Aporia has been nominated for the 2026 World AI Awards in the Agent Governance & Policy Control category, recognising its work on technology designed to monitor AI applications and enforce rules around how those systems interact with users.
It is a problem that has become harder as generative AI has moved into customer service, internal business tools and autonomous agents. An AI application can produce a useful answer one moment and disclose sensitive information, follow a malicious instruction or wander outside an organisation’s approved subject matter the next.
Aporia built its AI Guardrails around that problem: instead of relying entirely on the underlying model to behave as intended, another control layer checks interactions and can intervene when predefined policies are violated.
The World AI Awards recognises organisations, individuals, products and technologies contributing to the development and real-world application of artificial intelligence across industries.
Putting policy between the AI and the user
Aporia’s approach is fairly direct. Its Guardrails sit between an AI application and its users, inspecting messages and model responses against policies configured by the organisation.
Those controls can address issues including hallucinations, prompt injection, personally identifiable information, data leakage, profanity and conversations that move into restricted topics. Aporia also developed policy controls for company-specific requirements rather than limiting the platform to generic safety filters.
When the system detects a violation, developers can configure different responses. An interaction can be logged or flagged, while higher-risk content can be blocked, overwritten or rephrased before reaching the user.
That makes the technology particularly relevant to Agent Governance & Policy Control. The issue is not simply watching what an AI system has already done. It is enforcing operational boundaries while the interaction is happening.
Small models take on specific control jobs
Aporia took an unusual architectural approach to the problem.
Instead of using one large language model as a universal judge of another AI system, the company developed what it calls a multiSLM Detection Engine. Multiple specialised small language models handle individual policies, with different models focused on areas such as hallucination detection or prompt-injection attacks.
Aporia reported an average latency of 0.34 seconds and a 90th-percentile latency of 0.43 seconds in its 2024 benchmark tests with five policies enabled. Those are company-reported benchmark results rather than independent measurements, but they illustrate the engineering challenge behind real-time governance: a safety layer is considerably less useful if every check makes an AI application noticeably slower.
Aporia also expanded Guardrails beyond text. In 2024, it introduced controls designed for text, audio and visual AI applications, reflecting the shift toward multimodal models that can process several kinds of information within the same interaction.
Governance becomes an operational problem
There is a practical distinction between writing an AI policy and enforcing one.
An organisation might decide that its customer-facing agent should never expose personal information, discuss certain subjects or accept instructions intended to bypass its safeguards. The difficult part is translating those requirements into controls that work across thousands or millions of interactions.
Aporia’s restricted-topic Guardrail, for example, allows organisations to specify subjects an AI application should not discuss without rewriting the underlying system prompt. Other policies can inspect both user inputs and model outputs, giving developers control on either side of the interaction.
The technology does not eliminate the risks associated with generative AI. Guardrails themselves rely on detection systems that can make mistakes. What they provide is an additional enforcement layer that organisations can configure independently from the model generating the answer.
Aporia becomes part of Coralogix
Aporia’s story also changed significantly before its 2026 World AI Awards nomination.
Observability company Coralogix acquired Aporia in December 2024. Aporia said its guardrails and observability technology would be integrated with Coralogix, while its founders and team would help establish Coralogix AI, a dedicated research and development centre focused on AI monitoring, security, governance and control.
TechCrunch reported that Aporia had raised approximately $30 million before the acquisition. The purchase price was not officially disclosed by the companies.
The technology has continued to play a role in Coralogix’s AI strategy. In 2025, Coralogix described its offering as providing AI companies with observability and guardrails covering model performance as well as the quality, security and governance of AI responses, capabilities it gained through the Aporia acquisition.
That evolution gives Aporia’s work a broader context. AI governance is moving away from policy documents alone and increasingly becoming an engineering problem: rules have to be translated into controls that can operate while AI systems are actually being used.
Graham Cooke, President of the World AI Awards, said:
“Aporia’s nomination highlights an increasingly important question for organisations deploying AI agents: how do you turn governance policies into controls that actually operate in real time?
“Its work on AI Guardrails and configurable policy enforcement demonstrates the opportunity to give businesses greater oversight of how AI applications interact with users, information and organisational rules. We congratulate Aporia on its 2026 World AI Awards nomination in Agent Governance & Policy Control and look forward to following how this area develops.”
Aporia joins organisations, researchers, entrepreneurs and technology developers being recognised through the 2026 World AI Awards.
The programme recognises organisations, individuals and technologies contributing to the development and application of artificial intelligence across industries. The Agent Governance & Policy Control category addresses a particularly immediate part of that landscape as businesses move from experimenting with generative AI to putting agents and AI applications into everyday operations.
That shift changes the governance problem. Policies written on paper still matter, but organisations increasingly need technical mechanisms capable of monitoring behaviour, applying restrictions and responding when an AI system crosses an established boundary.
Aporia’s work sits squarely in that emerging layer between AI models and the people who use them.
Learn more about Aporia and its AI Guardrails technology at https://www.aporia.com/.
Discover the World AI Awards 2026, explore the nominees and learn more about the awards at https://www.worldawards.ai/.
