Lilac AI

Summary

• One-liner: Scalable tool for data scientists to search, cluster, and analyze text datasets for generative AI
• Raising: $650K pre-seed at $8M cap
• Location: Boston, MA
• Website: https://lilacml.com

Investment Thesis

1000x Opportunity?

Strong founding team (ex-Google, co-created TensorFlow.js) tackling data quality for LLMs — a real pain point. However, acquired by Databricks within months of raising, so no opportunity to see the 1000x path play out.

Kingmaker Fit?

N/A — couldn't get allocation.

Green Flags

• Elite technical founders (decade at Google, built TensorFlow.js)
• Clear product-market pull — Databricks acquired them quickly
• Solves real problem: data quality for generative AI training

Red Flags

• None identified before round closed

Key Questions

• N/A (deal closed before diligence)

Decision

Missed — Wanted in but too late to get allocation.

Outcome

• Acquired by Databricks: March 2024
• Terms: Undisclosed

Lilac AI — Contacts

Founders

Name	Role	Email	LinkedIn
Nikhil Thorat	Co-founder	nikhil@lilacml.com	LinkedIn
Daniel Smilkov	Co-founder	daniel@lilacml.com	LinkedIn

Background

Both founders spent ~10 years at Google. Co-created TensorFlow.js. Nikhil was former tech lead of Google Image Search UI.

Relationship Timeline

• N/A — no direct contact established

Last Touchpoint

• Date: N/A
• Status: Missed opportunity
• Next Action: None — acquired by Databricks

Lilac AI — Notes

2024-03 — Post-Mortem

• Type: Review
• Summary: Learned about Lilac after they were acquired by Databricks in March 2024. Had raised $650K at $8M cap in pre-seed. Wanted to invest but were too late to get allocation.
• Lesson: Move faster on promising AI infra deals with elite technical founders.

References

• Databricks acquisition announcement
• SiliconANGLE coverage