Latte

Deck: Here

Fundraise: $1.1m

Founders: Michael Martin & Timothy Wu

My Notes

Founding Team

• Michael co-founded a student media publisher, growing to over 11m social media followers and 8bn video views. He is also a self-taught full-stack engineer and previously built a consumer SaaS product reaching the top 10 mental health applications in the UK.

• Tim is a self-taught AI engineer who used to be a lawyer. He has previously built a legal tech search engine using state of the art techniques in information retrieval and neural search. He replicated a DeepMind research paper within just 6 months of teaching himself programming

Vision

• Augment the entire video marketing lifecycle, from smart storage to creation and distribution.

• Become indistinguishable from a human video editor.

Problem

• If you don’t repurpose content to short form you’re not making the most of your long-form content, but significant time and effort is required to repurpose content.

• Issues with understanding ROI on video content and optimising for socials.

• The pain point is bigger (ie., the process is longer) at the enterprise level.

Solution

• Content repurposing automation

• Smart content libraries

• Content analytics and optimisation

Product

• Users input long-form content (podcasts, videos, interviews etc), and Latte leverages AI to extract the ten most relevant parts and optimises for socials (captions etc).

• ST Differentiation = quality of the product → slicker UX, more accurate subtitles, speaker detection etc. LT the defensibility comes from the data they collect as they win customers and specialise models at each stage of the workflow to provide significantly better output.

• Roadmap / next steps

  • Editing existing short-form content

  • Smart storage and content management

  • Collaboration

Technology

• Proprietary tools around audio processing, segment cutting, scene detection etc.

• Fine-tuning open-source models e..g, LLaMA → Captured 90k data points so far.

• The goal is to bring specialised models in-house on their own infrastructure, whether that be their own foundational models or fine-tuned models.

Traction

• Currently in beta. 1300 users and 12.5k clips generated.

• In the 9 weeks since introducing a pricing model, it has grown to >140 paying customers & $5.5k in MRR.

Market

• Capitalising on the growing trend toward short-form content e.g., TikTok, YouTube Shorts, Spotify etc

• Startups that solve the full workflow and build really good products leveraging AI will win in this space. Whether you fine-tune or build your own models.

GTM

• Social-driven GTM (leveraging Michael’s experience). Receiving a lot of inbound but also reaching out directly on LinkedIn.

• Targeting small businesses where the sales cycle is smaller. PLG approach to enterprise customers.

• Figma approach → Community help grows the product but revenue is driven by larger enterprise embedding Latte in the workflow.

Fundraise

• Raising $1.1m, 80k committed from angels

• Milestones

  • Grow high-quality revenue (PLG driven) → $1m ARR

  • Team: hire engineers to mobilise the data and continue fine-tuning models. Growth hires to hit ARR target.