From AI Agents to 3D Worlds, These Tools Know What Devs Need

Welcome to this week’s DevNewsletter, our small but growing roundup of last week’s devtools news. Focussing on start-ups and scale-ups where AI and developer ingenuity take centre stage. These teams are proving that innovation thrives when tools fit real needs.

 

Funding Wins

Graphite Secures $52M and Spins out Diamond to Squash Coding Bugs Faster

Graphite began as an internal code review tool which is now trusted by 500+ companies including Shopify, Snowflake, and Figma! Last week, the team announced $52M in Series B funding and introduced Diamond, a new standalone AI product built to catch bugs and code errors automatically. Founded by Tomas Reimers, Greg Foster , and Merrill Lutsky in 2020, Graphite saw revenue grow 20x in 2024 and has made its core platform free for teams of any size. With Diamond in the mix, they’re doubling down on their mission to make code review faster, smarter, and less painful.

Explore what’s next for Graphite and Diamond

Rerun Raises $17M to Build the Data Stack for Physical AI

Since 2022, Rerun‘s mission has been to help robotics and spatial computing teams make sense of their data; from real-time sensor streams to offline training pipelines. Their open-source framework is already used by the likes of Meta, Google, and Hugging Face, and now they’ve secured $17M in seed funding to build a full cloud data platform around it! The round was led by Point Nine, with backing from Seedcamp, Costanoa, and angels like Guillermo Rauch and Nicolas Dessaigne. Co-founder Nikolaus West said it best:

“Physical AI has gone from intractable to inevitable.”

Rerun’s tools make complex visualisation, querying, and experimentation seamless, speeding up the path from messy logs to smart machines.

See how Rerun is powering the future of Physical AI

Confident AI Lands $2.2M to Help Teams Test LLMs with Confidence

Confident AI (YC W25), the team behind DeepEval (4.3k GitHub stars and over 400k monthly downloads), has raised $2.2 million in seed funding to improve how teams benchmark, test, and safeguard LLM applications. Co-founded by Jeffrey Ip and Kritin Vongthongsri, the startup was bootstrapped for a year before joining YC’s W25 batch, then closed its round in just five days. Backers include Y Combinator, Flex Capital, January Capital, Rebel Fund, and Oliver Jung. Ip writes:

“We’re going to double-down on providing better LLM evaluations for everyone… and this includes not just testing for functionality, but also for LLM safety.”

Find out how Confident AI is raising the bar for LLM testing

Nunu.ai Raises $6M to Power AI Agents for Game Testing

nunu.ai, the YC-backed startup building AI agents that can test and play any game, has raised a $6 million seed round co-led by a16z Speedrun and TIRTA Ventures. Founded in 2022 by Nicolas Muntwyler GDC, Kyrill Hux, and Jan Schnyder GDC, the team is aiming to automate QA testing with multi-modal agents that can adapt to any virtual environment. With this round, Nunu.ai plans to scale its infrastructure and grow its team as it moves beyond gaming into broader embodied AI applications. The new funding brings their total to $8 million following a $2 million pre-seed in 2024.

Head to nunu.ai to learn more about their vision.

 

3, 2, 1… Launches

SciPhi Brings Agentic Intelligence to your Data with R2R

R2R from SciPhi is an open-source agentic retrieval system that transforms how teams uncover insights across complex document collections and the web. Traditional RAG approaches fall short when it comes to multi-step reasoning and synthesis, R2R changes that with its Deep Research API. It combines citation-backed answers, knowledge graph generation, and transparent agent workflows, all accessible via a clean REST API. Ideal for enterprises working with diverse data types, without the infrastructure burden. Congrats to Owen Colegrove and the SciPhi team on the launch!

Explore R2R and sign up here.

Roblox Creates Cube to Life with Open Source 3D Object Generation

Roblox has launched the first version of Cube, its AI-powered 3D generation model, now available both in-platform and as open source. Cube lets developers create “meshes” or 3D representations of objects like cars or scenery from a simple prompt, with tools to refine them directly in Roblox Studio. It’s designed to speed up creation for indie devs and studios alike, and lays the foundation for future “4D” experiences with interactive AI-generated scenes. Roblox also announced upcoming AI features for text, voice, and speech, making its world-building toolkit even more powerful.

Read the full announcement here from Anupam Singh (VP of Engineering) and Nick Tornow (VP of Creator Engineering)

Sideko Sweeps Away API Chaos with Fully Automated Developer Tools

Brooklyn-based Sideko has launched a powerful new platform that takes the pain out of managing API programs at scale. The platform automates the creation and maintenance of API documentation, SDKs, mock servers, and CLIs, keeping everything in sync whenever an API changes. It’s already in use by over 200 companies including Prudential, and is particularly well-suited for enterprises adopting AI agents, thanks to its robust “action layer” and support for hybrid deployments. Less time wrestling with docs, more time building. Patrick Kelly, CEO and co-founder of Sideko said:

“The role of APIs has fundamentally shifted from being a technical integration point to becoming the primary way enterprises deliver products….Our platform ensures that this new front door to companies’ services is welcoming and efficient for developers.”

Learn more about Sideko’s launch and what it means for enterprise dev teams.

Unify Builds a Flexible Framework for Observability without the Fluff

Frustrated by bloated, rigid observability tools, the team behind Unify has launched a lightweight, fast, and fully hackable observability framework for products with or without LLMs. Founded by Daniel Lenton, Unify lets you track exactly what matters using just three primitives: tables, views, and plots. Developers can compose flexible interfaces for monitoring, evaluation, and optimisation without being boxed in by rigid dashboards. The core building block is simple and from there, users can visualise anything from LLM evaluations to production traffic in seconds. It’s designed to get out of your way so you can ship faster, stay focused, and only track what actually matters.

Try Unify and get one month free via Product Hunt

Timescale Launches Major Upgrades to Make Postgres Lightning Fast at Scale

Last week, Timescale rolled out a series of updates aimed at making Postgres faster, more scalable, and better suited for real-time analytics—no sharding or bolt-ons required. Highlights included new support for secondary indexes on compressed data (with 1,185x faster lookups), bulk triggers for hypertables, and proof that a single Timescale instance can handle 2PB of data and 1 trillion metrics per day.

If you’re not already following co-founder Ajay Kulkarni on LinkedIn, he’s well worth keeping up with. Catch the full breakdown from Timescale Launch Week in their detailed wrap-up here.

Pruna AI Makes Its Model Optimisation Framework Open Source

Pruna AI has open sourced its model optimisation framework, giving developers an all-in-one toolkit for compressing and fine-tuning AI models. It supports methods like pruning, quantisation, caching, and distillation, and helps users evaluate both the performance gains and quality trade-offs. Co-founder and CTO John Rachwan explains:

“We are similar to how Hugging Face standardized transformers and transfusers… We are doing the same, but for efficiency methods.”

While many open-source tools focus on just one method, Pruna combines them into a cohesive framework – already in use by teams like Scenario and PhotoRoom.

Explore the details on Pruna’s official blog.

Briefer Launches AI Analyst Built to Think Like a Teammate

Briefer (YC S23) has launched its new AI analyst; an intelligent agent that sits inside its popular data workspace and helps teams turn questions into clear, auditable insights. Unlike typical text-to-SQL tools, this one can write SQL, plot graphs, run Python, and incorporate your business context and metadata to improve accuracy. Because it works directly inside Briefer notebooks, you can follow every query and action step by step, no black boxes here. Built for both technical and non-technical users, it’s a powerful way to scale data analysis across a team.

Congrats to founders Lucas da Costa and Lucas Vieira! Try the Briefer Agent by booking a demo at briefer.cloud.

Artie Announces Enterprise Solution with Real-Time Replication at Petabyte Scale

Artie (YC S23), founded by Jacqueline Cheong and Robin Tang, has just launched its enterprise-grade platform for high-volume, low-latency data replication. Built for teams managing massive datasets and mission-critical workloads, Artie for Enterprise replaces brittle ETL jobs with always-live data syncs across warehouses and hybrid deployments. New features include Multi-Step Merge for faster updates, Terraform support for painless scaling, and Datadog integration for deep visibility. It’s designed to make data movement faster, safer, and easier to manage, even in highly regulated environments.

Explore the full announcement on the Artie blog.

 

DevTools of the Week

Keywords AI Shines with Hot New Updates

Keywords AI (YC W24), founded in 2023 by Andy Li, Hendrix Liu, and Raymond Huang, is a platform for prompt engineering, observability, and evaluations, helping developers ship AI features fast. Last week, it stole the spotlight with a string of impressive updates. Part of Y Combinator’s Winter 2024 batch, it’s already helping over 40 YC startups ship solid AI features quickly.

It solves real developer headaches, like managing infrastructure and debugging LLM workflows with tools like a playground for 350+ models, request logging, and usage analytics.

Just days ago, their OpenAI Agents SDK integration launched, spiking signups by 50% on day one and doubling them over the weekend—proof they’re nailing developer needs. Co-founder Hendrix Liu highlighted how user feedback is fueling their roadmap. Then came API keys management via API, letting devs create and tweak keys with limits and expiry dates—no UI hassle. Add to that custom properties for faster log and dashboard searches, and it’s clear they’re on a roll.

Keywords AI is proving it’s a developer’s ally for taming AI chaos. These updates caught our eye this week—check them out at keywordsai.co and see why they’re worth your time!

From clever AI twists to tools that just get it, these teams are firing on all cylinders, proving brilliance lies in solving the right problems well. See you next week for more dev magic!

Next week will be highlighting devtool events in April, so if you have an event you would like to advertise or news to publish, please send us a message.

And as always, if you’re thinking about your next role, devtoolsjobs.com is your place to be.

Have a brilliant week!

Louise Ogilvy (Recruitment Director)

Becca Combe MIRP CertRP (Lead Recruiter)

Ben Chipchase (Lead Recruiter)

Natalie Harper (Administrator)