I once attended a talk by Anges Callard on the topic of āAspirationā. Itās the desire to reach some future state thatās different from today. To create that future state, one must go through a period of struggle. If you aspire to have a family when youāre old, you must spend raise children when youāre young. If you aspire to run a marathon, you must physically train several days a week for months. If you want to buy a house, you must save money for years. Starting a business is an act of aspiration. Itās difficult, expensive, and time-consuming. If itās truly innovative, few people will believe in it until your version of the future has arrived. Startups begin with an idea, then iterate until they become relevant. Twitter started as an internal service at a podcasting company, Twitch started as a live stream of one person, and Lyft started as a long-distance carpooling service between universities.Ā The founders of these companies were living in the future, willing to endure many years of experimentation and uncertainty. Venture capitalist Mike Maples invested in each of the aforementioned companies. The breakaway startups heās invested in were not proven, predictable, obvious business plans. They started with an aspirational insight about the future, and iterated to become massive companies.Ā Pivoting with one leg firmly plantedIām the founder of Velvet, a startup thatās made several informed pivots since our inception. In the world of 0-1 startups, tiny pivots happen every day. Itās all about increasing the number of opportunities to succeed. Most startups die because they donāt pivot, even when the signals are exceedingly clear. Holding on to a stagnant idea is good for a founderās ego. But it hurts customers, employees, and investors.Ā Hereās the highlight reel of our pivots. These happened over several years, beginning before the company was incorporated or had investors: 1. Wallet: Seamless consumer crypto walletĀ 2. Authentication: Unified sign in for any appĀ 3. Customer data profile: Query data from any sourceĀ 4. AI SQL editor: Text-to-SQL on top of your database We killed some ideas before building anything, such as the crypto wallet. For others, we built an entire product, onboarded customers, and then continued building on top of them. This was the case with the AI SQL editor, which attracted a group of passionate customers. The AI SQL editor worked really well as a product, but we had limited insights into what happened between our app and OpenAI. We started storing requests to our database, giving us the ability to query AI logs directly. It gave us a level of flexibility and control over our AI features that we couldnāt get anywhere else. We didnāt think of the warehousing tool as a product until one of our customers (Find AI) asked to use it. We warehoused over 3 million requests for them in the first week, logging 1,500 requests per second during their launch. Which led us to develop Velvetās most recent productā¦Ā 1. AI Gateway: Warehouse every AI request to your database Now we have customers using Velvet daily for caching, evaluations, and analysis of opaque endpoints like the Batch API. Announcing the new VelvetIām excited to announce the new self-serve Velvet product. We built an AI gateway to warehouse every OpenAI and Anthropic request to your database. Itās a simple 2 line code change to get started. Requests are formatted as JSON. You can include custom metadata in the header - like user ID, org ID, model ID, and version ID. This means you can run complex SQL queries unique to your app. Granularly calculate costs, evaluate new models, or identify datasets for fine-tuning. Check out Velvetās launch on Product Hunt today! The path to product-market fitIf youāve followed Velvetās startup journey from the beginning, you know weāve explored many different ideas. Try our product today, and youāll see important vestiges from each iteration - from our seamless login system to the AI SQL editor. None of it mattered, until we made something people wanted. I believe this iterative process is the only way to find product-market fit. Set a hypothesis, get feedback, ship something cohesive, and attempt to drive adoption. Then repeat, continuously. The AI Gateway weāre launching today builds on months of compounding customer feedback and momentum. Itās purpose-built infrastructure for AI feature development. Iām excited to get it into the hands of more customers today. Just add 2 lines of code to get started. Learn more at usevelvet.com.