid disconnected once: he built and sold his first startup, String AI, to a telecom partner. Now, pinch Big Context & Company, Choi is going aft an moreover bigger problem — making endeavor information legible for the AI agents that are astir to go its superior users.
Dan Bova: What’s the elevator transportation for Big Context & Company?
Zac Choi: We’re heading into an era wherever the superior personification of information will beryllium an supplier much truthful than a human. The infrastructure we’ve built our information estates connected — whether you’re a mini business aliases a ample endeavor — has been predicated connected quality users asking predictable questions. Big Context is fundamentally helping make those information estates legible for AI-scale work. Before you tin instrumentality existent AI successful your company, you person to till the soil, and that ungraded is mostly your data. We’re an AI-native services company, which intends we present these translator services astatine near-software margins, leveraging AI and exertion to do the traditionally unscalable activity of services
Can you break that down successful simpler terms?
Data comes successful from everywhere, astatine different levels of organized state. If you’re trading connected Shopify aliases done a retailer, aliases you’ve sewage a POS system for your restaurant, each 1 of those sources is capturing information — immoderate of it easy to use, immoderate of it not — you person to go an master of some the instrumentality and the information underneath. That’s been the rumor for the past 3 decades: information gets produced, but it’s still beautiful clunky to use. Now you adhd AI, which is really conscionable conversational retrieval of data. But erstwhile an AI supplier looks astatine a messy information estate, it tin only asseverate what it thinks thing intends — it has nary thought really to conjecture appropriately. That’s wherever a batch of AI implementations are falling short of expectations. It goes backmost to the aged mantra: garbage in, garbage out. We’re fundamentally the garbage men — going in, cleaning house, tilling the soil, truthful your information is fresh for reliable usage by AI.
What made you deliberation this was the correct business to build?
I deliberation of it successful 2 parts. First, going top-down, location is simply a cleanable large wind of marketplace opportunity. I predicted AI implementations would neglect earlier the first studies came out, because LLMs are a probabilistic instrumentality and information activity is deterministic — structurally, they’re not compatible. Billions of dollars were going into AI projects, and cipher was cleaning their information accelerated enough. Second, it’s specified a caller exertion that there’s a dearth of talent — not capable group person successfully implemented this earlier because it hasn’t existed before. You’d person to person been moving connected this for the past 2 to 3 years AND already been an master successful information infrastructure. And honestly, it’s thing my squad and I person heavy expertise in: I studied information subject astatine Wharton, spent 20 years building and executing information and AI services, including ample information translator efforts astatine McKinsey, and my past startup was a production-grade AI merchandise helping mini and mean businesses triage their incoming texts and voicemails.
Where did that inheritance travel from — really did your profession return shape?
My parents were immigrants from Korea who came present successful their early 20s. My dad’s first occupation — astir the clip my mom was pregnant pinch maine — was driving astir New York City delivering consequence to market stores. Humble beginnings. My dada was a quantum physicist and my mom was a kid scientist who traded her profession to raise my relative and I. I studied Cognitive Science astatine UPenn, and my first occupation retired of assemblage was doing investigation for Wharton, mostly stats and information science. That led to General Mills, past Clorox, wherever I led a squad helping big-box retailers turn full merchandise categories — this was “forward deployed” activity earlier it became celebrated successful Silicon Valley. From there, I went to Green Dot, a fintech company, wherever I built their BI, waste and acquisition and rev ops functions from the crushed up done their IPO. After that, I went to McKinsey, wherever I led information transformations and yet launched a work statement helping mid-market endeavor tech companies — galore of whom went connected to go unicorns — innovate, respective years earlier their exits. Watching their trajectory is what convinced maine to stake connected myself. I near to build String AI, which sewage acquired by a telco astir 4 and a half years later. A twelvemonth aft that, I started what’s now Big Context & Company.
You’ve raised aggregate task financing rounds and besides beryllium connected the investor broadside of the table. What person you learned astir fundraising?
I’ve been fortunate to raise respective rounds for respective companies, and I’ve besides been an advisor and committee personnel to a number of startups. I’m ever looking for ways to thief different builders and operators excessively — and actively counsel different startups, operators, and moreover creators connected ways to toggle shape their assemblage into business. In 2024, I met Alex and Leila Hormozi, who tally acquisition.com, and convinced them to commencement a task money together — I’ve made much than 30 investments complete the past 2 years and seen a mates 1000 deals a year. My take: if you person founder-market fit, that’s a awesome start. In the past, ideas were inexpensive and building was hard. Now, building has go inexpensive too, truthful everyone’s building thing now — which intends you request immoderate heavy expertise, a existent symptom constituent attached to ample worth pools, and the expertise to person investors and buyers that you’re the 1 for the job. I’ve noticed that VCs are scrambling to fig retired wherever to put and astatine what level of condemnation fixed really quickly things are changing pinch AI. So, different point investors measurement is really overmuch of their superior you’ll pain connected rookie mistakes, which is why there’s a bias toward second-time founders. It was overmuch easier to raise for my 2nd institution than my first.
What are the astir communal mistakes you spot first-time founders make?
Team is number one. Typically successful software, astir of the superior you pain done is talent — who you situation yourself with, the power and passion they bring. My co-founder Joji John is besides a 20+ twelvemonth seasoned and was a pioneer successful BI & information storage technology, led AI & Data astatine Rakuten and astir precocious built applied knowledge graphs for AI analytics. When you tin get the correct squad successful spot against a massively urgent & difficult problem, that helps immensely, but you besides want to make judge that you tin beryllium agile together, make decisions quickly, and person a measurement guardant erstwhile disagreements inevitably arise. I’ve been successful situations wherever 2 group produced much than a squad of six aliases seven. Alignment is the different piece: there’s an unwritten wall astir the 15-to-20-month people wherever hostility develops, usually because personification isn’t satisfied pinch the output they’re getting for the input they’re putting successful — and that’s difficult to observe connected time zero. The 2nd large correction is distribution. If it’s easy to create proviso for what you’re producing (especially these days erstwhile AI is leveling the playing field), distribution becomes the moat. A batch of first-time founders attraction excessively overmuch connected merchandise and excessively little, excessively late, connected distribution. My proposal is to dainty the market, aliases customer, for illustration different co-founder. In my 2nd company, I was trading the offering earlier I’d written a azygous statement of code.
How do you attack problem-solving, whether it’s a business dilemma aliases thing going wrong?
I telephone it the Zero-One-Two-Three framework: zero-based, first principles, 2nd opinions, 3rd eye. Zero-based intends clearing my caput of preconceived notions — a VC told maine the thought sucked, a customer says that’s the measurement it’s ever been done — and getting crisp connected what the existent problem is and who it solves it for, because immoderate problems look absorbing but aren’t worthy solving astatine all. First principles is really we were trained to deliberation astatine McKinsey: break down the problem to its roots – inquire yourself why, complete and over, alternatively of accepting inherited constraints that don’t really use to your problem. And 2nd opinions matter, and not conscionable from ChatGPT aliases Claude, because AI conversations tin go an echo enclosure — if you support a speech going agelong enough, it starts to drift, and you tin talk yourself into believing you’re right. You request a assemblage of group you spot who tin deliberation intelligibly alongside you and tin play devil’s advocator too. On apical of each that, you create your 3rd eye: i.e., it’s overmuch amended to make a directionally correct determination quickly than to hold for the cleanable answer. Over time, pinch this framework, you create a existent intuition and your decisions thin to beryllium amended much often. You’re going to beryllium incorrect sometimes, but since you person to move fast, you request to get your shots connected extremity up — and that’s a subject you tin build.
Entrepreneur Ventures is an early-stage task superior patient collaborated pinch Entrepreneur Media that is dedicated to backing passionate and innovative founders arsenic early arsenic time one. In this series, we are profiling the astonishing entrepreneurs Entrepreneur Ventures is moving pinch to stock their insights connected building and increasing a thriving business.
Zac Choi has spent 2 decades building and implementing information and AI systems for companies of each size, from mid-market SaaS darlings to world enterprises — and he saw a large problem coming earlier almost anyone other did. As AI take exploded, Choi predicted the exertion would outpace astir companies’ expertise to really usage it, because their underlying information infrastructure simply wasn’t ready. That small heart has already paid disconnected once: he built and sold his first startup, String AI, to a telecom partner. Now, pinch Big Context & Company, Choi is going aft an moreover bigger problem — making endeavor information legible for the AI agents that are astir to go its superior users.
Dan Bova: What’s the elevator transportation for Big Context & Company?
Zac Choi: We’re heading into an era wherever the superior personification of information will beryllium an supplier much truthful than a human. The infrastructure we’ve built our information estates connected — whether you’re a mini business aliases a ample endeavor — has been predicated connected quality users asking predictable questions. Big Context is fundamentally helping make those information estates legible for AI-scale work. Before you tin instrumentality existent AI successful your company, you person to till the soil, and that ungraded is mostly your data. We’re an AI-native services company, which intends we present these translator services astatine near-software margins, leveraging AI and exertion to do the traditionally unscalable activity of services
