INNODATA INCINOD

Innodata focuses on AI agent trust, safety evaluation and enablement for federal and enterprise use.

“The appointment also builds on the momentum of Innodata’s Federal Practice, announced last Fall, which brings the Company’s data engineering, model evaluation, and AI enablement capabilities to U.S. government and defense-related missions. It also underscores Innodata’s deepening focus on AI agent trust and safety — a capability the Company is bringing to both Federal missions and enterprise adopters of agentic AI. As government agencies and enterprises move from experimenting with AI to deploying autonomous agents at scale, Innodata believes independent evaluation and safety assurance will be essential to trusted adoption. Admiral Rogers’ experience at the intersection of national security and cyber operations is expected to help guide the Company’s work in this critical area.”

Innodata provides training data, alignment, evaluation and operationalization for agentic AI systems.

“Today, leading AI innovation labs and Big Tech companies (including five of the so-called “Magnificent Seven”) building frontier generative AI models and leading enterprises engage us to provide (i) training and post-training data development; (ii) alignment and preference optimization; (iii) capabilities, alignment, and safety evaluation; and (iv) AI enablement and operationalization, including support for agentic and tool-using systems.”

Frontier agentic systems require specialized behavioral and agentic evaluations.

“Early AI evaluation focused primarily on output correctness. In contrast, today’s frontier systems - particularly agentic and tool-using systems - require behavioral and agentic evals that assess how models plan, reason, and act over time. These evals may examine reasoning coherence, tool selection and invocation, multi-step task execution, adherence to system instructions, and robustness under adversarial or ambiguous inputs.”

Innodata discusses importance of structured human judgment for agentic AI evaluation and deployment readiness.

“This shift toward agentic evaluation materially increases the importance of structured human judgment, domain expertise, and scalable evaluation operations. We believe that the ability to measure not only what a model outputs, but how it arrives at those outputs, is increasingly central to deployment readiness and long-term safety.”

Innodata offers red teaming, adversarial evaluations, and cybersecurity services for LLMs and AI agents.

“AI safety has expanded to include misuse, exploitability, and unintended system behaviors - particularly as models are connected to retrieval systems, code execution environments, autonomous agents, and enterprise tools. Innodata conducts structured red teaming and adversarial evaluations to surface failure modes that are not observable through standard benchmarks. These efforts include probing prompt-injection and jailbreak vulnerabilities, testing misuse scenarios involving retrieval-augmented generation, agent workflows, and tool use, identifying degradation under distribution shift, and supporting mitigation through targeted post-training datasets. In parallel, we have expanded our cybersecurity capabilities as applied to LLMs and AI agents, including threat modeling for agent-based systems, assessment of data exfiltration and privilege-escalation risks, evaluation of secure tool invocation and sandboxing controls, and testing of monitoring and guardrail mechanisms designed to reduce exposure to adversarial attacks and enterprise security breaches.”

Innodata launched Evaluation and Observability Platform as control plane for agentic systems.

““We are also continuing to innovate at a rapid pace. This quarter we launched our Evaluation and Observability Platform in beta - a control plane for agentic systems - and shortly after launch closed our first platform engagement, valued at $1 million in revenue, with a hyperscaler customer. Fifteen additional companies are actively evaluating the platform, and we are in discussions with two leading hyperscalers about potential channel partnerships that we believe could meaningfully expand the platform’s reach. The strength of our research bench was further validated when one of our researchers had two papers accepted at the 2026 International Conference on Machine Learning (ICML), one of which received the prestigious ‘Spotlight’ designation - placing it among approximately 2% of the nearly 24,000 papers submitted to ICML this year.””

Innodata shifted to single-segment reporting focused on agentic AI technologies.

“Effective the first quarter of 2026, the Company is reporting its financial results as a single operating segment. The Company previously reported three operating segments: DDS, Agility, and Synodex. The shift to single-segment reporting reflects the transformation of the Company’s business strategy and operating model, driven by its focus on agentic AI technologies and the increasingly integrated way in which it manages and delivers its services.”

Innodata provides data development, evaluation, and enablement services for agentic and tool-using AI systems.

“Today, leading AI innovation labs and Big Tech companies (including five of the so-called “Magnificent Seven”) building frontier generative AI models and leading enterprises engage us to provide (i) training and post-training data development; (ii) alignment and preference optimization; (iii) capabilities, alignment, and safety evaluation; and (iv) AI enablement and operationalization, including support for agentic and tool-using systems.”

Innodata describes need for behavioral and agentic evaluations of frontier AI systems.

“Early AI evaluation focused primarily on output correctness. In contrast, today’s frontier systems - particularly agentic and tool-using systems - require behavioral and agentic evals that assess how models plan, reason, and act over time. These evals may examine reasoning coherence, tool selection and invocation, multi-step task execution, adherence to system instructions, and robustness under adversarial or ambiguous inputs.”

Innodata states agentic evaluation increases importance of human judgment for deployment readiness.

“This shift toward agentic evaluation materially increases the importance of structured human judgment, domain expertise, and scalable evaluation operations. We believe that the ability to measure not only what a model outputs, but how it arrives at those outputs, is increasingly central to deployment readiness and long-term safety.”

Innodata performs red teaming, adversarial evaluations, and cybersecurity assessments for LLMs and AI agents.

“AI safety has expanded to include misuse, exploitability, and unintended system behaviors - particularly as models are connected to retrieval systems, code execution environments, autonomous agents, and enterprise tools. Innodata conducts structured red teaming and adversarial evaluations to surface failure modes that are not observable through standard benchmarks. These efforts include probing prompt-injection and jailbreak vulnerabilities, testing misuse scenarios involving retrieval-augmented generation, agent workflows, and tool use, identifying degradation under distribution shift, and supporting mitigation through targeted post-training datasets. In parallel, we have expanded our cybersecurity capabilities as applied to LLMs and AI agents, including threat modeling for agent-based systems, assessment of data exfiltration and privilege-escalation risks, evaluation of secure tool invocation and sandboxing controls, and testing of monitoring and guardrail mechanisms designed to reduce exposure to adversarial attacks and enterprise security breaches.”

Innodata announces innovations in dataset creation for LLMs, AI agents, and robotics.

“Announces Innovations in Dataset Creation and Evaluations for LLMs, AI Agents, and Physical AI (Robotics)”

Innodata sees accelerating demand for frontier model training, agentic AI, and physical AI.

““We believe we are entering 2026 with extraordinary momentum. Demand across frontier model training, agentic AI, and physical AI is accelerating, and we believe our role is evolving from data supplier to strategic lifecycle partner for some of the world’s most advanced AI initiatives.”

Innodata reports gains in AI agent constraint satisfaction under real-world complexity.

“Up to 25-point gains in AI agent constraint satisfaction under real-world complexity;”

Innodata supports AI enablement including for agentic and tool-using systems.

“Today, leading AI innovation labs and Big Tech companies (including five of the so-called “Magnificent Seven”) building frontier generative AI models and leading enterprises engage us to provide (i) training and post-training data development; (ii) alignment and preference optimization; (iii) capabilities, alignment, and safety evaluation; and (iv) AI enablement and operationalization, including support for agentic and tool-using systems.”

Agentic systems require behavioral and agentic evaluations of planning, reasoning and acting.

“Early AI evaluation focused primarily on output correctness. In contrast, today’s frontier systems - particularly agentic and tool-using systems - require behavioral and agentic evals that assess how models plan, reason, and act over time. These evals may examine reasoning coherence, tool selection and invocation, multi-step task execution, adherence to system instructions, and robustness under adversarial or ambiguous inputs.”

Shift to agentic evaluation increases need for structured human judgment and scalable operations.

“This shift toward agentic evaluation materially increases the importance of structured human judgment, domain expertise, and scalable evaluation operations. We believe that the ability to measure not only what a model outputs, but how it arrives at those outputs, is increasingly central to deployment readiness and long-term safety.”

Innodata performs red teaming, adversarial evaluations, and cybersecurity testing for LLMs and AI agents.

“AI safety has expanded to include misuse, exploitability, and unintended system behaviors - particularly as models are connected to retrieval systems, code execution environments, autonomous agents, and enterprise tools. Innodata conducts structured red teaming and adversarial evaluations to surface failure modes that are not observable through standard benchmarks. These efforts include probing prompt-injection and jailbreak vulnerabilities, testing misuse scenarios involving retrieval-augmented generation, agent workflows, and tool use, identifying degradation under distribution shift, and supporting mitigation through targeted post-training datasets. In parallel, we have expanded our cybersecurity capabilities as applied to LLMs and AI agents, including threat modeling for agent-based systems, assessment of data exfiltration and privilege-escalation risks, evaluation of secure tool invocation and sandboxing controls, and testing of monitoring and guardrail mechanisms designed to reduce exposure to adversarial attacks and enterprise security breaches.”

Market for data labeling and agentic AI safety services is projected to grow rapidly.

“As AI investment scales, we believe the need for high-quality data and the ability to validate model behavior at scale expands in parallel. By way of illustration, according to Data Collection and Labeling Market (2025–2030) published by Grand View Research, Inc. in November 2024, the global data collection and labeling market was valued at approximately $3.8 billion in 2024 and is projected to reach about $17.1 billion by 2030, representing a compound annual growth rate of approximately 28.4% from 2025 through 2030. We believe these estimates understate the longer-term opportunity associated with next-generation evaluation, safety, and assurance services required for agentic systems and regulated AI deployments, which are increasingly viewed as necessary components of AI commercialization.”

Innodata is investing in evaluation frameworks for agentic trust, safety, and workflows.

“We intend to continue developing new capabilities in response to evolving customer needs and advances in AI technologies. As AI systems become more autonomous and agentic, customers increasingly require solutions that support trust, safety, and control across development and deployment. We are investing in capabilities related to agentic trust and safety, including evaluation frameworks designed to assess agent behavior, tool use, and adherence to constraints across multi-step workflows.”

Innodata is working with companies to develop and benchmark autonomous agents.

“Agentic AI : We believe Agentic AI will unlock the next wave of enterprise productivity. We’re working with leading companies to develop and benchmark autonomous agents across real-world use cases - measuring consistency, reliability, and efficiency.”

Innodata expands capabilities around agentic AI, generative AI training data and enterprise AI.

“Strategic Alignment. We believe we are aligned strategically to support our customers on their generative AI journeys. We continue to innovate and expand our capabilities around collecting and creating generative AI training data; agentic AI; enterprise AI; and large language model trust & safety.”