HomeCompaniesGreptile

Greptile

AI expert on your codebase that you can query via an API

Greptile is an AI that understands large codebases that you can query via an API. You can: - Use it to chat with your codebase - Plug it into Jira to generate descriptions for tickets, into GitHub to get automate PR reviews etc. with full knowledge of the codebase - Build custom internal tools and automations on top of the API! Greptile can be used via the cloud, or be hosted 100% on your cloud in an air-gapped VPC. Today, 800+ software teams including Wombo, Metamask, Warp, Exa AI, Bland, Leya, and more use Greptile to accelerate their software development lifecycle.

Jobs at Greptile

San Francisco, CA, US
$75K - $115K
0.20% - 0.50%
1+ years
San Francisco, CA, US
$120K - $175K
0.50% - 1.50%
Any (new grads ok)
San Francisco, CA, US
$150K - $200K
0.50% - 2.00%
1+ years
Greptile
Founded:2023
Team Size:6
Location:San Francisco
Group Partner:Brad Flora

Active Founders

Daksh Gupta, Founder

Co-founder of Greptile - the AI expert that has full context of your codebase. I studied CS at Georgia Tech, and interned at Amazon, where I was so good that they named an Amazon locker after me.
Daksh Gupta
Daksh Gupta
Greptile

Soohoon Choi, Founder

founder@greptile
Soohoon Choi
Soohoon Choi
Greptile

Vaishant Kameswaran, Founder

Founder @ Greptile. CS from Georgia Tech, minor in Linguistics.
Vaishant Kameswaran
Vaishant Kameswaran
Greptile

Company Launches

Hi everyone! We’re Soohoon, Vaishant, and Daksh from Greptile. We are building the AI expert on any codebase.

tl;dr Greptile is an AI platform that lets LLMs understand large codebases. Teams and devs at 250+ companies including Stripe and Amazon use it to get answers to questions about large codebases via our web app, automatically update their docs, build PR review bots on our API, and more. Try it now → app.greptile.com

PROBLEM

As software companies grow:

  1. Codebases get more complex.
  2. Legacy code and tech debt accumulates.
  3. Docs are usually outdated.
  4. Original engineers leave, and new engineers are less familiar with the code.

To get around this, developers:

  1. Just brute force their way to grokking the code - which is hard and time-consuming.
  2. Seek help from the subject matter expert, which is difficult if they have left, work remotely, or are just busy.
  3. Try to use tools like GitHub copilot, which are not designed to answer with full code context, so they fail to deliver.

In fact, the thing that makes coding hard is usually the codebase, not the code!

We found a way to index and search large codebases such that LLMs can always find the right context, allowing any developer to build internal developer tools powered by AI.

Here’s Felix from Twenty (open source CRM with 9k GitHub stars).

🦎 Greptile in action →

Here are some ways engineers use Greptile 👇

  1. Narek from Lantern (W24): grokking PostgreSQL’s source code, which Lantern is built on.
  2. Rahul from AgentHub (W24): parsing OSS frameworks that AgentHub integrates with.
  3. Jack Chapman from Cardinal Gray (S23): generating E2E test cases for multi-file services.
  4. Isaiah Granet from Bland AI (S23): integrating with the Twilio node SDK.
  5. SWE at Thera (S22): writing new features that require complex, multi-file changes.
  6. YC Medtech (S15): building an internal PR review bot on top of our API.
  7. SWE at Stripe (S09): debugging internal Rust libraries.
  8. Principal Engineer at Microsoft: reviewing his team’s code (open source).
  9. TMO (old B2B software company): working with their legacy .NET code.

ASK →

YC Sign Photo

YC Sign Photo

Selected answers from Greptile's original YC application for the W24 Batch

How long have each of you been working on this? How much of that has been full-time? Please explain.

We have been working on projects together for over a year. We have been full-time since May 2023 (graduated college). We pivoted to this 3 days ago.

Update: We have been working on this for just under 2 months now.

What is your company going to make? Please describe your product and what it does or will do.

AI assistant that helps devs onboard onto new, unfamiliar codebases, especially ones with bad or no docs. You give it a link to a repo, it generates docs for every file and directory, then lets you navigate and understand the code by asking questions in a chat interface.

An example of a question is "Where is the code that inserts stuff into the user database?" and the AI takes you to the specific file and explains how it works.

YC W24 Application Video

YC W24 Demo Day Video