Brian Chesky, CEO of Airbnb, spoke at a Y Combinator event this summer. (Y Combinator Photo)

Seattle-area startups that just graduated from Y Combinator’s summer 2023 batch are tackling a wide range of problems — with plenty of help from artificial intelligence.

The companies include:

  • Talc AI, a service for assessing large language models.
  • Watto AI, an AI program that generates consulting reports.
  • Neum AI, a platform designed to assist companies in maintaining the relevancy of their AI applications with the latest data.
  • Gleam, a platform for insurance brokers to implement self-funded health plans for SMBs and startups.

Founded in 2005, YC is one of the leading startup accelerators. Airbnb, Stripe, DoorDash, Instacart, Reddit, Twitch, and many others went through the program. More than 90 of its graduates are now valued at more than $1 billion.

Seattle startups that graduated from YC include LifeAt, ShelfEngine, Strac, Needl, Humanly, and others.

YC received more than 24,000 applications for this year’s summer batch. It funded 229 companies, down from last year’s 240. The Silicon Valley accelerator has been scaling back its investment pace.

Many of the companies in the summer cohort are working on AI-related products or services.

The latest cohort also includes Line.Build, led by Seattle-based CEO Dasha Cherepennikova. The startup sells tools that help users search for funding options for decarbonization projects.

GeekWire spoke to founders from Talc, Watto, Neum, and Gleam to learn more about their companies and how they are leveraging AI.

Talc AI

Talc.ai co-founders Matt Lee (left) and Max Kerr. (Talc Photo)

Co-founders: Matt Lee and Max Kerr

Explain what your startup does in two sentences: We provide end-to-end evaluation of large language model apps. When developers make a change to their AI, from prompt changes to retrieval strategy updates, we provide details on exactly how accurate their new build is.

What’s your secret sauce? We borrow proven techniques from the latest in NLP (natural language processing) academia to build evaluation tooling that any software engineer can use. Devs shouldn’t be neck-deep in evaluation pipelines just to test their software, so we solve that complexity for them.

The smartest move we’ve made so far: Teaming up together. We’ve been climbing partners for almost 10 years, forging our relationship through cold nights in the Cascade mountains, but it wasn’t until the YC interviews that we decided to join forces for a startup as well. We’ve trusted each other with our lives, from sharing the last few scraps of food to catching 20-foot falls off of granite cliffs, and that working relationship and deep trust is already paying off in the high demand environment of a startup.

The biggest challenge we’re facing: Telling [YC Managing Director] Michael Seibel that we haven’t moved to SF yet.

How are you using AI and how do you see it affecting your industry? We use Github Copilot daily, but haven’t begun to use it for email. We’re considering it for outbound sales.

There was no demand for us 2-to-3 years ago as large language models weren’t viable yet and so the industry didn’t exist, but Pandora’s Box has been opened and every knowledge work industry is about to be shaken up — it’s a matter of when, not if.

Watto AI

From left: Watto AI founder and CEO Rishabh Panwar, co-founder Ishita Bhandari, and CTO and founder Suryansh Soni. (Watto AI Photo)

Co-founders: Rishabh Panwar, Ishita Bhandari and Suryansh Soni

Explain what your startup does in two sentences: Watto AI uses LLMs to generate McKinsey quality reports in seconds. One of our customers uses us to generate custom white papers and marketing collateral for their enterprise prospects, by combining customer notes in Salesforce with market reports in Google Drive.

What’s your secret sauce? Watto integrates with petabyte-scale that pervades a modern enterprise across tools like Google, Atlassian, Microsoft suites, Slack, Jira and more. Watto securely uses this contextual data to build high quality documents/reports that employees spend quarters in writing and getting reviewed. Over time, our proprietary LLMs fine-tune and learn to become your team’s star performer.

The smartest move we’ve made so far: We use Watto ourselves to synthesize all the client meetings and public information of our customers to build a hyper-personalized game plan for every prospect. This not only helps us familiarize with their pain points, but helps demonstrate early value to those prospects.

The biggest challenge we’re facing: AI is progressing at breakneck speed and companies are scrambling resources to build internal AI committees, security teams to understand its potential and impact. In general, that has been slowing a lot of similar startups; however, we are helping our prospects in setting up these guidelines to make sure that every company leverages AI and empower their users and employees in turn. 

How are you using AI and how do you see it affecting your industry? Watto uses AI to automatically generate high quality documents and reports. We seamlessly integrate with an enterprise’s internal and external tools to build context-aware reports. Watto empowers users to do high quality work that directly impact business and user metrics.

AI will positively impact every vertical in every industry to not only boost their productivity but to help build better products faster. Manual, repetitive busywork will be a thing of the past and everyone will be empowered to do more high-quality work in their respective domains. 

Neum AI

Neum AI co-founders David de Matheu (left) and Pinhas Kevin Cohen. (Neum AI Photo)

Co-founders: David de Matheu and Pinhas Kevin Cohen

Explain what your startup does in two sentences: Neum AI is the next generation of data pipelines built specifically for retrieval augmented generation (RAG). Neum AI helps companies create vector embeddings and abstracts away the complexity of managing and keeping them up to date in real-time at a massive scale. 

What’s your secret sauce? Neum AI supports vector embedding workflows at a large scale while helping reduce the amount of time it takes to generate embeddings and saving on costs of maintaining them up to date.

The smartest move we’ve made so far: Getting into Y Combinator.  

The biggest challenge we’re facing: Continuing to evolve and keep up with the market needs. Generative AI applications continue to change as new use cases are developed. This is especially challenging when it comes to the needs of enterprise customers.

How are you using AI and how do you see it affecting your industry? Neum AI at its core is an enabler for generative AI applications by helping connect data into vector databases and making it accessible for RAG.

We are also using AI as part of the platform to help make data pipelines more contextual and easier to build for customers. By analyzing their data, we can predict the best processes and transformation to ingest that data into vector databases. We use this process to also analyze results and continuously improve the performance of the pipeline over time.

Gleam

Gleam founders Emeka Itegbe (left) Oliver Keh. (Gleam Photo)

Co-founders: Emeka Itegbe and Oliver Keh.

Explain what your startup does in two sentences: Gleam is a platform that helps insurance brokers implement self-funded health plans for SMBs and startups. Brokers can give their clients access to high-quality insurance at a fraction of the cost, without having to worry about plan design and administration, banking, or compliance.

What’s your secret sauce? We’ve got a great team. We’re highly technical founders with experience building products in the healthcare space.

The smartest move we’ve made so far: Easy — joining Y Combinator.

The biggest challenge we’re facing: There’s so much we want to do and build, and not enough time to do everything. 

How are you using AI and how do you see it affecting your industry? We have plans to incorporate more AI into Gleam in the near future. Today, the insurance industry relies heavily on manual, human processes that add to administrative burdens and rising costs. There’s a huge opportunity for AI to automate those processes and make the industry more efficient and affordable.

Like what you're reading? Subscribe to GeekWire's free newsletters to catch every headline

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.