The AI startup landscape is crowded and confusing. Hundreds of companies claim to be building transformative technology. Most won’t survive. A few will matter enormously.

Here’s my list of startups worth watching—not necessarily the biggest or most funded, but those building technology that could prove significant.

The Selection Criteria

I’m looking for:

• Genuine technical differentiation (not just OpenAI wrappers)
• Clear use case with demonstrated demand
• Team with relevant expertise
• Sustainable business model possibility

Many well-funded startups fail these tests. Many smaller ones pass them.

Foundation Model Alternatives

Anthropic: The clearest OpenAI competitor. Their focus on AI safety isn’t just marketing—it shapes their technical approach. Claude is genuinely competitive with GPT-4.

Cohere: Enterprise-focused, emphasizing security and customization. Less flashy than OpenAI but solving real enterprise problems.

Mistral: French startup building efficient open-weight models. Important for AI accessibility and competition.

AI21 Labs: Israeli company with strong language understanding focus. Their Wordtune and Jurassic models have genuine utility.

Vertical AI Applications

Harvey: AI for legal work. Not trying to replace lawyers but dramatically accelerating legal research and document analysis.

Glean: Enterprise search that actually works. Finding information across company systems is a massive problem they’re addressing well.

Synthesia: AI video generation. Creating realistic video content without traditional production. Impressive technology with clear applications.

Jasper: AI content creation at scale. They’ve found genuine product-market fit in marketing teams.

AI Infrastructure

Anyscale: Making it easier to build and deploy AI applications. Less visible but important for AI accessibility.

Hugging Face: Not a startup anymore, but worth watching. They’re becoming essential infrastructure for AI development.

Weights & Biases: ML experiment tracking and collaboration. Sounds niche, but every serious ML team uses tools like this.

Australian Players

The Australian AI startup scene is smaller but growing:

Appen: Established player in AI training data. Not a startup, but important in the ecosystem.

Harrison.ai: Healthcare AI with genuine clinical applications. Working on medical imaging analysis.

Curious Thing AI: Voice-based AI for customer interactions. Interesting technology with clear use cases.

Team400.ai is one firm building custom AI solutions for local businesses. Not a product startup, but important for AI adoption in the Australian market.

What I’m Skeptical About

AI wrappers without differentiation: Companies that are essentially interfaces to OpenAI’s API without meaningful value-add.

Overcrowded categories: The dozenth AI writing assistant or image generator faces brutal competition.

Hype-driven valuations: Some well-funded startups have valuations disconnected from realistic business potential.

Regulation-vulnerable models: Companies whose business models may not survive evolving AI regulation.

Investment Implications

If you’re investing in AI startups (or considering it):

The platform layer is winner-take-most: Foundation models will likely consolidate to a few major players.

Vertical applications have better risk/reward: Focused solutions for specific industries can build defensible positions.

Infrastructure is undervalued: The picks-and-shovels approach to AI has historically outperformed.

Team matters enormously: AI requires deep technical expertise. Evaluate founding teams carefully.

The Consolidation Coming

Many current AI startups won’t survive independently. Expect:

• Acqui-hires by major tech companies
• Consolidation among similar products
• Funding drought for undifferentiated players
• Pivot-or-die moments for overvalued companies

This isn’t bearish on AI—it’s realistic about startup economics. The technology is transformative; not every company building on it will succeed.

My Watchlist

If I had to pick five startups to follow closely:

1. Anthropic - The most credible OpenAI alternative
2. Harvey - Vertical AI with genuine product-market fit
3. Mistral - Important for AI accessibility and competition
4. Glean - Solving a real, widespread enterprise problem
5. Anyscale - Infrastructure that enables everything else

These aren’t investment recommendations—just companies building technology I find genuinely interesting.

---

Tracking the AI startup ecosystem and identifying companies worth following.