Jessica Wu of Sola on Odds-Based Decisions, Selling Before Building, and Automating Enterprise Work

Open on YouTube ↗
Overview

Jessica Wu is co-founder and CEO of Sola, an intelligent process automation platform that uses AI to help businesses automate critical operational workflows. In this interview, Wu traces a path from competitive piano and math, through MIT, to quantitative finance, and finally to founding a startup. A thread runs through the whole conversation: startups are emotional, and Wu argues that founders make better choices when they reduce problems to their root and weigh decisions statistically rather than by feel. Wu then applies that view to concrete cases: turning down early revenue, adopting Y Combinator's "sell first" advice, winning enterprise trust as a small company, and coping with the highs and lows of founding.

15 min read

What Sola Is and Where It Stands

Wu describes Sola as a platform that helps businesses automate their "most critical and operational workflows using AI in a much easier and faster way than traditional RPA." The company came out of Y Combinator about two years before the interview. It raised a seed round led by Sarah at Conviction, and more recently a Series A led by a16z, which Wu attributes to strong traction with enterprise customers.

Wu reports several growth figures. Revenue grew 5x this year, and execution volume on the platform has doubled month over month since the start of the year. Customers include companies in the Fortune 100 and the Am Law 100, along with some of the largest private businesses in logistics and health care.

Competition, Discomfort, and MIT

Wu grew up competing in piano and math and believes this built a strong will and discipline. Wu counts risk-taking as a personal strength, meaning a willingness to put oneself in uncomfortable situations and to act before feeling fully ready. Wu says they do this often and believes it builds resilience and character.

Wu visited MIT several times in high school and describes it as a place that does a good job of making you "the dumbest person in the room," which Wu finds the most fun kind of environment because there is so much to learn from others. In Wu's telling, MIT is the one place that really is as the movies portray it, with students building roller coasters on the front lawn and training models in dorm basements. Wu also says MIT "puts a ceiling" on how technically hard things get. Even two years into running the company, Wu says, nothing since college has required as much sheer brainpower as the hard problems there.

Wu's advice for students is to jump into things as quickly as possible. That means learning what researchers are working on, spending as much time as possible with those researchers, and making a habit of taking in whatever is at the frontier. Wu argues this habit matters especially now, when the tech world changes so quickly.

Reducing Problems to the Root

Wu sees one main benefit of a technical background: it lets you break problems down easily. If you can state what you are trying to solve in simple terms, Wu says, it becomes much easier to solve. Wu thinks this matters more in an environment where "new model basically comes out every single week." Their answer is to keep coming back to what you are solving for your customer or user, and build from there.

Wu also distinguishes the root problem from "the problem that people are saying." Once you narrow things down to the underlying problem, Wu argues, you can deliver far more value to the people who use what you build.

What Finance Taught About Decision-Making

Before Sola, Wu had no plan to start a company. Wu had tried venture capital and worked at a couple of hedge funds, including one where Wu was the youngest quant researcher. An advisor had recommended time in finance because it teaches a special way of thinking, especially in trading. Wu summarizes that way of thinking as being objective and always calculating the odds.

Wu admits to not always thinking in a numerical or standardized way and says the years in finance helped a great deal with objectivity and rational analysis. Because startups can be very emotional, Wu says, going back to first principles and thinking statistically helps with decisions.

Wu gives startup examples. One is weighing a very large deployment. Another is deciding which features to build. Wu suggests asking questions like what the odds are that a particular customer converts by a particular date, and what you are giving up in exchange, while noting "this isn't perfect math." The point, Wu says, is to have a statistical framework rather than deciding because "I really like this customer, so I want to work with them." Wu agrees with the common claim that poker teaches useful startup lessons. In Wu's view much of it comes down to odds, rationality, and making decisions in the best direction.

The Origin of the Problem: RPA and Brittle Legacy Systems

The idea for Sola started at one of the hedge funds. Staff there did a lot of manual work in a very old brokerage system the firm had built on, and Wu had been trying to automate it with RPA tools. Wu explains that RPA, or robotic process automation, automates manual work by copying how humans work: moving the mouse, typing, clicking, and using browser and desktop applications.

At the time, Wu did not know how big the RPA market was. Wu did know that large companies have a lot of manual work and that no easy tools existed for it. Even with a computer science background, Wu found it hard to build a simple browser or desktop automation. Wu calls this "the seed of the problem."

Wu's co-founder had a similar experience building hospital systems at MGH. The old tools there surprised him with how brittle they were and how hard they were to implement. Wu calls these two experiences "really lucky glimpses" into how work is done in the real world.

Wu contrasts this with life at MIT, which is very tech-forward. Tools there are easy to use, everything has APIs that connect to each other, and automations are simple to build. At most real companies, Wu says, the work is very manual, and people move across many systems that don't connect: spreadsheets, internal portals, external systems, files, "and just about everything in between."

From YC Without an Idea to an RPA Company

Wu says the team entered YC without much of an idea and spent roughly the first month working out what to build. RPA appealed to them because they understood the problem well and knew how they wanted to solve it. Wu calls the goal of automating all digital work "a perennial and obvious one." What excited the team was the combination of applying AI to real enterprise work and the technical and model improvements coming from the other side.

The MVP and How the Product Changed

Sola's first version was simple. Like today's product, it had a recorder for capturing a workflow, which could then be uploaded. Unlike today, it did a poor job of showing what the workflow actually did. It could run the workflow, but only on the user's own computer, repeating the same steps. Wu says it was "not quite as intelligent or easy to use as we had hoped."

Recording and uploading still work the same way. Now, Wu says, workflows can run on hundreds of VMs at scale. Users can edit them in fine detail, add logic and extra information, and choose from many options for orchestrating and running them at scale.

Tradeoffs: Catching Up and Saying No to Revenue

Wu says the team probably underestimated one thing: existing tools "offer everything under the sun" after 20 years of development, which raises the question of how a newcomer catches up. The balance they found was to become useful to customers quickly enough, without trying to match everything at once.

Wu describes some painful moments early on when the team had to give up exciting early revenue to focus on the core product. They turned customers down at the very beginning, including some deals that were "almost there, but not quite." In the early months, Wu says, a company gets pulled in many directions. Some conversations meant saying no to big names the team was excited about. Wu gives two reasons: those deals did not fit the direction the team wanted to take, and the team was capacity-constrained, so taking them on would have meant sacrificing too much elsewhere in the business.

When panicked or under pressure, Wu relies heavily on advisors. Wu says they ask for a lot of advice and keep people with more experience close. The first step in a crisis is to gather those people's views, which Wu says usually helps.

YC's "Sell First" Advice and Its Limits

The team chose YC because of its reputation and because early advice told them they would be surrounded by great people and have three months to lock in. Wu says YC pushes founders hard to sell. Even without a working product, a team can put up a fake front end, for example with a tool like Lovable, and try to sell it.

Wu explains the reasoning. When you build something, you often don't know if you are going in the right direction. The clearest sign that people want it is that they will pay for it and keep paying. YC's message, as Wu puts it, is "don't spend 6 months to build a product. First try and sell something. See if it works." Wu adds that you may "burn a couple bridges" this way, so "your mileage may vary." Even so, you come away with a clear picture of whether people would pay, rely on the product, and see it as solving a real problem.

Wu also sets limits. Selling first stops being appropriate as a company grows: "don't sell a fake feature." The mindset still applies, though. Even on the right product, there are many possible directions, and Wu recommends experimenting, building mock-ups, talking to customers, shipping something, and then iterating and polishing.

Wu says the team met its first customer very early, around YC, while they were selling a product that did not yet exist. They were honest about it. They told the customer it would exist in a few months, and the customer said to come back when it did. Wu calls the advice counterintuitive, since people are taught all their lives to finish something before presenting or selling it. Wu says it is "not for the faint of heart." If you succeed and actually make the sale, you then have to deliver, and YC's advice at that point is roughly to "spend a week in the basement coding." Wu still thinks the approach points you the right way, because it gives a clear yes or no on whether the thing is worth doing at all.

Customer Delight as a Small Company's Advantage

Wu names Delivering Happiness as a favorite book. Wu describes it as a book about how small things matter and how customer delight works as a north star, and says delight takes many forms beyond delivering on time. Wu's team has found that "if your customers are happy, most other things fall in place."

Wu stresses that this is hard in their field. Sola is asking companies that have existed for decades, or even centuries, to trust a startup with their most critical operations. Wu says most customers now come through word of mouth, which Wu credits to how happy existing customers have been.

Wu lists ways a small, new tool can make up for its newness. The first, and obvious, one is to build a great experience that truly solves customer problems. The second is to ship fast and act on good feedback. Wu says this works in both directions: choose customers who give good feedback, and move quickly on what they ask for. Wu also mentions more traditional ways to win enterprise customers, such as being supportive and listening, and calls listening "the most important thing you can do for customers." The team still holds weekly calls with some early design partners, collecting feedback each week and delivering quickly. Wu believes this is part of why companies choose a startup over an old incumbent, beyond the tool simply being better.

Christmas Deployments and "We Can Never Go Down"

Wu tells the story of one of Sola's largest customers, who wanted a deployment last year during their quietest period, which was Christmas Day. The customer had used Sola for a long time and had been championing it internally. The team came in and worked that week. Wu's view was that if the deployment was going to happen on Christmas, they would do it, though Wu adds, "I hope that we don't have to ask this of the team again."

Wu says this reflects the stakes of Sola's work. Customers use it to send invoices, run accounts payable, send out shipments, and enter patient data. The team realized early that Sola "can never go down," because an outage could mean a real business with hundreds or thousands of employees has late billing that day. Wu contrasts this with consumer tools or AI tools that generate copy or find sales leads. When those go down it disrupts people's work, but "it's not the end of the world." For Sola, Wu says, "when we go down, it is sort of the end of the world." Wu says the team has worked "a million weekends and a million different holidays" to keep the service running and is very grateful to the customers who bet on an early startup. Wu hopes the company can keep delivering on its promises.

The Compressed Roller Coaster of Founding

Wu calls building a startup "really, really unimaginably difficult": a roller coaster of very high highs and very low lows, compressed so tightly that your best morning can be followed two hours later by some of the worst news you've heard, day after day or week after week. Wu gives examples of the bad news: a team member leaving, or a champion at a customer leaving in the middle of a sales process.

What Wu has learned is that things tend to return to normal. Wu has come to trust that things will be okay even when they don't feel okay, with two conditions: you are building in the right market, and you have the right people around you. As such events keep happening, founders learn to step back and handle much more than before. Wu says the early team learns this too, and seeing the bigger picture helps a lot with peace of mind.

Why Keep Going: A Ten-Year Problem

Wu closes with the reasons for staying committed. The problem is big enough that "there will be no problem of us having 10 years to spend on it." Wu also believes deeply in Sola's mission: much manual, tedious work shouldn't have to be done by people. Wu hopes businesses will rely on Sola for the rote work nobody really wants to do, so people can spend their time on the more fulfilling, creative, and interesting parts of work, such as leadership, strategy, and decision-making.