The AI Agent Index

Documenting the technical and safety features of deployed agentic AI systems

Astra


Basic information

Website: https://web.archive.org/web/20241219095324/https://deepmind.google/technologies/project-astra/

Short description: Phone-based multimodal assistant with tool use.

Intended uses: What does the developer say it’s for? Prototype of a universal AI assistant [source]

Date(s) deployed: Not deployed, “trusted tested waitlist” debut Dec 11 [source]


Developer

Website: https://web.archive.org/web/20241231224948/https://deepmind.google/

Legal name: Google LLC [source]

Entity type: LLC

Country (location of developer or first author’s first affiliation): Incorporation: Delaware, USA (GOOGLE LLC 3582691) [source]

Safety policies: What safety and/or responsibility policies are in place? Several relevant commitments [source] [source] [source]


System components

Backend model: What model(s) are used to power the system? Gemini [source]

Publicly available model specification: Is there formal documentation on the system’s intended uses and how it is designed to behave in them? None

Reasoning, planning, and memory implementation: How does the system ‘think’? 10-minute memory, remembers past interactions [source]

Observation space: What is the system able to observe while ‘thinking’? Live speech, images, video, text [source]

Action space/tools: What direct actions can the system take? Uses Google Search, Maps and Lens for conducting research [source]

User interface: How do users interact with the system? Live video feed with voice and chat interface in demo [source]

Development cost and compute: What is known about the development costs? Unknown


Guardrails and oversight

Accessibility of components:

  • Weights: Are model parameters available? Closed source
  • Data: Is data available? Closed source
  • Code: Is code available? Closed source
  • Scaffolding: Is system scaffolding available? Closed source
  • Documentation: Is documentation available? Unavailable

Controls and guardrails: What notable methods are used to protect against harmful actions? Unknown

Customer and usage restrictions: Are there know-your-customer measures or other restrictions on customers? Currently behind a waitlist

Monitoring and shutdown procedures: Are there any notable methods or protocols that allow for the system to be shut down if it is observed to behave harmfully? Unknown


Evaluation

Notable benchmark evaluations: None

Bespoke testing: Demos [source]

Safety: Have safety evaluations been conducted by the developers? What were the results? None

Publicly reported external red-teaming or comparable auditing:

  • Personnel: Who were the red-teamers/auditors? None
  • Scope, scale, access, and methods: What access did red-teamers/auditors have and what actions did they take? None
  • Findings: What did the red-teamers/auditors conclude? None

Ecosystem information

Interoperability with other systems: What tools or integrations are available? Uses Google Search, Maps and Lens for conducting research [source]

Usage statistics and patterns: Are there any notable observations about usage? None


Additional notes

None