Bloop

Most of the Fortune 500 still use mainframes for some of their compute, but innovation, skills and cost pressure means that most are looking to reduce their mainframe workloads. We build tools that help teams modernise, write and understand legacy codebases, with the help of generative AI.

Sales & Marketing Agents

Visit Website

Founded

2021

Location

London, United Kingdom

Employees

Funding

$8M Seed

Bloop — AI Legacy Modernization and COBOL‑to‑Java Agents

Bloop is an AI company focused on modernizing legacy codebases—especially converting COBOL into readable, “human‑quality” Java—for large enterprises with mainframe workloads. The team is research‑driven across compilers, static analysis, LLMs, and Rust systems, with presence in New York and London and backing from [Y Combinator (S21)](https://www.ycombinator.com/companies/bloop). Their motion is sales‑led, targeting Fortune 500 and mainframe‑heavy organizations via [enterprise engagements](https://bloop.ai/contact-us).

Core focus: **AI‑assisted legacy modernization** and **COBOL‑to‑Java conversion**, plus code understanding and refactoring

Positioning: Java output “as a human would write,” ready for further development and consolidation

Roots: Open‑source, Rust‑based semantic code search and VS Code workflows

Why Enterprises Choose Bloop

**Reduce modernization risk:** Compiler‑backed, agentic pipeline with test generation and quality gates

**Readable target code:** Java tailored for maintainability and consolidation, not just syntactic translation

**Proven research pace:** Open technical posts and benchmarks, including [SWE‑bench Verified 71.14% in <19 steps](https://bloop.ai/blog/swe-bench)

**Enterprise alignment:** Sales‑led pilots for mainframe‑heavy verticals (banking, insurance, public sector)

Product and Technology

COBOL‑to‑Java Modernization

Agentic pipeline integrates compilers, static analysis, and tests to deliver auditable conversions

Evaluation rigor on COBOL using a HumanEval‑derived benchmark to track LLM progress and failure modes

Research and Benchmarks

[SWE‑bench Verified: 71.14%](https://bloop.ai/blog/swe-bench) with simplified tools and parallel tool use (Claude 4 Sonnet)

Open technical write‑ups and engineering deep dives

Open‑Source Code Search Roots

Rust and TypeScript stack with semantic + regex search, precise navigation, and GPT‑4 Q&A over code

VS Code extension and multi‑repo indexing improved developer workflows

Who It’s For

Large enterprises with substantial **COBOL/mainframe** estates

Banks, insurers, and public sector organizations undergoing multi‑year modernization

Engineering leaders and platform teams seeking **safe, testable code conversion** and consolidation

High‑Value Use Cases

Converting COBOL applications to readable, maintainable Java

Auditable modernization with **test generation and validation in the loop**

Legacy system understanding and refactoring at scale

Cross‑repo and monorepo code search and comprehension

Consolidating fragmented services into fewer, maintainable modules

Integrations and Stack

Historical developer tooling: VS Code extension; local and remote repo indexing

Vector search: Qdrant integration showcased in a public case study

LLM stack: Claude 4 Sonnet used in published SWE‑bench runs

Likely enterprise integrations include standard SCM/CI systems based on prior workflows; details are not publicly listed

Market Feedback

Pros

Fast, developer‑friendly **semantic code search** with natural language and regex; strong multi‑repo navigation

Handy **VS Code** workflow for surfacing examples and patterns

Effective for exploring unfamiliar codebases with semantic + regex search

Cons

Some users questioned long‑term support when the open‑source project paused or shifted direction

Product evolution from OSS search to enterprise modernization caused confusion about the current feature set

Sales Motion, Pricing, and Trial

Enterprise‑focused pilots; pricing not publicly listed for modernization

Historic OSS tool had a free tier; current GTM emphasizes **sales‑led evaluation**

To explore a pilot or proof of value, contact Bloop via the [sales form](https://bloop.ai/contact-us)

Company, Team, and Timeline

YC S21: [Company profile](https://www.ycombinator.com/companies/bloop)

Founders: Louis Knight‑Webb and Gabriel Gordon‑Hall; team strengths in compilers, Rust, and systems

Coverage in independent trackers reflecting the pivot to legacy modernization: [Dealroom](https://app.dealroom.co/companies/bloop_1), [CB Insights](https://www.cbinsights.com/company/bloop)

Policies and Contact

Contact and enterprise inquiries: [bloop.ai/contact-us](https://bloop.ai/contact-us)

Company presence: [LinkedIn](https://www.linkedin.com/company/bloopai)

Notable Resources

Company site and positioning: [Bloop](https://bloop.ai)

Research and engineering: [Blog](https://bloop.ai/blog)

Agentic modernization deep dive: [Building an AI agent](https://bloop.ai/blog/building-an-ai-agent)

COBOL LLM evaluation: [Evaluating LLMs on COBOL](https://bloop.ai/blog/evaluating-llms-on-cobol)

SWE‑bench results: [71.14% Verified in <19 steps](https://bloop.ai/blog/swe-bench)

Open‑source roots: [Bloop GitHub](https://github.com/BloopAI/bloop), [Qdrant case study](https://qdrant.tech/blog/case-study-bloop/)

Search‑optimized summary: Bloop is an enterprise AI agent company specializing in COBOL‑to‑Java conversion and legacy application modernization. With a compiler‑backed pipeline, rigorous benchmarks, and open research, Bloop helps Fortune 500 teams de‑risk mainframe modernization, deliver readable Java, and accelerate consolidation initiatives.

Related Companies

11x AI

11x is an AI and Automation company that builds autonomous digital workers.

Actively AI

Custom AI that replicates the depth of your best sellers to increase qualified pipeline and revenue. Founded by Stanford AI researchers and loved by scale-ups and enterprises like Ramp, Verkada, Navan, and Ironclad. Precision over volume.

AdsGency AI

#1 innovative CDP, connect your data and turn it into actionable campaigns!

Alta HQ

PEAK DATA PERFORMANCE Turn Data Into Customer Action ALTA is a cognitive computing platform that hyper targets consumers. We use analytics and cognitive science to understand audience personalities at an individual level. We analyze massive amounts of data to predict market performance. We then apply unique linguistic scores to create a predictive fingerprint based on subconscious motivators. - Turn anonymous transactions into rich buyer profiles - Communicate more effectively with existing customers - Develop and optimize message variants - Verbiage recommendation and variant optimization for your audience - Hyper target based on personality - Deploy in your ad network or marketing automation stack

Artisan AI

At Artisan, we're creating AI Employees, called Artisans, and software which is beautiful, easy to use, and replaces the endless stack of point solutions. We're starting with outbound sales. Our platform contains every tool needed for outbound sales - B2B data, AI email sequences, deliverability optimization tools and so much more. Ava, our AI BDR, operates within our platform. She automates: - Lead discovery with access to over 300M B2B contacts - Lead research, with 10s of data sources - Choosing a sales strategy and writing + sending hyper-personalized emails - Managing deliverability with a suite of tools - from email warmup to placement tests We're on a mission to create the final boss of software, with every SaaS product needed for sales and AI employees consolidated together in one exceptional platform. This is the next Industrial Revolution.