Silmaril GTM brain

Silmaril provides an adaptive runtime security firewall for organizations deploying AI agents and coding assistants.

category

AI agent security

market profile

  • text: Tool-using agents create a security boundary around external content, model decisions and actions in connected systems.
  • text: Prompt injection defense is a developing market with runtime enforcement, assessment and agent governance products.

personas

  • title: Application security director
    pains:
    • Untrusted content may cause agents to misuse tools or expose data
    • Security teams need evidence and context around blocked actions
  • title: AI platform engineering lead
    pains:
    • Guardrails must integrate into existing agent frameworks without excessive latency
    • Agents need safe access to external tools and data
  • title: Engineering leader rolling out coding agents
    pains:
    • Repository content may contain hidden instructions
    • Agents have access to sensitive source code and credentials

icp

  • text: Software and enterprise teams deploying production AI agents that read outside content and invoke tools; no public size threshold identified.
  • text: Organizations adopting coding assistants or connected agents with sensitive systems and security leadership involved in deployment.

message pillars

  • text: Update defenses as the threat environment changes.
  • text: Intercept dangerous agent outcomes at runtime.
  • text: Treat external information as an agent attack surface.

objections

  • objection: Will this slow agents or block legitimate work?
    response: Benchmark latency and false positives on the buyer's real workflows before broad rollout.
  • objection: Does it fit our frameworks and deployment requirements?
    response: Test the integration and hosting options with the actual models and tool calls in scope.
  • objection: Can security teams investigate what was blocked?
    response: Review the evidence and trace detail available during a pilot.

competitors

  • name: Lakera
    why: Offers AI security and prompt injection defenses.
  • name: Prompt Security
    why: AI security vendor listed in the agent security competitive landscape.
  • name: Zenity
    why: Agent security platform addressing runtime behavior and governance.
  • name: PromptArmor
    why: Provider focused on AI security and prompt injection risks.

influencers

  • name: Eduardo Velasco
    url: https://www.linkedin.com/in/ejvelasco
    why: Co-founder and CTO discussing the company's AI security approach.

keywords

  • AI agent security
  • prompt injection defense
  • agent firewall
  • AI coding agent security
  • LLM tool call security
  • AI agent red teaming

voice

  • Threat-led and technical
  • Direct
  • Confident

signals

  • text: An organization is moving tool-using agents into production with access to sensitive data.
  • text: An engineering team is rolling out coding agents across repositories and reviewing the associated trust boundaries.

stack

  • LangGraph
  • Codex
  • Claude Code
  • GitHub Copilot
  • Cursor

pricing

model: unknown

motion

sales_led

channels

  • website
  • LinkedIn
  • security community

confidence

medium

Founders

  • name: Aum Upadhyay
    title: Founder/CEO
    role: CEO
    bio: Co-Founder & CEO at Silmaril. I built the security and privacy framework at AWS that prevented over $1.8B in damages. Now I bring that same defensive mindset to Silmaril, working with customers to ship prompt injection defense.
  • name: Eduardo Velasco
    title: Founder/CTO
    role: CTO
    bio: Co-Founder & CTO at Silmaril. I prove the risks are real by finding exploits myself. I chained a prompt injection into root access inside ChatGPT. Ex-Amazon tech lead specializing in low latency ML models that generated $400M in annual revenue.

Back to YC companies