FILTARReal-time AI guardrails
Inspect AI interactions for unsafe, malicious or non-compliant content and return a clear decision.
ALLOW · BLOCK · REDACT · WARN · REVIEWFILTAR COREONE API
Protect your
AI interactions.
FILTAR Core adds real-time guardrails and sensitive-data protection to applications, assistants, chatbots, agents and inference endpoints.
Developer-firstReal-timeSaaS or your environment
01 interaction.received02 inspect: complete03 context: financial_services04 policy: applied05 decision: BLOCK + REDACT
EXPLAINABLE DECISIONPROTECTION APPLIED ✓
CORE CAPABILITIES
Start with the capabilities that matter most.
Customers activate the FILTAR Core modules that match their stack and risk profile.
Sensitive-data protection
Detect and prevent sensitive personal or business information from reaching models or users.
INSPECT · UNDERSTAND · ENFORCE · PROVEPolicy enforcement
Apply organisation-specific rules, thresholds and actions consistently across AI workloads.
POLICY · THRESHOLD · ACTIONWHY FILTAR CORE
Purpose-built protection for fast, focused security decisions.
Specialised Small Language Models are designed for focused security decisions. Every interaction is protected without relying on a general-purpose LLM.
Specific security and semantic tasks
Models designed to interpret the signals and business context that matter to each decision.
CONTEXT OVER GENERIC KEYWORDSLess unnecessary overhead
Reduce latency, compute and unnecessary infrastructure overhead through a focused-model approach.
PURPOSE-BUILT COMPUTEOne engine across AI environments
Use FILTAR Core across applications, assistants, agents and inference environments.
SAAS · CUSTOMER ENVIRONMENTEvidence for every decision
Return actions, scores and evidence for security, risk and compliance teams.
DECISION + TRACEPRODUCT ROADMAP
Extend FILTAR Core across the AI lifecycle.
Upcoming components connect to the same FILTAR Core security engine.
FILTAR SDK
Developer tooling for embedding FILTAR into build and test workflows.
BUILD · TEST · PROTECTFILTAR Agent Gateway
Pre- and post-inference security for agents, models, cloud and neocloud endpoints.
PRE + POST INFERENCEEnterprise browser extension
Extend FILTAR protection to workforce AI interactions through the browser. Currently used for customer POCs; broader distribution remains under review.
WORKFORCE AI
MARKETPLACEPROCURE THROUGH AWS
Plan your AWS Marketplace procurement.
Contact the team to discuss AWS Marketplace availability, deployment and purchasing requirements.