Heron Data

We are Heron Data, a Series A stage company based in New York and London.

• At Heron, we use the latest advances in AI and LLMs to automate document processing. Our software handles emails, analyzes attachments, and takes actions like updating CRMs or requesting more info to execute workflows end-to-end for our customers. We serve a range of businesses including some of the world's largest fintechs, banks and other specialty finance companies, commercial insurers, and more.
• Our largest product automates underwriting for small business lenders and other specialty finance companies. We eliminate cumbersome manual processes and enable our customers to provide their services with much greater speed, precision, and fairness.
• We are a rapidly growing company with over 150 customers in financial services and beyond who trust us to be embedded in critical processes - without us, their businesses stop. We analyze tens of thousands of documents a day, including over 20 million individual bank transactions daily for our customers. We continue to grow into more segments and workflows: further into lending, banking, insurance. And we think there's much more we can do!
• We are a lean, scrappy and committed team with a flat structure and collaborative culture. Our team brings together a wealth of experience from great tech companies including Facebook, Spotify, Revolut and many other startups from around the world.
• With team members from five different continents, we embrace multiculturalism and diversity. From company events to everyday interactions, we make sure everyone feels welcome and accommodate all cultural and dietary requirements.
• Since our launch in September 2020, at the end of the S20 batch, we've experienced a lot of growth! Today, we serve over 150 customers. Read more about our mission here

• For Frontend -- our customer dashboard is built with NextJS, using MUI, React, and TypeScript. We deploy our frontend apps on Vercel
• For Backend -- our backend applications are primarily written in Python using the Flask or FastAPI frameworks. We also use celery for async task processing. We have both a main monolith and also microservices. We containerize & deploy these applications on Kubernetes clusters or on managed compute (serverless)
• For ML — we have a variety of classification & prediction tasks that we deploy different models for, ranging from classic ML models / NLP techniques to fine-tuned LLMs
• For infrastructure — we use both GCP and AWS for different components. Some of our main components include Postgres database, Redis and RabbitMQ. All of our code repositories live in GitHub and we use GitHub actions extensively for CI/CD