HHUD builds a Runtime Code Sensor that links live production software with AI-powered coding assistants. Developers add a lightweight SDK to their apps to stream real-time information about how code behaves—such as errors, latency, performance, and how functions are used—directly to AI tools like GitHub Copilot and other code generators. The product works by feeding ongoing, context-rich data from a running system into AI models so generated code can be aware of real-world conditions rather than just static code. This makes AI-generated code safer and more reliable in production. HUD differentiates itself by enabling production-aware development, providing continuous feedback from live environments rather than relying on static analysis or old logs. The company’s goal is to help software teams build higher-quality software faster with fewer bugs and better performance by bridging the gap between AI-generated code and real-world usage.
Work here?Claim your company
Industries
Company Size
51-200
Company Stage
Early VC
Total Funding
$21.2M
Headquarters
San Francisco, CA
Founded
2025
See people who can refer or advise you
Help us improve and share your feedback! Did you find this helpful?
Total Funding
$21.2M
Above
Industry Average
Funded Over
2 Rounds
ExpertCrowd, Toptal's platform for training and improving AI models, has teamed up with HUD, a Y Combinator-backed startup, to release ExpertCrowd-SWE. The software engineering benchmark evaluates frontier-grade large language models (LLMs) and coding agents. Toptal's experts wrote the tasks and grading rubrics. HUD's infrastructure hosted the task environments and verified the benchmark through quality assurance pipelines. Researchers from both companies, including more than a dozen PhDs, reviewed the methodology and results. "ExpertCrowd puts practitioners at the center of defining the work AI systems should perform and judging how well they perform it," said Taso Du Val, founder and chief executive officer of Toptal. The benchmark is the first release of the collaboration. More initiatives are in development.
ExpertCrowd and HUD Collaborate to Advance AI Training and Evaluation, Release SWE Benchmark. SAN FRANCISCO, Nov. 6, 2026 /PRNewswire/ - ExpertCrowd, Toptal's platform for training and improving AI models and systems, today announced a collaboration with HUD, a Y Combinator-backed startup that provides tools for building AI training data through its DataVendor(TM) marketplace. Together, the teams released ExpertCrowd-SWE, a software engineering benchmark that evaluates the performance of frontier-grade LLMs and coding agents. As AI systems take on increasingly complex work, measuring their capabilities requires tasks that reflect professional practice and grading criteria that capture what successful execution actually means. This collaboration brings together ExpertCrowd's expert-led task development and HUD's infrastructure for running complex RL environments. ExpertCrowd and HUD co-developed ExpertCrowd-SWE, bringing together engineers from Toptal's global network of highly skilled domain experts with HUD's experience collaborating with frontier research teams. After Toptal's experts authored tasks and grading rubrics, HUD's infrastructure hosted task-run environments and verified the benchmarks against internal quality assurance pipelines. HUD and Toptal's AI research teams, composed of more than a dozen PhDs and researchers with educational or professional experience at Harvard, EPFL, the University of Liège, DeepMind, Qwen, and ByteDance, reviewed and assessed the methodology and results to build one of the most robust software engineering benchmarks in the industry. "ExpertCrowd puts practitioners at the center of defining the work AI systems should perform and judging how well they perform it. By combining that specialized expertise with HUD's testing infrastructure, we're advancing what AI systems can achieve and setting a higher standard for real-world performance," said Taso Du Val, founder and CEO of Toptal. ExpertCrowd-SWE is the first release of an ongoing Toptal-HUD collaboration, with additional initiatives in development and further announcements coming soon. Building on this foundation, ExpertCrowd and HUD will bring more specialized human expertise into the tasks, environments, and evaluations that support the next generation of AI. "We built HUD's infrastructure to give research teams the tools to create realistic RL training environments, run evaluations across AI models, and inspect model performance. ExpertCrowd put that infrastructure to work, bringing the professional expertise to translate complex, real-world workflows into rigorous tasks, reference outputs, and evaluation frameworks that test the limits of today's leading AI models," said Jay Ram, founder and CEO of HUD. About ExpertCrowd ExpertCrowd is Toptal's platform for training and improving AI models and systems through RLHF, RL data generation, data labeling, and model evaluation. It is powered by Toptal's global talent network, with more than 30,000 specialized domain experts across 140+ countries, including more than 2,000 AI and big data specialists, 12,000 professionals with advanced degrees, and 1,000 PhD-level AI specialists, available on demand. For more information, please visit ExpertCrowd.com. HUD is a platform for building reinforcement learning environments, used by startups, Fortune 500 companies, and frontier labs. It provides infrastructure for training models, evaluating AI agents, and creating the post-training data that aligns AI to organizations' work. HUD's mission is to build reliable, fair, and open infrastructure for AI data with the goal of making data valuable for the people who create it and trustworthy for the labs that train on it. DataVendor(TM) is HUD's marketplace, which allows AI labs to license training and evaluation data directly from the companies that created it. For more information, please visit HUD.ai. FONTE Toptal
Hud launches agentic rollouts powered by function-level runtime intelligence. FinanceWire Oct. 29, 2026, 10:00 AM New York, United States, October 29th, 2026, FinanceWire Hud has released agentic rollouts, a capability that uses function-level runtime data to validate code changes from the pull request stage through production. It lets coding agents ship changes when they are safe and gives those agents direct feedback from production when something goes wrong. Understanding how it works starts with understanding what Hud measures. The Runtime Code Sensor At the core of the system is Hud's runtime code sensor. Teams add it to their backend services with a one-line SDK. Once running, the sensor automatically learns how each service behaves in production, down to individual functions. That function-level baseline is central to Hud's approach. Hud knows how specific functions normally behave under real conditions, with real parameters and real code flows. Every later decision in an agentic rollout draws on that baseline. What Makes a Rollout "Agentic" Hud describes an agentic rollout as a rollout built for agents. Coding agents can already plan, write, and test code and open a pull request in minutes. What they have lacked is any view of production. They cannot see how their code actually runs once it ships, which makes it difficult for a team to let an agent carry a change all the way through. Getting to production has depended on human-driven validation, which lengthens time to delivery. Agentic rollouts supply that missing view. Agents and engineers work from the same runtime context, so decisions about merging and shipping rest on evidence from production rather than on a reviewer's limited time. Stage One: Before the Merge When a pull request is opened, Hud analyzes it against a production impact map. The map reflects the real parameters and code flows Hud has observed in production. From that analysis, Hud classifies the change's production risk level. A change classified as safe can be merged at the speed it was written. When a change is classified as risky, Hud provides clear context about it, which helps reviewers move through the review faster. Stage Two: As Traffic Shifts From the moment the new version hits production, Hud closely watches its behavior at the function level. As it takes traffic, Hud continuously compares its behavior with the learned baseline. Because the comparison runs at the function level, a regression is caught immediately, and Hud can trace it to both its root cause and the pull request that introduced it. Stage Three: When Something Regresses If a regression does occur, Hud's context supports rolling back the release, reverting the change, and fixing the underlying issue. The coding agent receives deep context about the failure, which helps it avoid making the same mistake in future changes. Taken together, the three stages give teams knowledge of each change's production-level impact before, during, and after the rollout. A Working Example at Riverside Riverside, which provides a platform for recording and editing studio-quality audio and video, shows why function-level runtime data matters. Much of Riverside's backend is real-time media code, and that kind of code only shows its true behavior under production load. Riverside also ships fast with AI. The company routes every deploy through a canary alongside Hud's agentic rollout and automatic rollback. Flawed deploys are detected quickly and reverted in about eight minutes, usually before a customer notices. Shilo Magen, VP of R&D at Riverside, explained the impact: "Our engineers use AI agents every day, and writing code is no longer the bottleneck. The real challenge is shipping that code to production with confidence. For a real-time media platform like Riverside, even a small regression can have a major impact on the user experience. With Hud, we can automatically detect regressions and roll back problematic releases in minutes. Our engineers now spend less time worrying about deployments and more time building." The Engineering Problem Behind the Design The design responds to a measurable pattern in AI-assisted development. Per Faros AI's 2025 vendor survey, teams with high AI adoption merge 98% more pull requests, while review time rises 91%. Put plainly, engineers using AI heavily produce about twice as many pull requests, and each takes about twice as long to review. Meanwhile, Harness's 2025 vendor survey found that only 6% of organizations describe their continuous delivery as fully automated. The technical consequence is that the fastest part of engineering now feeds directly into the slowest. A pull request waits for someone willing to merge it. After deploy, someone watches dashboards. Hud describes this as the missing link in the AI software factory: agents build quickly, but they cannot see production. Roee Adler, Co-Founder and CEO of Hud, described the gap the technology is meant to fill: "AI already changed how code gets written. The bigger problem now is getting that code into production safely, at the speed agents write it. Every engineering leader I talk to is trying to solve this problem. Coding and triage have agents, but the merge and rollout still run on people, because it's scary to give coding agents accountability over the stability and reliability of the production environment. That's the missing link in the AI software factory, and it's the problem we built Hud to solve. With agentic rollouts, agents and engineers work from the same runtime context, at the function level, so a change is merged to production based on evidence. When something does regress, it's immediately detected and fixed, but more importantly the agent that wrote it gets a feedback loop to learn and improve." The feedback loop is also where the technology is headed. Each rollout teaches the coding agent how its changes behave in production, so later changes ship with less risk than earlier ones. Contact. Editorial Team [email protected] Sponsored Financial Content
ClickHouse and Hud connect the dots between AI code, production data and engineering decisions. New York, USA, August 13th, 2026, FinanceWire The promise of AI-powered development has largely centered on speed. Coding agents can generate changes quickly, allowing engineering teams to increase the amount of software they produce. But faster code generation also creates a new pressure point: teams need to understand those changes quickly enough to decide whether they belong in production. Hud says AI now generates or assists with 42% of the code developers ship, with that figure expected to reach 65% by 2027. The company's new integration with ClickHouse is built around the idea that production behavior can provide some of the context needed to manage that acceleration. The integration connects ClickStack, ClickHouse's open-source observability stack, with Hud's Runtime Code Sensor. Rather than treating observability as a separate troubleshooting function, the companies are connecting it with workflows for assessing, releasing, and fixing AI-generated software. From Infrastructure Signals to Code-Level Context The distinction between the two platforms is central to the integration. ClickStack provides broad visibility across applications and infrastructure, helping teams determine the service, deployment, or endpoint associated with an issue. Hud operates at the code layer, following function-level behavior and connecting production activity to the changes behind it. Shared trace IDs allow a coding agent to connect the two systems. An engineering team investigating an issue in ClickStack can move directly toward the corresponding code-level context in Hud. "Our users already trust ClickHouse to store and query their Open Telemetry data at scale," said Mike Shi, Head of Observability of ClickHouse. "The shift now underway is from simply watching systems or investigating issues to using that data to make day-to-day engineering decisions. Hud connects observability data to the code and changes behind it, making the entire stack more useful for teams building and shipping software with AI." That approach expands the role of production data. Instead of becoming useful only after a deployment goes wrong, runtime information can help inform decisions about changes before they are released. Making the Release Process More Dynamic The integration supports pre-deployment risk assessment for code changes. Teams can evaluate changes against real-time information about how affected code behaves in production, with higher-risk changes held for additional review and deeper context while safer changes can move faster or be automatically merged. The workflow continues after deployment. ClickHouse and Hud support release verification, automatic reversion upon regression, automated detection and investigation, and agentic workflows that can create pull requests to fix underlying issues. Rollback and remediation workflows are also part of the integration. That means the same connection can support several decisions around a single change: whether it should ship, whether it is behaving as expected after deployment, and what should happen if production behavior diverges from expectations. "AI is accelerating how quickly teams can generate code, but safely shipping it with high confidence requires production context," said May Walter, CTO of Hud. "Hud and ClickHouse bring real production behavior at an unparalleled breadth and depth, so teams can build a production-aware AI SDLC: gating changes before they ship, proactively verifying them once deployed, and fixing issues as they arise - all using real runtime truth. Together with ClickHouse, we are bringing that intelligence across the entire AI SDLC." The Case for a Production-Aware AI SDLC The practical use case becomes clearer when a seemingly small change produces an unexpected production result. A query can slow down, a function can behave differently under a particular workload, or a code path can consume more resources. Hud is designed to detect these issues at the function level and provide forensic context into their causes. ClickStack contributes the wider operational view, helping teams connect the code-level behavior to what is happening across the application. "Like every modern engineering organization, a growing share of our code is now written with AI," said Rom Kadria, Senior Software Engineer, monday.com. "We write code much faster, but the challenge has shifted to shipping just as quickly while maintaining confidence that new code won't cause harm. ClickHouse gives us the wide operational picture at scale, while Hud gives us the runtime intelligence and next-level introspection needed to evaluate and ship AI-generated code confidently. When issues do arise, combining ClickHouse and Hud allows us to triage and resolve them quickly. For a company building with AI, that combination is the obvious choice." Engineering teams can get started by installing the Hud SDK and connecting it to their ClickStack service. Hud's runtime intelligence then flows alongside the OpenTelemetry data teams already collect. The broader proposition is a shift in where production information sits within software development. As coding agents generate more of the code, ClickHouse and Hud are attempting to make real-world application behavior part of the feedback loop that determines how that code is reviewed, released, monitored, and repaired. Contact. Editorial team [email protected]
Hud brings on Shai Alani to expand Runtime Intelligence category for ai-native engineering. The software industry has spent the past several years focused on accelerating development through artificial intelligence. AI coding assistants can now generate code in seconds, helping engineering teams deliver features faster than traditional development cycles allowed. But generating software is only part of the equation. Once applications are deployed, engineers still face the challenge of understanding how that code behaves in production, and why it sometimes fails. Hud is positioning that challenge as the foundation of a new software category it calls Runtime Intelligence. To help establish that market, the company has appointed Shai Alani as Vice President of Marketing. Beyond code generation. AI has fundamentally changed software creation by reducing the time required to write code. According to Hud, however, the industry's tooling has not evolved at the same pace in understanding production behavior. Conventional observability platforms can alert teams when an application encounters an issue, but determining the precise cause often requires combing through logs and integrating information from multiple sources. AI coding agents face similar limitations because they can analyze source code without necessarily understanding how it executes in real production workloads. Hud's platform is designed to bridge that disconnect by providing function-level production evidence and forensic context whenever software encounters problems. The goal is to enable engineers and AI coding agents to investigate incidents, identify root causes, validate changes, and deploy fixes with greater confidence. "AI has changed the speed of software creation, but production is still where code proves itself," said Roee Adler, Co-founder and CEO of Hud. "The next major category in the AI SDLC is Runtime Intelligence: production behavior resolved to the function level, coupled with deep forensics when things go wrong, so humans and agents can understand, fix, and validate software with confidence. Shai brings the experience we need to build that category and scale Hud into a defining company for AI-native engineering teams." Hud's latest executive appointment reflects more than an expansion of its leadership team. Alani will oversee global marketing strategy while leading the company's efforts around category creation, brand development, and demand generation. His background includes serving as VP of Marketing at Lightrun, as well as previous leadership roles at Coralogix and Aporia. At Hud, his responsibilities will include communicating why runtime evidence may become an increasingly important component of AI-assisted software development. Alani believes the industry is approaching a point where producing software quickly is no longer the primary challenge. "Runtime Intelligence is the missing layer in the AI software stack," said Shai Alani, VP Marketing at Hud. "AI has made it easy to generate code, but it has not made it any easier to stand behind that code once it is running in production, where reliability is actually decided. That gap is fast becoming one of the defining problems for AI-native engineering teams, and it is exactly the kind of category you build a company around. That is why I joined Hud, and it is the story I am excited to take to market." Integrating production into the AI development loop. Hud says its Runtime Intelligence platform operates continuously in production through a runtime code sensor that runs alongside every function. When software behaves unexpectedly, the platform captures detailed forensic information that can help explain what occurred while providing the evidence needed to validate a solution. Rather than requiring developers to reconstruct events from scattered telemetry, Hud aims to bring production behavior directly into the software development process. The company says this allows engineering teams to resolve incidents more efficiently while giving AI coding agents a stronger factual foundation for recommending fixes. The platform is already deployed across millions of production services at organizations including Monday.com, Lemonade, Axonius, and Cyera. Hud has also raised $21 million in funding led by Aleph and SquarePeg. As AI continues reshaping software engineering, attention is increasingly shifting from how quickly code can be written to how confidently it can be deployed and maintained. Hud's latest leadership appointment underscores its effort to define Runtime Intelligence as an essential part of that next phase.
Find jobs on Simplify and start your career today
Don't see your dream role? Check out thousands of other roles on Simplify.Browse all jobs
Industries
Company Size
51-200
Company Stage
Early VC
Total Funding
$21.2M
Headquarters
San Francisco, CA
Founded
2025
Find jobs on Simplify and start your career today
Don't see your dream role? Check out thousands of other roles on Simplify.Browse all jobs