Rare Disease Tech

What the fail of KovakDx says about HealthTech incentives.

Hey — It’s Nico.

Today’s issue takes 5 minutes to read.

If you only have one, here are the five main things:

This issue is brought to you by Intercom’s early stage program, offering a 100% discount to startups.

Let’s get into it.

Intercom for Startups

Join Intercom’s Early Stage Program to receive a 100% discount.

Get a direct line to your customers. Try the only complete AI-first customer service solution.

This Week In Startups

🔗 Resources

How LaunchDarkly’s CPO created a six-figure AI product manager as a side hustle.

Maven's founder shares how they've navigated product-market fit

Growth tactics from Canva, Figma, ChatGPT, and more.

📰 News

IIya Sutskever launches a new AI company.   

OpenAI acquires remote collaboration startup Multi.

Prosus slashes its $2.1B in Byju’s to zero but remains hopeful for recovery.

Zendesk launches VC arm to back early-stage AI startups

Fintech unicorn Unit lays off 15% of its workforce.

💸 Fundraising

10-minute delivery app Zepto raises $665M at a $3.6B valuation.

Accounting Startup Klarity secures $70M Series B to continue building AI capabilities.

Decagon emerges from stealth with $35M to provide ‘human-like’ AI agents for enterprises.

Khosla-backed Marble raises $5M to offer affordable group therapy for teens.

Ex-HubSpot exec launches AI-powered CRM Day.ai and raises $4M.

Fail(St)ory

Rare Disease Tech

KovaDx, a health tech startup focused on helping people with sickle cell disease, announced it is shutting down.

Why It Matters:

  • Sickle Cell Disease (SCD) is one of the most poorly funded diseases.

  • It is a genetic disorder that primarily affects people of African ancestry. In the US, it affects 1 in 365 black Americans. Those with the disease have a life expectancy that is 34 years shorter than the average.

  • KovaDx tried to help SCD patients by combining machine learning and 3D phase imaging. 

  • KovaDx's closure is primarily due to a lack of funding, highlighting the difficulties in securing financial support for rare disease research.

  • This is part of a bigger trend. The health tech industry has seen a severe decline in funding.

What Was KovaDx: KovaDx was a health tech startup founded by Dr. Yaw Ansong-Ansongton, who was inspired by his work as a doctor in Ghana, where he treated children with Sickle Cell Disease (SCD).

SCD is a group of genetic red blood cell disorders that can lead to severe complications, such as strokes. It is a lifelong illness with no known cure and predominantly affects people of African ancestry.

KovaDx developed a proof-of-concept Red Blood Cell Health Index, combining advanced imaging and machine learning. They hoped that this could help SCD patients by preventing severe episodes through routine monitoring.

The Numbers:

  • 📅 Founded in 2019.

  • 🚀 Joined the ABCT accelerator and received a $25K grant.

  • 💰 Raised $500K in different grants.

Reasons For Failure:

  • Lack of Funding: KovaDx struggled to secure sufficient funding. While they managed to obtain some money through various grants, it was ultimately not enough. In a post-mortem analysis, Dr. Yaw wrote, "There was no money! I attribute this to the fact that we were targeting a rare disease that affects a typically low-resource community."

  • Conflicts among Co-Founders: In the post-mortem, Dr. Yaw also detailed numerous conflicts that arose between the three co-founders. Disagreements ranged from which market to target (Dr. Yaw wanted to focus on the African market, while the other co-founders chose to pivot to the US market) to disputes over equity percentages. These internal conflicts are common among co-founders and can be a significant, though often hidden, reason for startup failure.

  • Decline in HealthTech Funding: Securing funds for rare disease research is always challenging, and the recent decline in the HealthTech industry further hindered KovaDx. As shown in the graph below, venture capital funding in the industry peaked in 2021 but has experienced a sharp decline since then.

Go Deeper:

Win or Fail?

Safe Superintelligence Inc (SSI)

Last week, Ilya Sutskever, co-founder of OpenAI, launched Safe Superintelligence Inc. (SSI).

  • The company aims to pioneer "Safe Superintelligence," emphasizing responsible AI development.

  • Sutskever parted ways with OpenAI a month ago amid disagreements with its leadership on AI safety strategies.

  • He claimed the Superalignment team was not given enough resources to ensure AI safety correctly.

  • Following his departure, OpenAI no longer maintains a dedicated safety team.

  • Sutskever claims that SSI’s team, investors, and business model will allow them to focus on achieving Safe Superintelligence instead of having to follow short-term commercial pressures.

Will SSI become the next OpenAI?

Login or Subscribe to participate in polls.

Trend

Claude 3.5 Sonnet

Anthropic has released its latest AI model: Claude 3.5 Sonnet.

Why it Matters:

  • Anthropic claims the model can go toe-to-toe against GPT4-o in reasoning, math, and coding.

  • It is now the most intelligent model in the Claude family and is also cheaper than its brother, Claude 3 Opus.

  • Like GPT4-o, Sonnet is a multimodal model. It has vision and can understand any image you upload.

  • This model positions Anthropic as the most important OpenAI competitor.

The Claude Family

The Claude family originally consisted of three models:

  • Claude 3 Haiku: The affordable, lightweight, but less intelligent model.

  • Claude 3 Sonnet: The mid-tier option.

  • Claude 3 Opus: The most intelligent yet most expensive model.

Claude 3.5 Sonnet is an upgrade to the mid-tier model. It significantly boosts intelligence while maintaining the same price. This upgrade makes Claude 3.5 Sonnet the best overall model in the Anthropic lineup, offering a perfect blend of speed, intelligence, and affordability. Consequently, Claude 3 Opus now finds itself in an awkward position, being more expensive and less intelligent than Sonnet.

The Benchmarks

Anthropic benchmarks claim that Claude 3.5 Sonnet outperforms GPT, Gemini, and LLaMA in areas like reasoning, coding, and math.

Additionally, Anthropic states that Claude 3.5 Sonnet is twice as fast as Claude 3 Opus. Its vision capabilities have also been significantly enhanced, bringing it much closer to GPT-4's multimodal capacities. 

However, Claude still lags behind GPT in some aspects:

  • It lacks voice interaction, so you cannot talk to the model as you can with GPT.

  • It does not have real-time vision like GPT-4. Its vision is limited to images.

Artifacts

Claude 3.5 Sonnet introduces a cool new feature called Artifacts. Essentially, Artifacts are separate windows that Claude creates when asked to generate content like text documents, website designs, or code.

If you ask Claude to write code, a separate window that functions like an IDE will appear. Claude can even run the code, allowing you to preview it directly within the Artifact window. This will enable you to create apps or games inside the Claude platform.

Help Me Improve Failory

How Was Today's Newsletter?

If this issue was a startup, how would you rate it?

Login or Subscribe to participate in polls.

That's all of this edition.

Cheers,

Nico