🦄 Unicorner Startup of the Week:

Dawn

✍️ Notes from the Editors

Today's article is brought to you by Fidelity Private Shares. Managing your equity and investors should be simple—Fidelity Private Shares is the solution.

LAST CALL: we’re hosting two massive founder and funder events for SF and LA Tech Week in October. Join us and hundreds of founders and investors for two evenings of meaningful connections. Food, drinks, and vibes on the house.

Apply here:

• SF Tech Week 🌉

• LA Tech Week 🏖️

See you in SF tomorrow!

- Ethan and Arek 🦄

Analytics for AI Products

Dawn is an analytics platform built exclusively for AI products. Dawn integrates into a company’s codebase through three lines of code, and it extracts deep insights into how its customers use its products and interact with the language models within them. Through Dawn’s dashboard, teams can explore the topics their users are talking to their models about, create collections of events, search through user conversations, detect LLM response errors, translate conversations, and customize privacy controls. Dawn’s platform also has a Slack bot, which gives teams visibility through daily updates on key usage-based KPIs.

🔗 Check it out: dawnai.com

Shouldn't managing your equity and investors be simple?

Who here is feeling buried by cap table management? Day-to-day due diligence? Juggling all your investors? 🙋🏻‍♀️ 🙋‍♂️ 🙋🏽

Dealing with equity can be confusing. And fundraising shouldn’t have to be your full-time job. 


A solution: Fidelity Private Shares^SM

Fidelity Private Shares helps you:

• More easily manage your cap table and data room

• Get faster, more accurate 409A valuations

• Do sophisticated scenario modeling

• Fully automate your next funding round

Fidelity even has a vibrant startup community for access to exclusive events for founders and VCs.

Mention Unicorner to get 20% off your first-year paid subscription.

This is sponsored content. Unicorner and Fidelity are not affiliated. For sponsorship inquiries, please reply to this email.

💰 Business Model

Dawn currently sells directly to select AI companies and startups, as access to its product is not yet “self-serve.” Dawn works directly with AI companies to onboard them onto the platform. It charges its customers a base fee to set up and cover the upfront integration costs initially and then charges additionally based on usage. As Dawn continues to develop its product and scale, it aims to offer self-service functionality to integrate with its product.

📈 Traction and Fundraising

• Participated in Y Combinator’s Winter 2024 cohort

• Publicly launched two weeks ago on Sept. 25

• Onboarded early customers including Can of Soup, Type.ai, and Atlas

👫 Founders

• Zubin Singh Koticha, CEO: Previously CEO @ Opyn, Co-Founder @ Mechanism Labs, Researcher @ Thundercore, Head of Research @ Blockchain at UC Berkeley, Undergraduate @ Berkeley

• Alexis Gauba, Co-Founder: Previously Co-Founder @ Opyn, Co-Founder @ she256, Researcher @ Mechanism Labs, R&D @ Blockchain at UC Berkeley, EECS @ UC Berkeley

• Ben Hylak, Co-Founder: Previously Human Interface Design @ Apple, UX Engineer Intern @ Google, Avionics Software Engineer @ SpaceX, CS & Robotics @ Worcester Polytechnic Institute

📖 Founder Story

Before Dawn, Koticha and Gauba were building Opyn, a decentralized finance platform for people to sell speculative cryptocurrency derivatives and assets. After scaling Opyn and transitioning its leadership, they were on the hunt for new opportunities to innovate. They joined forces with Hylak, their close friend and Koticha’s roommate, and started hacking on all sorts of problems together. Eventually, they discovered the lack of visibility they had while hacking AI products together, ultimately leading to Dawn’s creation and their YC application.

💼 Opportunities

None at the moment, but Dawn is looking to onboard new customers. If you’re an AI Company that wants more visibility into your product, reply to this email and we’ll connect you.

🔮 Our Analysis

A few years ago customers would click and tap on products, but now, they talk to them—and founders aren’t listening. While clicks were quantitatively measurable, the rise of AI and our ability to now talk directly to products in natural language has made it difficult for companies with AI products to gain visibility into their product and customers. All the while, these conversations are more important than ever. In the current AI landscape where everyone is relying on the same foundational models, it’s really hard to build a better product than your competitors, meaning unlocking the valuable insights to do so is more essential than ever.

Measuring quantitative KPIs like retention and usage is relatively easy, but determining when users are having a positive or negative experience with LLMs when both inputs and outputs are natural language is challenging and often involves implementing complex sentiment analysis. One of Gauba’s insights was that rather than clicking the thumbs up or thumbs down buttons on ChatGPT (which is easily measurable), many users re-prompt the LLM with more specifics, or even simply swear at the model in frustration—responses that are extremely hard to measure but provide extremely valuable insights into a product’s performance. These direct conversations with models serve as a very rich dataset about a company’s product, but this data is meaningless without analytics. That’s why Dawn is so valuable to AI companies—its technology digests their vastly unstructured usage data and gives them the ability to learn from their customer experiences.

While others are also building to increase visibility into AI models, Dawn’s approach focuses on security and language context, which positions it extremely well to scale into being the go-to solution. Users of all kinds are worried about access to data and the unsettling pattern of companies using their data to further train models, potentially causing concerns around data security. Dawn’s focus on customizing security while still preserving the ability to extract insights makes it extremely attractive for privacy-minded users, especially those building in verticals with heavy data stringency, like healthcare or defense. Furthermore, Dawn’s approach to extracting meaning from our LLM conversations depends heavily on context—an ethos that others have yet to embrace. The real insights within LLM data come from the context within the conversations, continuity between messages, and domain-specific knowledge, and these are all areas in which Dawn is focusing, as evident through its topic discovery and event collection features.

Naturally, Dawn itself leverages AI to extract insights to make sense out of billions of tokens of natural language data. As AI technology continues to develop and more people build in the AI space, Dawn’s product improves while its customer base also grows.

📚 Further reading

• dawn: Analytics for AI products. [Y Combinator]

• Ben Hylak Building in Public [X]

Written by Rohan Desai

Reimagining life-saving injections

Pirouette is transforming life-saving injections with easy-to-use auto-injector technology.

With a 93% intended patient adoption, 21 patents, support from the National Institutes of Health, and backing from YC, they’re set to disrupt the $750B market.

Invest alongside Y-Combinator

Sponsored content

If you enjoyed today's article, forward this email to a friend!

If you're just seeing this email now, subscribe here.

Made with 💜 by the Unicorner team 🦄