By admin 
In a rapidly evolving landscape where the deployment of Artificial Intelligence agents is accelerating across enterprises, a critical challenge has emerged: maintaining developer control and optimizing costs as these intelligent systems proliferate. Addressing this growing need, San Francisco-based B2B machine learning startup TrueFoundry has launched TrueForge, a new custom AI agent harness designed to empower developers with greater agency and reduce operational expenditures. Co-founded in 2021 by former Meta engineers, TrueFoundry has released TrueForge under the permissive MIT License, making it freely available on GitHub. This move signifies a strategic effort to foster wider adoption and collaboration within the AI development community, allowing for extensive customization, modification, and self-hosting of the harness.
TrueForge’s open-source nature means it can be seamlessly integrated with any preferred AI models, whether chosen by individual developers or mandated by their parent enterprises. This vendor-neutral approach liberates organizations from being locked into specific proprietary ecosystems, offering unparalleled flexibility in model selection and integration. The harness can be forked, adapted, and even incorporated into commercial products, a testament to its robust design and TrueFoundry’s commitment to open innovation.
The company highlighted the efficiency and cost-effectiveness of TrueForge in a detailed blog post, presenting compelling benchmark data. In their testing, TrueForge, when paired with the open-source GLM-5.2 Large Language Model (LLM), successfully completed 11 out of 14 tasks on DevRev’s Enterprise-Bench. This benchmark specifically evaluates multi-step tool utilization across complex enterprise systems such as Customer Relationship Management (CRM), issue tracking, and document management. The results were striking: achieving these outcomes with TrueForge cost a remarkable 75% less than utilizing Anthropic’s Claude Managed Agents harness, powered by the high-end Claude Opus 4.8 model. The incurred cost for TrueForge was a mere $2.90, starkly contrasting with the $11.80 spent on the Claude-powered solution for the same tasks.
Further substantiating these claims, TrueFoundry conducted a comparative analysis using the same underlying LLM, Opus 4.8, across both TrueForge and Claude Managed Agents. Even in this more direct comparison, TrueForge demonstrated a significant cost advantage, reportedly saving approximately 30% in operational expenses. The cost using TrueForge was $8.50, compared to $11.80 for Claude Managed Agents, underscoring the harness’s inherent efficiency in resource utilization and cost management. These figures are crucial for enterprises grappling with the escalating costs associated with large-scale AI deployments.
The decision by TrueFoundry to offer such a powerful and efficient harness as open-source might raise questions about their business model. However, Anuraag Gutgutia, TrueFoundry’s co-founder and COO, explained the strategic rationale in an exclusive interview with VentureBeat. "We’ve had this ask from a bunch of customers," Gutgutia stated. "You have an ability where you bring in agents and MCPs – can we also get something where you can actually launch these managed agents? I think that is the need we are satisfying. It is not a replacement. People will use this alongside other harnesses, like the cloud-managed ones or the commercial-provider-managed ones, but this will serve as a way for people to use them in a vendor-neutral way and also at a lower cost."
This statement reveals TrueFoundry’s broader vision. While TrueForge addresses the runtime and orchestration layer for AI agents, the company already offers a premium "AI Gateway" for enterprises. This commercial offering provides centralized control over model and Model Capability Provider (MCP) access, credentials, permissions, budgets, and observability. TrueForge, therefore, complements the AI Gateway by handling the sophisticated reasoning and execution loop – the process where an AI model repeatedly analyzes information, calls upon necessary tools, processes the results, and continues working until a task is successfully completed.
For enterprise developers, the practical implementation of TrueForge is designed for ease of use and scalability. They can initiate their work locally with a single command and a lightweight SQLite database. As their projects mature, they can seamlessly transition to a shared deployment environment using Docker Compose or Helm, leveraging more robust database solutions like Postgres and Redis. This incremental deployment strategy allows for rapid prototyping and iterative development without the immediate overhead of complex infrastructure setup. However, TrueFoundry explicitly cautions that the local configuration is intended solely for individual developer use and not for internet-facing production services, emphasizing the need for proper security and scaling measures in production environments.
Gutgutia articulated TrueFoundry’s overarching ambition: to establish its AI Gateway as the universal control layer beneath any chosen agent and harness infrastructure within an enterprise. "There will be a set of companies that will use our harness as the way to launch managed agents," he explained. "Others may continue using Claude, other open-source harnesses, or internal systems. But all that traffic should still be flowing through our gateway." This strategy positions TrueFoundry not just as a provider of an open-source tool but as a foundational platform for enterprise AI governance and management.
The core innovation of TrueForge lies in its sophisticated approach to context management, aiming to minimize waste and optimize the efficiency of LLM interactions. The harness intelligently controls the amount of information fed back into the model at each step of an agent’s operation. This includes delaying the loading of MCP tool schemas until they are actively required, delegating isolated tasks to specialized sub-agents, and intelligently handling large tool results by saving them to files rather than overwhelming the active context window. Furthermore, TrueForge processes structured results through dedicated code execution and automatically compacts long-running conversations to maintain efficiency. The default context compaction threshold is set at 50,000 tokens, a parameter that can be adjusted on a per-agent basis to suit specific needs.
TrueForge also differentiates itself in its handling of the sandbox environment. Unlike runtime systems that maintain an agent within an isolated environment for its entire duration, TrueForge operates a core agent loop on its server. A sandbox is provisioned only when the agent needs to execute code or interact with files. This dynamic allocation of resources significantly reduces unnecessary compute overhead and allows a single server to manage a greater number of agents concurrently. TrueFoundry asserts that these architectural choices directly contribute to reducing overall model spending, a critical factor for cost-conscious enterprises.
To provide a clearer picture of TrueForge’s positioning, a comparative analysis against Claude Managed Agents and other leading orchestration harnesses is essential. While the provided data does not include a comprehensive table with all fields populated, the core differentiators can be inferred. Claude Managed Agents, for instance, are a proprietary solution from Anthropic, offering managed services with a focus on their proprietary LLMs like Claude Opus. Their licensing is closed, and pricing is based on API usage, which can become substantial for complex, multi-step tasks. Model flexibility is generally limited to Anthropic’s offerings, and deployment is managed by Anthropic.
In contrast, TrueForge, with its MIT license, offers maximum flexibility. It supports a wide array of AI models, including open-source options, and allows for self-hosting and extensive modification. Its pricing, when integrated with the AI Gateway, is designed to be more economical, particularly for high-volume or complex operations. The deployment options are also more diverse, catering to both local development and scalable cloud deployments. The key differentiator for TrueForge is its focus on developer control, cost optimization through efficient context management and resource allocation, and its open-source, vendor-neutral nature.
A crucial distinction for enterprise adoption lies between TrueForge operating independently and when it is integrated with TrueFoundry’s commercial AI Gateway. While the open-source harness provides the core agent runtime capabilities, it does not inherently enforce enterprise-level access policies or governance. "If you are using just the open-source version of our agent harness, yes, you will need to put the right controls therein or in front of some other internal control system," Gutgutia advised. This means that organizations utilizing the standalone harness must implement their own security, access control, and compliance measures.
However, when paired with TrueFoundry’s AI Gateway, agents can inherit and leverage existing enterprise identity and access management frameworks. The gateway acts as a central enforcement point for Single Sign-On (SSO), identity providers, and granular permissions, ensuring that access to models, MCP servers, tools, skills, and other agents is managed consistently and securely across the organization. This integration transforms TrueForge from a developer tool into a governed enterprise AI component. This distinction is paramount for platform engineering teams evaluating the project; TrueForge is free software, but the commercial control plane provided by the AI Gateway is where TrueFoundry derives its revenue and offers enterprise-grade governance.
Early adopters and beta users have already demonstrated the practical benefits of TrueForge. NetApp, a significant enterprise storage and data management company, was a beta user and actively contributed to TrueForge’s development by providing essential requirements. NetApp’s IT organization has reportedly leveraged the technology for critical functions such as incident response and expedited ticket triage, while also enabling developers to access internal agents as self-service tools. Automattic, the company behind WordPress.com, is also cited as an early adopter, highlighting the harness’s appeal across diverse technology sectors.
TrueFoundry’s journey began in 2021, with a foundational mission to assist enterprises in deploying and managing machine learning models. Their initial focus was on providing robust solutions for Kubernetes-based model serving, training, and infrastructure management. The company was founded by Nikunj Bajaj, Abhishek Choudhary, and Anuraag Gutgutia, individuals with prior experience at prominent tech organizations like Meta and WorldQuant. Their shared insight was the critical need for mature, control-oriented systems as AI transitioned from research to production environments within large corporations. As generative AI gained traction, TrueFoundry strategically expanded its offerings beyond traditional MLOps to encompass the management of LLM applications and the intricate ecosystem of models, tools, and agents that support them.
By 2025, the AI Gateway had become a central pillar of TrueFoundry’s business strategy. This gateway serves as a crucial intermediary layer, connecting enterprise applications with model providers. It manages essential functions such as request routing, authentication, access controls, observability, budget tracking, and failover mechanisms. This strategic pivot towards comprehensive AI governance has been supported by substantial external investment. In February 2025, TrueFoundry secured a $19 million Series A funding round, led by Intel Capital, with participation from existing investors Eniac Ventures and Peak XV’s Surge, as well as Jump Capital and prominent angel investors like Gokul Rajaram and Mohit Aron. This funding brought the company’s total financing to approximately $21 million, as detailed in an Intel Capital announcement, signaling strong investor confidence in their vision for enterprise AI.
At the time of the Series A announcement, TrueFoundry reported a fourfold year-over-year increase in its customer base and was managing over 1,000 clusters dedicated to machine learning workloads. The company’s trajectory continued to accelerate, with an increasing focus on large-scale enterprise AI traffic. In January 2026, VentureBeat reported on TrueFoundry’s TrueFailover launch, an automated solution for rerouting enterprise AI traffic. By then, the company had surpassed 30 paid customers globally, achieved over $1.5 million in annual recurring revenue (ARR) in the preceding year, and was processing more than 10 billion requests monthly through its AI Gateway.
TrueFoundry’s client roster includes notable organizations such as NetApp, Siemens Healthineers, ResMed, Automation Anywhere, Nvidia, and Games24x7. Gutgutia further highlighted collaborations with Fortune 1000 companies including NetApp, Siemens, Synopsys, and Automation Anywhere during his interview. In a move to further consolidate its market position and enhance its agentic AI capabilities for enterprise clients, TrueFoundry also made a strategic acquisition in June 2026, purchasing UK-based Seldon AI. Seldon AI, a long-standing vendor in the MLOps space, was known for its Seldon Core software, widely used for production model serving and inference.
This acquisition underscores TrueFoundry’s strategy of unifying traditional ML, LLMs, tools, and agents under a single, cohesive deployment and governance framework. While the company’s commercial focus has historically been on the control plane beneath enterprise AI workloads – managing access, routing, policy enforcement, cost monitoring, and service availability – the introduction of TrueForge extends this strategy to the agent runtime itself. TrueForge provides an open-source execution layer above the existing control plane. Gutgutia emphasized the complementary nature of these offerings: organizations can deploy TrueForge independently or continue using alternative agent harnesses, while TrueFoundry’s long-term business objective remains to provide the overarching governance and infrastructure that underpins any chosen agent deployment strategy for enterprises.
