The Venture Codex

Led · Series B · Aug 2022

Rookout offers a developer-first observability platform focused on giving engineers live visibility into their code in production via dynamic instrumentation and “non-breaking breakpoints.” Its core features enable live debugging and the collection of runtime data, and the product integrates tracing standards like OpenTracing and OpenTelemetry as well as third-party logging services. The company emphasizes developer ownership of production behavior and supports workflows that surface business logic rather than just infrastructure metrics. Customers include Amdocs, Cisco, Dynatrace, Jobvite, Santander Bank and UPS. The team is relatively small (the article references 20 engineers) and the company says it has seen its revenue grow 20x since its last funding round in 2019. Rookout is positioning to scale its developer-focused observability offering and expand further into enterprise accounts. Rookout offers a production debugging platform that collects code-level data on demand without redeploying, supporting serverless (including AWS Lambda), containers, Kubernetes, and PaaS like Google App Engine and Elastic Beanstalk. The company emphasizes visibility into ephemeral platforms that are traditionally hard to observe. Over the past year customers have found new uses for Rookout’s code-level data collection capabilities, prompting the company to expand beyond pure debugging. Rookout plans to use its latest funding to broaden support and enhance its observability and pipelining features. The startup launched in 2017 and maintains offices in San Francisco and Tel Aviv. It has a team of about 20 employees and has raised more than $12 million to date. Rookout builds debugging tools for applications that run on modern container and serverless platforms. The company launched a live Kubernetes debugger feature that brings the equivalent of breakpoints to Kubernetes. Rather than stopping code execution, Rookout collects full-stack data so developers can inspect variable state with a few clicks. That design allows use in both development and production environments and aims to reduce reliance on complex logging and distributed tracing setups. CTO and co-founder Liran Haimovitch said the product restores the breakpoint-like visibility developers and DevOps need without halting apps. The company has raised $4.2M for its rapid production debugging platform.

Participated · Series B · Jul 2022

Deci develops a platform that builds and optimizes AI-powered systems, relying heavily on neural architecture search (NAS) to automatically discover lower-cost, high-performance models across text, audio and image tasks. The company says its technology can generate image-classification models that deliver more than 2x runtime improvement while improving accuracy, enabling deployment on CPUs rather than large GPUs. Deci was co-founded in 2019 by Yonatan Geifman, Ran El‑Yaniv and Jonathan Elial and has about 50 employees. Its customers include a videoconferencing provider that used the platform to roll out client-side background blurring; others use Deci to reduce inference and deployment costs. Deci has a strategic collaboration with Intel to optimize models on Intel processors, producing accelerated question-answering and image-classification models for Intel CPUs. The company faces technical limitations and evaluation challenges inherent to NAS and competition from firms like OctoML, Neural Magic and OmniML, even as it expands R&D and go-to-market efforts. Deci provides a deep learning platform that uses proprietary AutoNAC (Automated Neural Architecture Construction) technology to build and continuously update models optimized for available data and hardware. The company has focused on computer vision use cases and offers a family of models called DeciNets to reduce compute needed for inference. Deci also generates synthetic datasets to supplement training data and keeps customers' data within their own environments during model development. It has traction with Fortune 500 tech companies, including a major videoconferencing customer using Deci to run background-blur inference on edge, CPU-based devices, and collaborated with Intel on MLPerf to accelerate ResNet-50 on Intel CPUs. Deci plans to expand from computer vision into building better NLP models for voice and language applications. Financially, the company recently completed a $21M Series A and previously raised a $9.1M seed round a year earlier; the new capital will be used to expand product and customer bases. Deci offers an end-to-end platform that uses AI to optimize pre-trained deep‑learning models, benchmarking and tuning them for production. Engineers bring models into Deci, which manages optimization and packages models for deployment via a runtime container or Edge SDK. The company says Deci‑optimized models run five to ten times faster on the same hardware while maintaining comparable accuracy. Deci supports inference on both CPUs and GPUs and is already working with customers in autonomous driving, manufacturing, communications, and healthcare. The company frames its product as a way to make it easier and faster for businesses to take AI workloads into production and to improve model performance and accuracy. Deci was founded by Yonatan Geifman, Jonathan Elial, and Ran El‑Yaniv and is based in Tel Aviv.

Led · Seed · Feb 2022

SegmentStream offers a cookie-less, predictive attribution solution that applies proprietary machine learning to first-party user behavior to evaluate the incremental performance of marketing channels and campaigns. The platform generates “Probabilistic Conversions” and feeds insights back into ad platforms like Facebook Ads and Google Ads to automate ad optimization. The product aims to replace cookie-based approaches that are becoming ineffective for accurate ROAS calculation. The company highlights its focus on helping marketing teams gauge the value of channels and campaigns that don’t directly drive sales. SegmentStream was founded by Pavel Petrinich, Oleg Katryshev and Constantine Yurevich and is a TechStars alumnus. The startup currently serves over 100 customers worldwide, including KitchenAid, Nespresso, and Freshly Cosmetics, and has raised $2.7 million in seed funding. SegmentStream is a SaaS platform for analyzing and improving digital-marketing effectiveness using machine learning. Its core capabilities include multichannel ML attribution to determine the incremental contribution of each marketing source to final sales and automated budget reallocation to improve results. The company says more than 100 companies use its solutions, with clients such as Lenta, Petrovich, Nespresso, United Colors of Benetton, and Kaspi.kz. SegmentStream does not disclose full financials but reported that dollar monthly revenue has grown more than 38% since the start of 2021. Founded in 2018 by Konstantin Yurevich, Pavel Petrinich, and Oleg Katryshev, the startup is based in London and currently employs about 20 people. After the pandemic the company closed offices and moved to remote work. New funding will be used to grow technical and product teams in the CIS and to scale marketing and sales across the UK, the US, and Western Europe.

Participated · Series C · Nov 2021

Rescale provides a unified digital engineering platform that combines cloud high-performance computing resources, intelligent data management tools, and applied AI to speed modeling and simulation workflows. The company maintains a market-leading library of over 1,250 engineering and R&D applications and a global network of more than 500 cloud datacenters through partners such as AMD, AWS, Arm, Intel, Microsoft Azure, NVIDIA, and Oracle Cloud Infrastructure. Rescale serves hundreds of enterprise customers across aerospace, automotive, energy, life sciences, manufacturing, public sector, and semiconductors; customers including Arm, General Motors Motorsports, Samsung, SLB, and the U.S. Department of Defense spend over $1 billion annually on HPC infrastructure that the platform supports. Planned roadmap priorities include expanding available workflows and compute technologies, establishing a unified data fabric and digital thread for modeling and simulation, adding AI-native search, tagging, and automation, and strengthening enterprise security and compliance with additional government and defense authorizations. The company emphasizes democratizing AI-powered engineering and automating AI Physics tool use, targeting large speed improvements in design validation. Financially, Rescale has raised a $115 million Series D and has now brought total funding to over $260 million. Rescale delivers a platform that enables organizations to run scientific and engineering simulations on public clouds (AWS, Azure, GCP, IBM, Oracle) and manage on-premises HPC resources, schedulers, and licenses. Its network spans 8 million servers with over 80 specialized architectures and resources (e.g., Nvidia Tesla P100 GPUs, Intel Skylake processors, >1TB RAM) delivering a combined 1,400 petaflops of compute. The platform supports workloads across aerospace, automotive, oil & gas, life sciences, electronics, academia, and machine learning, and includes a recommendation engine trained on billions of core hours to optimize performance. Rescale offers on-demand and long-term computing environments and pricing for single batch jobs, optimization jobs, and large designs of experiments, and has expanded its software catalog to more than 800 apps. Since its most recent February funding round the company added over 100 new customers and now counts 200 enterprise subscribers and 400 subscribers overall, including several Fortune 50 businesses. CEO Joris Poort says the new proceeds will be used to grow Rescale’s platform, service offerings, and workforce (the company has 200 employees and expects to grow to 300 in a year). Rescale is a hybrid high-performance computing (HPC) cloud platform delivering intelligent computing for digital R&D. Led by CEO Joris Poort and CTO Adam McKenzie, the company serves more than 300 customers ranging from startups to Fortune 50 enterprises. Rescale enables research scientists and engineers to run simulation and artificial intelligence workloads on the cloud provider of their choice using specialized hardware architectures and software templates optimized for their use cases. The platform requires no setup or maintenance from users. Rescale closed a $50m Series C round, bringing its total funding to over $100m. The company stated it intends to use the funds. Rescale helps companies move legacy high-performance computing (HPC) applications to the cloud or hybrid environments, providing HPC resources and tuning applications for maximum performance. The platform supports bare metal, virtual machines and containers and can leverage on-prem capital assets while extending workloads to different cloud vendors. Rescale targets engineering, aerospace, scientific and other verticals and currently serves about 150 customers, including Sikorsky Innovation, Boom Aerospace and Trek Bikes. The company was founded in 2011 and is based in San Francisco, with roughly 80 employees. Financially, Rescale has raised a total of $52 million to date and recently closed a $32 million Series B. The company plans to use the new capital to hire developers and solutions architects, expand in Asia and Europe, and build relationships with systems integrators to broaden market reach. Rescale provides enterprise-class high performance computing (HPC) platforms that help large enterprises transition from legacy on-premise systems to scalable, agile cloud computing infrastructure. Its platform is deployed securely via a web-based application environment powered by simulation software providers and backed by large HPC infrastructure. The company serves global Fortune 500 customers across aerospace, automotive, life sciences, marine, consumer products and energy sectors. Led by CEO Joris Poort, Rescale has attracted notable seed backers and strategic investors. The company raised a $14M Series A and intends to use the funds to expand worldwide. Total capital raised to date exceeds $20M.

Participated · Equity · Sep 2021

Buyk operates a real-time retail service promising delivery of groceries and essential items to customers' doors in 15 minutes or faster, with no minimum spend and no delivery fee. The company uses hyperlocal dark stores stocked with roughly 2,000–3,000 SKUs and proprietary technology to optimize assortment, picking strategies, and delivery routes. Buyk is led by founders Rodion Shishkov and Slava Bocharov, who previously co-founded and ran Samokat, a large European ultrafast delivery operator. The service launched in Manhattan on September 14 and Buyk plans to expand across all New York City boroughs by the end of 2021 and to the largest U.S. metro areas in 2022. The company emphasizes tailoring assortments to hyperlocal demand, featuring local partners, and intends to develop a private-label offering. Buyk raised $46 million in a recent funding round from CM Ventures, Fort Ross, Citius and other investors ahead of its New York launch.