Building the security platform for the AI era
From inception to launch and beyond, our roadmap reflects a simple belief: securing AI should be fast, effective, and compatible with how teams already build. Here’s how we got here—and where we’re headed.
Project inception at Visionary Technologies ApS
Bastio AI Security is launched as a research-driven initiative under Visionary Technologies ApS to explore practical defenses for production LLM systems.
Platform foundations & field research
We begin building the Bastio.com security platform. Real‑world testing and customer trials highlight just how easy it is to jailbreak and prompt‑inject LLMs—and how little protection most organizations have in place. Development accelerates around a pragmatic, drop‑in gateway with policy enforcement and detection.
Public launch of Bastio Platform
The Bastio security platform goes live: OpenAI‑compatible proxy, inline policy controls, geolocation rules, lists & regex, analytics, and cost‑saving routing across multiple providers.
Training Bastio Security LLM
We are training a specialized Security LLM focused on detection, triage, and policy recommendations—grounded in real traffic telemetry and red‑team datasets.
What’s next
Model tuned for jailbreak detection, sensitive‑data leakage, and policy recommendations.
First‑class TS SDK with streaming helpers and declarative guard policies.
Attack taxonomy, cohort analysis, and savings breakdowns for finance & security teams.