Agents Gain Ground as Open Source and Testing Attract Big Funding

Last week showed just how quickly the developer stack is evolving. From fresh funding for smarter agents to new tools that put testing back at the heart of shipping code, the energy is on making AI less risky and more reliable. The biggest launches doubled down on open source, tapping into community trust and speed of iteration. And with investors writing massive checks into core infrastructure, it’s clear the foundations for the next wave of devtools are being laid right now.

This week brought in some serious funding, $523M in total for startups building the next wave of developer tools (yes, vibe tools included).

And with conference season in full swing, we’ve also pulled together a round‑up of October events to highlight. So grab your diary, pour a hot drink, and settle in for the latest.

🏆 Funding Wins

Flox Secures $25M Series B For Reproducible Dev

Development environment startup Flox has raised $25M in Series B funding led by Addition with NEA, D. E. Shaw, Hetz and Illuminate also backing the round. Founded by Ron Efroni and Michael Brantley, Flox makes setting up consistent dev environments as simple as running a single command. Already adopted by over 300 organizations, including PostHog and D. E. Shaw, the platform automates reproducibility at scale. The Flox Catalog now supports 180,000 packages along with NVIDIA CUDA binaries, so teams can jump seamlessly into GPU‑intensive work. The new funds will go toward expanding AI security, governance features, and real‑time vulnerability detection.

🌀 Big step for Flox — read the update

Greptile Snaps Up $25M To Scale AI Code Reviews

Greptile, the San Francisco startup behind AI‑powered code reviewers, has raised a $25M Series A led by Benchmark with Y Combinator, Initialized Capital and Cory Levy also in. Co‑founder Daksh Gupta says version 3 of its agent can now catch three times as many bugs, earning early adoption from Brex, Substack and PostHog. In the past month alone it reviewed over 5M lines of code and flagged 180,000 issues before they hit production. Gupta calls Greptile v3 “a complete rewrite of our core agent architecture.” With rivals like CodeRabbit and Graphite racing ahead, Greptile will be using fresh capital to double down on speed and reliability.

🐊 Kudos to Daksh Gupta and team — see the announcement

Rocket Blasts Off With $15M Seed Round

Indian startup Rocket, founded this year by Vishal Virani, Rahul Shingala and Deepak Dhanak, has raised $15M in seed funding led by Salesforce Ventures with Accel and Together Fund also joining. Rocket is pitching itself as the first “vibe coding platform” built not just for code generation but also iteration, deployment and scaling. In just 16 weeks it onboarded 400,000 users in 180 countries, 10,000 of them paid, and hit $4.5M ARR. The startup already counts Meta, PayPal and PwC developers testing serious apps built on its system. Virani notes:

“we are building the first vibe solution platform, solving not just day one but day two problems.”

🚀 Big congrats to Rocket — explore Rocket.new

Synthesized Raises $20M To Push AI Testing

London- and New York‑based Synthesized has closed a $20M Series A led by Redalpine with backing from IQ Capital, Mercia, UBS and Seedcamp. Deutsche Bank, both a customer and existing investor, also joined. Founder Nicolai Baldin says the rise of vibe‑coding makes deep automated testing essential, explaining:

“We are making sure we really identify those things which are going to break your app, at the data level, on the environment level, and help you expose those breakage points… This is absolutely critical because those traditional [testing] coordinators, they don’t do that.”

Synthesized’s software halves test data prep times for enterprise users and cuts QA costs by 40%. With demand for autonomous QA agents rising, the team plans fast growth in North America and Europe.

🧪 Read the announcement on Fortune.com and if you haven’t already, explore their website

Factory Forges $50M Series B For Agent‑Native Dev

San Francisco startup Factory has raised $50M Series B at a $300M valuation. NEA led, with Sequoia, J.P. Morgan, Nvidia, Abstract Ventures, Mantis Ventures, Frank Slootman, Nikesh Arora and Aaron Levie among backers. Founded by Matan Grinberg and Eno Reyes, Factory’s “Droids” help developers delegate refactors, debugging, migrations and more across any IDE or model. Already powering engineering at MongoDB, EY and Zapier, teams report 31x faster feature delivery and massive cuts in on‑call times. With this raise, Factory wants to cement agent‑native development as the next stage after autocomplete.

🏭 Well done to Factory — learn more and watch announcement from founder

Ultralytics Banks $30M To Power Next‑Gen Vision AI

Ultralytics, the company behind the widely adopted YOLO object detection models, has raised a $30M Series A led by Elephant with SquareOne participating. Founder Glenn Jocher has grown Ultralytics into one of the world’s largest AI developer communities, with 115K+ GitHub stars and models used over 2B times daily. Customers include Siemens, STMicroelectronics and Duolingo. “Ultralytics’ success is built on the conviction that open source drives enterprise innovation,” Jocher said. Profitable since launch, the firm will use its new capital to expand R&D, enterprise solutions and global presence. The raise coincides with its YOLO Vision conference in London this week.

👁️ Kudos to Glenn Jocher and team — see more here

Testkube Raises $8M For Continuous Testing At Scale

New York‑based Testkube has raised $8M Series A led by Ratmir Timashev and Insight Partners. Founded by Ole Lensmar and Dmitry Fonarev, the startup emerged from Kubeshop in 2023 and has already powered over 100M automated tests for Cloudflare, Siemens and Adobe.

“AI is rewriting the rules of software development, yet it also makes testing significantly harder.” (Fonarev)

Testkube integrates into Kubernetes workflows, orchestrating frameworks like Playwright, Cypress, Selenium and Postman. With continuous testing becoming the ultimate safety net for AI‑driven dev, the team is expanding enterprise adoption and adding AI‑powered orchestration and remediation features.

⏹️ Nice win for Testkube — learn more

Modular Scores $250M Series C At $1.6B Valuation

Modular, founded by ex‑Apple and Google engineers Chris Lattner and Tim Davis, has raised a $250M Series C led by U.S. Innovative Technology Fund with DFJ Growth, GV, General Catalyst and Greylock. The company wants to be the “Switzerland” of AI computing, letting developers run AI apps across chips from Nvidia, AMD and others without rewriting. Already used by Oracle and Amazon, Modular is positioning itself as the VMware of AI, an “AI hypervisor” layer. Lattner said it’s about enabling fair competition, not crushing Nvidia. The new funding will fuel expansion into AI training and rapid hiring across engineering and GTM.

💻 Huge milestone for Modular — read the full story

📈 Series D and Up

• Cohere raised a $100M Series D extension at a $7B valuation, with BDC and Nexxus joining. The Toronto startup, founded by Aidan Gomez, continues to target global enterprises with its Command model family and newly launched North platform. It also announced Joelle Pineau (ex‑Meta AI) as Chief AI Officer and Francois Chadwick (ex‑Uber CFO) as CFO. Fresh capital will push Cohere’s sovereign AI and enterprise‑first strategy, strengthened by a growing partnership with AMD.

🚀 3, 2, 1… Launches

Klavis.ai Launches Strata To Guide Agent Workflows

Klavis AI (YC X25), founded by Xiangkai Zeng, has launched Strata, an MCP server designed to make AI agents smarter at picking and using tools. Instead of dumping hundreds of APIs into context, Strata guides agents step by step, reducing overload and boosting accuracy. On the MCPMark benchmark, it outperformed GitHub and Notion’s own servers with a +15% uplift. In real‑world tests it hit over 83% accuracy for complex workflows like CRM queries and multi‑app chains. Zeng, who previously worked on Gemini’s tool use team at Google, says Strata is about thinking like a human. Developers can sign up free or self‑host from open‑source code.

📊 Well done Klavis.ai — try Strata | show support on Product Hunt

Kilo Code Brings Open‑Source AI Agent To JetBrains IDEs

After more than 420,000 downloads on VS Code and Cursor, Kilo Code is now available inside JetBrains IDEs including IntelliJ IDEA, PyCharm and WebStorm. The open-source AI coding assistant is model-agnostic, pay-per-use, and comes with four key modes: Architect, Orchestrator, Code and Debug. Growth lead Job Rietbergen announced on Product Hunt:

“After 420,000+ developers started using Kilo Code on VS Code and Cursor, our #1 feature request was: ‘When will you support JetBrains?’ Well… we listened!”

The launch also brings persistent memory, Context7 for up-to-date libraries, and 100% open source transparency backed by a community of more than 8,000 developers.

🔓 Congrats to Kilo Code — explore here

SimpliflowAI Debuts Loop MCP For Seamless Integrations

SimpliflowAI has released Loop MCP, pitched as a universal gateway connecting AI assistants to unlimited integrations without breaking context windows. Launched by Nikhil Ingewad and co-founders, Loop tackles a common blocker: each MCP tool schema bloats context and caps functionality in apps like Cursor or Claude Desktop. Instead, Loop fetches and executes tools on demand, letting devs plug in Slack, Gmail, Notion, Stripe, or even external MCP servers from a single dashboard. No more juggling configs or hitting tool caps. The goal is “one MCP, infinite tools, zero context bloat.”

🔗 Big props to SimpliflowAI — see website | view on Product Hunt and show your support

Future AGI Introduces Agent Compass For Debugging Agents

Future AGI, founded by Charu G. and Nikhil Pareek, has launched Agent Compass, a zero‑config evaluation tool that helps teams debug AI agents with speed and clarity. Instead of endless logs and dashboards, Compass clusters similar failures, identifies root causes, and suggests ranked fixes — all from just four lines of instrumentation code. Pareek says current observability tooling is like “flying rockets blind.” With Agent Compass, debugging is 98% faster, turning hours into minutes. Teams can now see not just what broke but also why and how to fix it.

🧭 Congrats to Future AGI — start free | take a look and upvote on Product Hunt

Datadog Launches Monocle For Real‑Time Metrics At Scale

Datadog unveiled Monocle, its new timeseries storage engine written in Rust. Replacing multiple older backends, Monocle unifies ingestion, storage and queries into a lightweight system that’s both faster and cheaper to run. Benchmarks show up to 60x higher ingestion throughput and 5x lower query latency compared with earlier generations. The system relies on a shard‑per‑core, async worker model with LSM‑tree persistence and Radix tree buffers to accelerate aggregations. Datadog engineers say Rust’s safety and concurrency provide the foundation to simplify operations across its platform. Even better, Monocle components are being reused across other Datadog systems for long‑term maintainability.

📈 Hats off to Datadog — read their blog

🔦 DevTool of the Week: Guepard

Not every startup begins in Silicon Valley. A year ago, Guepard’s co‑founders, KC ( 🐆 KC – Koutheir Cherni), Ghassen Menaouar and Hani C., were tinkering in a basement in Tunisia, debating how to make databases safe enough for AI agents to touch without fear. Every developer they spoke to had the same anxiety: “What if my AI assistant breaks production with one query?”

That problem became their mission. Guepard’s new Managed Cloud Platform (MCP), launched this week alongside a $2.1M pre‑seed raise, gives databases the ability to “time travel.” Any AI agent can spin up instant, disposable database branches, experiment freely, and then merge changes back safely. Think git‑style versioning, but for your data.

Backed by Schematic Ventures, UC Berkeley SkyDeck and angels from OpenAI, Datadog and Mistral, the team is betting this is the missing layer of agent infrastructure. KC, now based in San Francisco, frames it simply: “Removing fear from innovation. Making AI agents trustworthy. Letting developers dream bigger.”

The product is already being piloted with enterprises like Orange, giving teams the power to delegate work to AI without sleepless nights about corrupted tables or vanishing records. For companies trying to balance speed with safety in the age of AI development, Guepard promises a huge step forward: fearless agents and fearless developers.

🐆 Congrats to the Guepard team — see the Product Hunt launch | install the MCP server | explore at guepard.run

📆 Events and Conferences

Event season has launched. Here’s a round‑up of some of our favourite October events in the devtools and engineering space:

• DevOpsCon + Kubernetes Bootcamp | Sept 29 – Oct 3 | New York / Online Focused on CI/CD, Kubernetes, Platform Engineering and DevSecOps, this hybrid event combines talks and workshops with a dedicated expo (Sept 30–Oct 1). Hosted at the Marriott Brooklyn Bridge Hotel or available online.

• Sifted Summit | October 8–9 | London A two‑day gathering with 200+ speakers, workshops, and side events. Attendees get full conference access, networking tools via the event app, and even an afterparty. The Summit also marks the launch of the Sifted 250 leaderboard.

• Pacific NW Software Quality Conference | October 13–15 | Portland, Oregon and Online Celebrating its 43rd year, PNSQC 2025 focuses on “Quality Intelligence.” Workshops run Oct 13, with the two‑day technical conference following. Hybrid options let participants join in Portland or online.

• Kubernetes Community Day | October 21 | Edinburgh, UK A community‑driven event featuring talks, discussions, and networking for Kubernetes enthusiasts and professionals. Designed to foster collaboration and knowledge sharing across the Cloud Native community.

• TestCon Europe 2025 | October 21-24 | Vilnius + Online Europe’s largest software testing and QA event. The 10th edition explores automation, AI and IoT testing, Agile and DevOps workflows, with a broad program of talks, workshops, and networking opportunities.

Got a launch, raise, or tool we should feature next week?

Reach out to one of our team.

Louise Ogilvy (Recruitment Director) – UK/USA

Becca Combe MIRP CertRP (Lead Recruiter) – USA

Ben Chipchase (Lead Recruiter) – UK

Natalie Harper (Marketing Administrator)

Looking for your next DevTools role?

Visit devtoolsjobs.com