llms.txt
# steffi.ai: AI Agents That Handle the Work Your Team Should Not Have to Do
> steffi.ai is a secure autonomous AI agent platform built in Rust. Deploy digital workers to handle everyday operations like CRM updates, inbox sorting, file parsing, web research, and document drafting, and scale up to factory sensor monitoring, Android device control, and multi-agent enterprise pipelines 24/7. Run tasks safely in the background under strict governance. Steffi is currently in private beta: contact us to request access.
> For exhaustive technical architecture, benchmark matrices, and crate specifications, see [llms-full.txt](https://steffi.ai/llms-full.txt).
Steffi is a comprehensive AI agent platform built in Rust. It is designed for teams of any size: from solo founders and small businesses to enterprise operations and factories. Steffi runs autonomous AI agents that operate reliably in the background with full governance, security, and control. By running orchestration logic inside a compiled Rust platform, Steffi achieves 5x token efficiency compared to traditional frameworks and ensures completely predictable behavior.
The core philosophy of Steffi is controlled autonomy: agents execute tasks independently, but the platform enforces strict security boundaries, credential protection, prompt injection scanning, and full audit logs. This robust control layer allows smaller local models (like Qwen or Gemma) to perform complex tasks with high accuracy, since the platform carries the model.
### Key Capabilities and Features
Free Trial, Lite, Pro, and Max plans include a dedicated, isolated cloud server instance that we host and manage for you. The separate Self-Hosting Option lets you deploy Steffi on infrastructure you own or control for HIPAA, financial audit, or industrial compliance. Rent agents starting at $5 per hour on a pay-as-you-go model, or deploy flat-rate monthly seats. Steffi compiles to a single binary for fast, secure, and self-hosted deployment.
Agents run for hours or days in the background to complete multi-step workflows, recovering from failures automatically. Droid Control allows wireless control of physical Android devices without taking screenshots, operating entirely via structured markdown UI maps to make calls, send SMS natively on physical SIM cards, and drive apps.
Built-in security includes 6-layer prompt injection defense, AES-256-GCM credential vaulting, output filters, and static file antivirus scanning. Teams can assign tickets directly to agents on a collaborative board with automated task rollup, and let agents run Python and Node.js code inside secure, isolated Docker environments gated by strict gRPC access policies.
### Core Technology
The SMI (Steffi Machine Interface) proprietary model architecture handles memory compaction and tool selection, choosing a maximum of 5 tools and 1 skill per message turn to lower costs and errors. The native Specter browser engine lets agents navigate single-page applications (SPAs) safely without WebGPU overhead. The supervised DAG scheduler gathers Directed Acyclic Graph orchestration details under Erlang-OTP style supervision trees for continuous stability and timezone-aware cron runs. Persistent 4-tier memory preserves business context across sessions.
### Pricing Plans
Free Trial, Lite, Pro, and Max plans include a dedicated, isolated cloud server instance for each customer's agent. We provision, host, maintain, and monitor the server, so customers have no server maintenance to handle. The Self-Hosting Option is separate and allows customers to deploy Steffi on infrastructure they own or control.
- Lite plan ($5/hour, Hosted and Managed): Includes a dedicated cloud server and a 24B to 40B parameter model.
- Pro plan ($15/hour, Hosted and Managed): Includes a dedicated cloud server and a 120B to 150B parameter model.
- Max plan ($50/hour, Hosted and Managed): Includes a dedicated cloud server and a 1T to 2T parameter model.
- Self-Hosted (Contact for pricing): Custom enterprise deployment on private infrastructure or VPC.
## Documentation
- [Home](https://steffi.ai/): Product overview, key features, and pricing
- [Security & Compliance](https://steffi.ai/security/): Enterprise security architecture, SOC 2 controls, and AES-256 vault
- [Technology & Architecture](https://steffi.ai/technology/): Compiled Rust runtime, SMI tool routing, and OTP scheduler
- [Use Cases](https://steffi.ai/use-cases/): Automation examples across operational workflows and compliance industries
- [Privacy Policy](https://steffi.ai/privacy/): Data protection practices, encryption standards, and GDPR compliance
- [Terms of Service](https://steffi.ai/terms/): Platform guidelines, licensing terms, and liability policies
## External Resources
- [Full LLM Context](https://steffi.ai/llms-full.txt): Comprehensive technical architecture and crate specifications
- [Olib AI Organization](https://github.com/Olib-AI): Open-source repositories and development tools
- [Owl Browser](https://www.owlbrowser.net): Anti-detect browser paired with Steffi for web automation
- [Olib AI](https://olib.ai): Company homepage and access requests
