Alithea Biotechnology logo
HomePlatformScienceAbout UsContact

Alithea Biotechnology GmbH

AI-native life science solutions with transparent scientific execution and publication-linked evidence.

HomePlatformScienceAbout UsContact

Addresses

Headquarters / Invoice Address

Liebühl 2 A

79104 Freiburg, Germany

 

Laboratory / Delivery Address

Robert-Rössle-Straße 10, Gebäude 95

13125 Berlin, Germany

Legal

Managing Director: Fanny Alicia Giannou

Commercial Register: Amtsgericht Freiburg

Registration Number: HRB 719811

VAT ID: DE325036990

Imprint & Privacy Policy

Contact

info@alithea-bio.com

Email

Social

LinkedInYouTubeInstagramFacebook

© 2026 Alithea Biotechnology GmbH. All rights reserved.

    Platform

    The governed operating system for immunopeptidomics.

    Four layers. Proprietary data, modular compute, a tool marketplace, and an AI-native interface — secured with row-level isolation, audit trails, and credit-based governance.

    Try now

    The platform

    Four layers. Each one a product. Together, a system no one else has.

    Database layerCompute layerTool marketplaceSelf-service UI

    Click a layer to explore

    Select a layer to see what's inside

    Each slab is one product surface in the stack. Start here for the quick system view, then click a layer for the detailed breakdown.

    Database layer

    Proprietary HLA ligandome, third-party data, and client datasets — unified.

    Compute layer

    Secure, modular, auto-scaling infrastructure for ML/AI workloads.

    Tool marketplace

    Containerized analytics tools with MCP descriptors for AI-driven execution.

    Self-service UI

    Web UI, AI copilot, and SDK — every user gets the right interface.

    Three ways in

    Every user gets the interface that fits their workflow.

    Web UI

    Direct tool usage, data exploration, and result review

    AI copilot

    Conversational automation — the copilot runs the workflow for you

    SDK + API

    Build, test, and publish tools. Integrate into existing pipelines

    Everything is LLM-readable

    Every tool publishes a machine-readable descriptor. Any LLM agent can discover, configure, and execute it — no custom adapters, no glue code.

    Auto-generated tool descriptors

    Each module publishes its input/output schema via the Model Context Protocol (MCP). LLM agents read it natively.

    Agent-safe execution

    Read data, run modules, poll results. Destructive operations are blocked. Rate limits and credit caps prevent runaway execution.

    Multi-model, bring your own key

    Claude, GPT-4, Gemini, Amazon Nova — connect your own API keys or use platform-managed models.

    Governance built in, not bolted on

    Every action is isolated, audited, and budget-controlled.

    Row-level security

    Database-enforced tenant isolation. Cross-org data access returns zero rows — not an error, not a flag, zero rows.

    Immutable audit trail

    Login, publish, execute, adjust — every action logged with actor, timestamp, org context, and result.

    Credit ledger

    Pre-run estimation, reservation, settlement. Immutable transaction log. No surprise bills.

    Supply chain attestation

    Module signing, container image verification, vulnerability scanning, and SBOM generation on every module.