Rephrase Media creates free AI tools to help students and content creators practice ways of articulating themselves through writing. With this advanced solution, which provides students with AI generated variations of their own thoughts,users can learn through highly relevant examples, which is how they learn best. The service, simply called Paraphrase Tool, is completely free, with 15 paraphrasing modes in over 100 languages, including the options to simplify, clarify, smooth, and correct errors in their writing; additionally, as further examples, they can choose to make their writing more diplomatic, academic, or formal, as their preferences and needs might be. We’ve only been live for four months, but have seen over 10x growth each month and are now serving over 10K users a day.
Matthew Ramirez is an educator, investor, and serial entrepreneur. He is a Forbes 30 under 30 alumni. Ramirez worked as a writing instructor at UC Berkeley selling his first company, WriteLab, to Chegg in 2018. There, he worked for three years as Director of Product Management and oversaw AI products.
For the last eight years, I’ve worked on AI products to improve literacy in low-income and minority communities. My first product was WriteLab, which provided students with automatic feedback, and it was designed particularly for students in community colleges, where most do not have access to one-on-one tutoring or mentorship to help them develop into clear and effective writers. The question I have repeatedly been asked is, “why try to enact change through for-profit companies rather than through teaching?” It’s a fair question, given that I began my education career as a writing instructor at UC Berkeley. My answer is simple though: the need to improve literacy at scale is complex and urgent, and the fastest way to navigate through complexity is to iterate in a bottoms-up fashion, listening to user needs, making changes, and working to better serve them.
EdTech providers have the same responsibilities that a University has, which largely centers around protecting student data and anonymity. To this end, all data coming from student users through a school account needs to be encrypted, and any student data used for machine learning needs to be stripped of any personal identifying information before training. This will not only protect the identity of the student, but will also help prevent biases in the models that may emerge from personal information.
Education technology is changing rapidly, largely due to advances in AI. Now it’s possible to augment teachers meaningfully and scale learning services at very low cost. This is an environment where learners will win, having access to lower cost and higher quality services as time goes on.
Edtech is also an environment with ever lower costs of producing high quality AI solutions. This makes it very inviting to participate in this space, which means that there will be more competition and some risk of completely competing away profits. To deal with this, many larger edtech companies have engaged in buying sprees, consolidating sectors by buying out their competition.
One thing that makes a company successful is keeping the business model simple. As companies get bigger, it can be tempting to add more product lines with different business models to accelerate growth. While expanding your product ecosystem can make sense, complicating the business model rarely does as it is far harder to optimize many different functions than a single well-understood payment flow. One well-tested way around this is to explore bundling your products so that customers get the richness and variety of your offerings, while you get the simplicity of a single business model.
Another thing to consider well is product-market fit. Do users love or need the product? The answer to this is the foundation for everything else. If users love or need the product, they will come back to it, increasing the customer lifetime value. With a higher LTV, the team has more options for reaching customers, including advertising, amazing content, direct mail, and many other things that would be too expensive for products with lower LTVs. With a strong product-market fit, the team can more effectively test pricing, upsells, and cross-sells, all to improve the financial condition of the business.
Align as closely as possible with the inevitable. Try to establish a business that isn’t resistant to long-term trends, which might be clean energy, waste reduction, or automation. Minimizing waste at scale is one of the most attractive features of a business, as those that reduce wasted time, energy, or resources can offer more customer value for a lower cost, which has been a trademark of most successful technologies over time.
Additionally, here are some concrete ways you can evaluate the health and potential of your startup.
- High Aggregate Demand –
To assess whether demand for a product or service is big enough to matter, you can look at aggregate demand across search terms. To do this, you need to define parameters for your product and brand. There is a halo of associations that search engines will include as part of your brand when users use particular keywords. By identifying these keywords, you can assess whether the aggregate of monthly search volume for these terms is enough traffic for you to build the kind of business you want. If not, you can either pivot to something else or make your brand and products more comprehensive so as to include more keywords within that halo.
- A Product that Keeps it Simple –
Given the advice above, it may seem that the answer to making a market big is to make your product do many things. But this is not quite true. Most companies that succeed sell simple products with big opportunities, whether that means the options to deepen customer experiences or to provide offerings that complement the original simple solution. Additionally, you need to ensure that your product is stable and does what users expect it to do. Only then will you have a firm foundation to build on.
- Get Free and Low Cost Users –
To build a profitable business, you can’t spend more on users than you make from them, either in ad revenue or subscription fees. To determine an acceptable amount to spend on a user, you need to determine how much revenue you get from that customer over some period, the average subscription period being a good proxy for this. After subtracting what it costs to support that user (cost of goods sold), you have the figure above which you should not pay to acquire a new one. If you are spending more to acquire your users than you can expect to get from them over their lifetime with the product, you need to explore other acquisition channels. A good practice in general is to rank all of your channels by a ratio of customer acquisition cost to lifetime value of users coming through that channel.
- Keep Overhead Low –
Many investors encourage entrepreneurs to raise more money to move faster, but you can actually achieve fantastic growth with low operating overhead and strong financial discipline. To do this, you should always operate under the assumption that you are spending your own money, because, as it turns out, you are, as you are spending the equity you sold for outside capital. To achieve this financial discipline, founders should define processes by doing tasks themselves to start, delegating these tasks to contractors when the processes are well understood, and hiring full-time members only when the founders believe that this role is core to the business. In addition, there are many ways to grow quickly, such as by seeking out credits for cloud compute, getting as much search engine traffic as possible through low cost content, and cultivating strong relationships with brand advocates, bloggers, and other journalists. If you can achieve strong growth early on, you will get to profitability more quickly, and this profitability will provide you with both leverage and credibility if you decide to raise money later.
- Build a Moat around your Business with a Flywheel –
If you are building a product that has high demand, high search intent, and therefore likely low customer acquisition costs, you will almost certainly have many competitors. To win, you must develop a moat around your products and services. The best way to do this is to use the actions and information users provide to create a personalized experience for them, making it more difficult and far less interesting to use one of your competitors’ products. If your product is highly commodified, you have to emphasize ease-of-use, great design, high quality, and reliability.
The reality is that mastery is very difficult to achieve in entrepreneurship, since so much is uncertain and most challenges are very new. The best we can hope for is to learn through failure, come out with more savvy, and succeed with less cost the next time around. But the only way to succeed in business is to study your failures, understand what went wrong, and actively modify your approach to internalize a new set of operating habits. Practice breaking down problems into smaller subunits and isolate which methods sent you astray. If you do this enough times, you will have refined your approach to problem solving and may even surprise yourself with how good you’ve become at tackling bigger challenges.