Fundraising Fox

Greptile

YC W24

San Francisco, US Β· Founded 2023 Β· 35 employees Β· 3 known investors

Find your way into Greptile

534 people in our graph share verified history with the Greptile team β€” schools, employers, funds. One of them is your warm intro.

Brando De Courtenunlockedknows Vaishant Kameswaran Β· together at Georgia Institute of Technology (overlapped)
Γ—2knows the team Β· via UWC South East Asia
knows the team Β· via Georgia Institute of Technology
Γ—4knows the team Β· via Qualcomm
Γ—4knows the team Β· via HubSpot

Greptile provides AI agents that review and test pull requests by constructing a graph index of the codebase and using a swarm of agents to detect code quality issues, security risks, and logical bugs across multiple files. The platform serves software development teams and integrates with GitHub and GitLab.

Also known as Onboard AI Β· Greptile Β· greptileai

Founders & leadershipΒ· Y Combinator alumni (W24)

Greptile was founded in 2023 by Daksh Gupta, Soohoon Choi, and Vaishant Kameswaran.

DGDaksh Gupta
Daksh Guptain𝕏Co-Founder, CEODaksh Gupta studied computer science at Georgia Tech and interned at Amazon. He is co-founder of Greptile, an AI code reviewer.
SCSoohoon Choi
Soohoon Choiin𝕏Co-founder Β· ctoSoohoon Choi is CTO and cofounder at Greptile, a Y Combinator W24 company.
VKVaishant Kameswaran
Vaishant Kameswaranin𝕏Co-FounderVaishant Kameswaran founded Greptile. He studied computer science at Georgia Tech with a minor in linguistics.

Investors Β· 3

Funding

SEC filings, press & company announcements

$25M disclosed across 1 of 4 rounds Β· 2024–2025

Source: company announcements and press reports β€” follow each round's link for the claim.

The YC applicationΒ· Winter 2024 batch

Company profile

researched Sep 2026

Greptile is a San Francisco-based developer tools company that makes an AI code review agent. The product connects to GitHub or GitLab, builds a graph index of an entire repository β€” files, functions, classes and dependencies β€” and then runs a swarm of parallel agents over each pull request to assess the impact of changes beyond the diff. Findings are posted as PR comments, typically in about three minutes, with suggested fixes. Unlike file-by-file linters, the system is designed to catch multi-file logical bugs, security risks and style violations by reasoning across the whole system.

The agent personalizes over time: it reads other engineers' PR comments to learn a team's coding standards, incorporates thumbs-up/thumbs-down reactions and replies to suppress irrelevant suggestions, and auto-detects existing convention files such as CLAUDE.md and .cursor/rules. Teams can also write custom rules in plain English. Greptile positions itself as a "central validation layer" that sits between coding agents and production: review comments can be handed off in one click to Claude Code, Cursor, Codex, Conductor or Devin, a Greptile MCP server exposes comment context and learned rules to any AI agent, and a /greploop command lets a coding agent iterate with Greptile until issues are resolved. A newer agent, TREX, writes and runs tests for each PR in a sandbox to surface missed edge cases. Additional integrations include Jira and Notion, so a review can check a PR against the requirements of its linked ticket.

Greptile is offered as a cloud service (SOC 2 Type II) or self-hosted via Docker Compose or Kubernetes/Helm, including air-gapped environments and customer-supplied LLM providers, with SSO/SAML and audit logs for enterprise buyers. Twelve languages are listed as fully supported (Python, JavaScript, TypeScript, Go, Elixir, Java, C, C++, C#, Swift, PHP, Rust), with other languages supported at lower quality. The company also maintains public repositories including on-prem Helm charts (akupara), a CLI, a VS Code extension, a Codex plugin and agent skills.

Founding story

The three founders β€” Daksh Gupta, Soohoon Choi and Vaishant Kameswaran β€” met through Georgia Tech, where Gupta and Choi graduated in 2023 and Kameswaran studied computer science. They entered Georgia Tech's CREATE-X Startup Launch program with a different product, an AI shopping assistant called Tabnam, and pivoted after customer discovery showed their technology was better suited to software development. Gupta has credited CREATE-X with introducing him to co-founder Choi. Press coverage reports the company began in 2023 during a hackathon in San Francisco's South of Market district.

Business model

B2B SaaS sold on a per-seat subscription with credit-based usage. A free Starter tier covers one active developer with unlimited repositories and 50 credits per month; the Pro tier is $30 per seat per month including 50 credits per seat, with additional credits at $1 each. One credit equals one standard review and three credits equal one TREX review. Enterprise deployments add self-hosting, SSO, audit logs and dedicated support, and the review engine is also available via API with discounted bulk pricing.

Recurring per-seat subscription revenue plus consumption-based credit purchases, with API licensing at bulk rates and enterprise/self-hosted contracts.

Traction

The company's site states that more than 22,000 teams use Greptile; a January 2026 Georgia Tech article cites more than 2,000 customers. Greptile reported reviewing over 500 million lines of code in September 2025 and preventing more than 180,000 bugs that month, and its documentation claims 100,000+ bugs caught in production per month, an average review time of about three minutes and 9x faster time to merge. At its Y Combinator launch the company cited 250+ companies using the product. Y Combinator lists a team size of 35; a December 2024 press report described six employees at that time.

Latest developments

In September 2025 Greptile announced a $25 million Series A led by Benchmark Capital, with continued participation from Cory Levy, Y Combinator and Initialized Capital, and released Greptile v3 β€” a rewrite of the agent architecture with learning from human PR comments, a Greptile MCP server, Jira and Notion integrations, auto-detection of CLAUDE.md and .cursorrules files, and copyable prompts for coding agents. A January 2026 Georgia Tech article reported total capital raised of $30 million and a $180 million valuation, with more than 2,000 customers. The company is also offering early access to TREX, an agent that autonomously writes and runs tests for every pull request.

β–ΈFull profile β€” market position, technology, go-to-market, geography, history, risks & controversies

Market position

Greptile competes in the AI code review category alongside other automated PR review tools, positioning itself as a model- and vendor-independent validation layer for code produced by both humans and coding agents such as Devin, Claude Code and Cursor. The company reports that its v3 agent outperformed the other code reviewers it tested and catches three times more critical bugs than v2.

Whole-repository graph context rather than isolated file analysis, a learning loop that adapts to a team's standards from existing human PR comments and reactions to reduce noise, one-click handoff of findings into third-party coding agents and IDEs, autonomous test generation and execution via TREX, and deployment flexibility including air-gapped self-hosting with customer-supplied LLMs.

Technology

The core technology is a graph index of the repository capturing files, functions, classes and dependencies, over which parallel review agents operate. Version 3, released in September 2025, was a complete rewrite of the agent architecture. Supporting components include a learning system that mines human PR comments and reactions (with learnings isolated per organization), an MCP server, integrations with Jira and Notion, auto-detection of agent rule files, and the TREX agent that generates and executes tests in a sandbox. Deployment options span Greptile Cloud, Docker Compose, Kubernetes with Helm and air-gapped installs with customer-chosen LLM providers.

Go-to-market

Self-serve product-led adoption via a free tier and a five-minute GitHub/GitLab app install, complemented by a sales-led motion for enterprise and self-hosted deployments. Greptile offers free usage to qualified open source projects and discounts to pre-Series A startups, and demonstrates the product publicly by posting reviews on prominent open source repositories such as NVIDIA DALI, Meta PyTorch's OpenEnv and Solana Foundation's kora. Distribution is extended through IDE and agent integrations (Claude Code plugin, MCP server, Codex plugin, CLI) and through the Y Combinator network.

Software engineering teams and enterprises, including regulated and security-sensitive sectors such as defense, healthcare and financial services. Named users include Brex, Substack, PostHog, Bilt, Whoop, WorkOS, Browserbase, Mintlify, Bland, RollCredits, Raycast, Podium, Vouch and Y Combinator's internal software team.

Geography

Headquartered in San Francisco, with offices reported in the Transamerica Pyramid. The product is delivered as a cloud service or deployed in customer infrastructure.

History

Founded in 2023, Greptile joined Y Combinator's Winter 2024 batch (primary partner Brad Flora), launching publicly as an AI expert on large codebases used by developers at 250+ companies for codebase Q&A, documentation updates and API-built PR review bots. It closed a $4.1 million seed round led by Initialized Capital in June 2024 and shipped Greptile 2.0 in October 2024. In November 2024 founder Daksh Gupta drew widespread attention for publicly describing an 84-hour-per-week, no-work-life-balance culture. Subsequent releases included a May 2025 update adding long-term memory, MCP integrations, scoped rules and lower pricing. In September 2025 the company announced a $25 million Series A led by Benchmark Capital alongside Greptile v3.

Risks & controversies

In November 2024, founder Daksh Gupta's public statements that Greptile offers "no work-life balance," with typical workdays from 9 a.m. to at least 11 p.m. Monday through Saturday and sometimes Sundays, went viral and drew criticism that the approach was exploitative and a path to burnout; Gupta said he received death threats alongside job applications and later wrote that the schedule "isn't supposed to be forever because it isn't sustainable." Reported figures for customer counts vary across sources (22,000+ teams per the company site versus 2,000+ customers in a January 2026 article), and performance claims such as 3x more bugs caught and 9x faster merges are self-reported.

Compiled by commissioned research from 8 cited public sources β€” announcements, filings, and press listed under research sources below.

Key figures

latest reported
Average review timeJan 2026~3 minutes
Bugs caught in production per monthJan 2026100,000 bugs/month
Bugs prevented in monthSep 2025180,000 bugs
Companies using Greptile at YC launchJan 2024250 companies
CustomersJan 20262,000 customers
EmployeesDec 20246 employees
Faster time to mergeJan 20269x
HeadcountAug 202637
Lines of code reviewed in monthSep 2025500,000,000 lines
Pro plan priceJan 2026$30
Software engineer salary rangeDec 2024$120,000-$200,000 per year
Team sizeJan 202635 employees
Teams using GreptileJan 202622,000 teams
Total capital raisedJan 2026$30M
ValuationJan 2026$180M

Company-reported or press-reported figures, each dated to when it was claimed β€” not independently audited.

Images

Greptile photo

6 people who came through Greptile went on to found or lead other companies.

Competitors Β· 10

by search overlap
Qodo115 shared keywordsQodo builds a code quality and review platform for engineering teams working with AI coding agents, providing multi-agent code review, self-learning enforceable coding rules, and a context engine that reasons over full codebase context across the software development lifecycle. It targets enterprise engineering organizations with security features such as zero data retention, SOC 2 Type II certification, and on-premises or single-tenant deployment.
Augment Code99 shared keywordsCosmos is a unified AI agents platform that enables enterprise development teams to automate software development workflows at scale, improving engineering throughput and reducing time-to-merge through intelligent agent coordination and codebase understanding.
CodeRabbit90 shared keywordsCodeRabbit provides AI-powered code review tools for developers to improve software quality and efficiency in development workflows. The platform helps teams automate code review processes while allowing humans to focus on higher-level reasoning and strategic decisions.
Graphite87 shared keywordsGraphite Chat is an AI-powered code review tool that integrates with GitHub to automatically analyze pull requests and identify bugs and security issues. It serves software development teams seeking to improve code quality and catch problems before deployment.
CodeAnt AI85 shared keywordsCodeAnt is a code security platform that identifies and prioritizes security vulnerabilities in applications using AI to surface the most critical issues from thousands of findings. The company serves software development teams and enterprises building with both human-written and AI-generated code.
Pieces68 shared keywordsPieces is an AI memory product that automatically captures a person's work as it happens β€” across meetings, docs, messages, tabs, and chats β€” and keeps it current so AI tools and agents are grounded in the user's actual context. Its data stays on-device by default, giving users control over what is captured.
Atlassian68 shared keywordsAtlassian offers a teamwork platform combining project planning, knowledge management, and AI orchestration tools for teams and their AI agents. Its products help organizations plan, execute, and deliver work at scale.
Bito56 shared keywordsBito provides AI-powered development tools designed to accelerate developer productivity and code quality. The platform targets software development teams seeking to reduce time on routine tasks, minimize bugs, and focus on higher-level innovation.
Netezza56 shared keywordsIBM is a global technology company whose business spans enterprise software (including Red Hat, HashiCorp, and Confluent), IT infrastructure such as mainframes, servers, and storage, and IT consulting services. The company is also investing heavily in quantum computing and AI-based enterprise offerings, including its Lightwell open-source software security clearinghouse and the Anderon quantum wafer foundry.
LogRocket55 shared keywordsLogRocket provides session replay, product analytics, and AI-powered issue detection for web and mobile applications. The platform helps product teams understand user behavior, identify technical and UX problems, and monitor frontend performance.

Companies competing with Greptile for the same Google search keywords, organic and paid, via search-intersection analysis.

Timeline Β· 8

launches, deals, and filings
Sep 2025
Greptile raises $25M Series A led by Benchmark Capital

Greptile announced a $25 million Series A led by Benchmark Capital with continued support from Cory Levy, Y Combinator and Initialized Capital.

$25M source β†—

Sep 2025
Greptile v3 released

A complete rewrite of the core agent architecture, adding learning from human PR comments, a Greptile MCP server, Jira and Notion integrations, auto-detection of CLAUDE.md and .cursorrules, and copyable prompts for coding agents. The company states v3 catches 3x more critical bugs than v2.

source β†—

May 2025
Product update adding long-term memory, MCP integrations and scoped rules

Update introducing long-term memory, MCP integrations, scoped rules and pricing reduced by 40%.

source β†—

Jan 2025
TREX testing agent early access

TREX, a Greptile agent that writes and runs tests for every pull request in a sandbox to catch bugs and missed edge cases, opened for early access.

source β†—

Nov 2024
Founder's 84-hour workweek policy draws public criticism

Founder Daksh Gupta publicly stated that Greptile offers no work-life balance, with workdays from 9 a.m. to at least 11 p.m. Monday to Saturday and sometimes Sundays. The posts went viral, generating debate over whether the approach was exploitative.

source β†—

Oct 2024
Introducing Greptile 2.0

Release of Greptile 2.0, described as a platform for codebase understanding with code reviews, JIRA integration and custom workflows.

source β†—

Jun 2024
$4.1M seed round led by Initialized Capital

Greptile closed a $4.1 million seed funding round led by Initialized Capital in June 2024; a June 6, 2024 news item covering the raise is listed on the company's Y Combinator profile.

$4.1M source β†—

Jan 2024
Y Combinator launch: AI expert that understands large codebases

Greptile launched via Y Combinator as an AI platform letting LLMs understand large codebases, with teams at 250+ companies using it for codebase Q&A, documentation updates and API-built PR review bots.

source β†—

Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.

In the news

β–ΈResearch sources Β· 8

primary sources listed

8 public sources were cited for this profile; the first-party ones are listed here.

Frequently asked questions

What does Greptile do?
Greptile builds an AI code review agent that indexes a codebase as a graph and reviews pull requests with full repo context.
Who founded Greptile?
Greptile was founded by Daksh Gupta, Soohoon Choi, Vaishant Kameswaran in 2023.
Who are Greptile's investors?
Greptile's investors include Y Combinator, Benchmark, Initialized Capital.
How much funding has Greptile raised?
Greptile has disclosed $25M raised across 1 of its 4 known rounds.
Where is Greptile headquartered?
Greptile is headquartered in San Francisco, US.