Innovation Challenges

;
Solicitation for Cyber Resilience Solutions
Challenge Owner(s) Cap Vista
Organiser(s) Cap Vista
Industry Type(s) Digital/ICT
Opportunities and Support Opportunities to validate solutions with challenge statement owners through project trials.
Application Start Date 29 May 2026
Application End Date 30 June 2026
Website Click here to learn more

About Challenge

We publish challenge statements and invite global companies and startups to submit proposals which could enhance Singapore's national security.

After evaluating the submitted proposals, we will invite shortlisted participants to pitch their proposals and to address any queries from our evaluation panel which will comprise all relevant parties.

Each successful participant will be awarded a fully-funded trial that covers the costs of validating their solution for our use case(s). The trial may be in the form of proofs-of-concept, experiments, demonstrations, or any other reasonable ideas as proposed in the submitted application

We aim to award suitable innovators within 60 days after the submission deadline. Awarded providers will gain the opportunity to work directly with our end users to showcase and validate your solutions for national security applications. If your solution demonstrates success and good potential, we aim to scale it quickly.

Challenge Owner(s)Cap Vista
Context

Modern Security Operations Centers (SOCs) face escalating complexity across both defense and commercial cyber domains. The sheer velocity and volume of alerts, combined with the sophistication of threat actors operating in gray zones, have outpaced traditional analytic and response frameworks.

What We Are Looking For
  • Autonomous Agentic AI capable of alert correlation with proactive reasoning, task execution and cross domain collaboration.
  • Explainability of decision and outcome of action undertaken with human oversight. Clear event timestamp / audit trail in the actions taken by the AI agent.
  • Solutions that have the ability to be deployed SaaS (Software as a Service) and On-Prem.
Key Metrics
  • Provide detection coverage against APTs (Advanced Persistent Threat).
  • False positive rates comparison between AI generated rule vs OOTB (Out Of The Box) rules.
  • Blind spot identification rate.
Your Technology Solution Should Demonstrate
  • Measurable productivity gains in SOC Operations by having agentic AI systems autonomously triage & prioritize alerts from integration of cyber sensors & multiple intelligence feeds, and
  • Surface weak signals from cyber sensors indicative of emergent or coordinated threat campaigns before they escalate.
  • High technical readiness and deployment maturity with commercial or defence customers.
What A Trial Looks Like
  • Phase 1 (Internal evaluation at no cost to us) – We seek a free trial for 2 months in our test environment (Cloud) to showcase your solution’s capability only with shortlisted applicants.
  • Phase 2 (Paid) – If successful in Phase 1, we seek to subscribe to your solution for 12 months to be implemented in our chosen environment.
What We Are Not Looking For
  • Blackbox reasoning - general purpose AI with cybersecurity prompts is not the same as an agent with native threat reasoning e.g. probe whether the model actually understands APT behaviours.
  • A platform without out-of-the-box agents to improve Cyber SOC operations.
  • Proprietary rule formats.
  • Fully autonomous and no feedback loop provided for analysts.
  • Point solutions with no integration with sensors.
Follow-on Implementation Cost
  • Long term pricing model for Saas or On-prem solution.
  • Manpower and technical support (day rates).
Challenge Owner(s)Cap Vista
Context

Current cyber defenses are largely rule based and are playing catchup to novel adversarial AI threats (e.g. novel jailbreaking prompts, data poisoning and Agentic AI threats). AI security testing can be resource intensive (e.g. throwing a consolidated list of jailbreaking prompts to test robustness of LLM - Large Language Model).

What We Are Looking For
  • Novel methodology, approach or even product / toolkit to defend threats during AI development and production, as well as to perform security testing on AI models / applications and libraries.
  • Proposal that shows technical depth in addressing stealthy adversarial AI threats.
Key Metrics
  • Accuracy of the tool (e.g. false positives, true positives) and amount of compute resources required.
  • Coverage of attack scenarios on LLM and agents (e.g. OWASP Top 10 for LLM and Agentic AI Applications Coverage).
  • Solution latency (e.g. Number of test cases & time needed for complete coverage).
  • Ease of setup & level of human involvement required.
Your Technology Solution Should Demonstrate
  • How the solution tests, and addresses practical threat scenarios on an enterprise Agentic AI System (e.g. AI coding assistant that interacts with a CI/CD pipeline).
  • Tool Customizability (e.g. hosting custom prompts and algorithms).
  • Support for multiple AI API (Application Programming Interface) format (e.g. OpenAI API Format).
  • Integrability (e.g. data sharing, alert forwarding) with commercial cyber solutions, e.g., SIEM (Security Information and Event Management).
  • Supportability for various deployment modes, (e.g. SaaS, self-hosted, on-premise, air-gapped).
What A Trial Looks Like
  • Phase 1 (Internal evaluation at no cost to us) – We seek a free trial for 1 month via SaaS to showcase your solution’s capability only with shortlisted applicants.
  • Phase 2 (Paid) – If successful in Phase 1, we seek to subscribe to your solution for 12 months to be implemented in our chosen environment.
What We Are Not Looking For
  • Generic or non-threat driven defence approaches.
  • Indifferentiable value from open source offerings (e.g. only static library of prompts for testing).
Follow-on Implementation Cost
  • Long term pricing model for Saas (Software as a Service) or On-prem solution.
  • Manpower and technical support (day rates).