AI Deal Watch 4.25.23
AI for peer-reviewed literature search, prenatal health diagnosis, and warehouse robots.
Regular updates on the latest VC-backed AI startups. Follow along to stay informed!
Consensus, a Boston-based AI search engine for research papers, raised $3M in Seed funding. The round was led by Draper Associates, with Nomad Capital and Winklevoss Capital participating.
Problem to be Solved
Conventional search engines make users manually sift through search results that may contain inaccurate or biased information, and generative search engines like ChatGPT don’t work for specific factual information because they “hallucinate”. Consensus is developing a search engine that uses AI to extract and distill findings directly from scientific research.
How They Use AI
Consensus uses AI for 2 main tasks: search and summarize. It is unclear what algorithms they use for returning relevant research papers, but it is probably more clever (and efficient) than a vector database given that they have 150M+ research papers in their database. They claim to use GPT-4 to create aggregate summaries across documents but do not specify how they create individual-level summaries. However, they are probably using a technique similar to LangChain that finds the relevant parts of a research paper based on vector similarity and then summarizes the answer per paper.
Business Model
TBD. Users can create a free account to use the beta product. Consensus criticizes ad-supported search, so it’s likely they will eventually launch a paid version of the service. Since launching in September, Consensus.app has attracted over 200K users.
IdentifAI Genetics, an Israel-based developer of AI prenatal diagnosis solutions, raised $3.3M in Seed funding. The deal was led by eHealth Ventures.
Problem to be Solved
Millions of babies are born every year with severe genetic diseases, but existing non-invasive screening methods allow the detection of less than 10% of known genetic disorders. Invasive genetic tests are risky, late, and thus rarely used. IdentifAI Genetics tests DNA found in maternal blood.
How They Use AI
The primary problem IdentifAI is solving is how to separate the fetus’ DNA from the mother’s DNA. They have a proprietary algorithm they call “Hoobari” that supposedly solves this problem with high accuracy. Hoobari is a Bayesian-based algorithm that calculates the probability of each chunk of DNA (cfDNA) belonging to the fetus.
Business Model
Pricing details are unclear but the company will generate revenue by selling its noninvasive prenatal testing services to healthcare providers and patients.
Robust.ai, a California-based AI robotics startup, raised $20M in Series A1 funding. Prime Movers Lab led the round with support from Future Ventures, Energy Impact Partners, JAZZ Ventures and Playground Global.
Problem to be Solved
Warehouse robots have been mainstream since Amazon acquired Kiva in 2012, but most are hard to use and only useful in narrowly defined tasks. Robust is trying to solve this with robots based on “collaborative productivity” which are built to work better with humans.
How They Use AI
It is not obvious what the role of AI is exactly in this product, however, many robotics tasks involve some degree of reinforcement learning. Additionally, the robots probably have pretty sophisticated computer vision sensors that allow them to perceive and classify objects in their environment. Additionally, the Grace software suite seems to use lidar sensors on iPads/iPhones to construct 3D environment maps to make deploying the robots as simple as possible.
Business Model
Carter — an autonomous warehouse cart — is paired with Grace, Robust’s software offering. The company will offer customers the two products bundled, with a robot-as-a-service (RaaS) payment model (though larger corporations can opt to pay upfront if desired).