Dimension scores are derived from public data and fields; weighted into the composite. Reference only.
Tilde.run is positioned as a way for AI agents to run more safely on production data and real code. Its core idea is to turn every agent run into a rollbackable transaction: code in GitHub, data in S3, and documents in Drive are presented as a unified versioned filesystem, while all outbound calls are inspected and logged. This makes it feel more like a safety runtime layer for autonomous code rather than just a conventional sandbox.
Based on the available information, Tilde focuses on isolated sandboxes, a versioned filesystem, network policy, and auditing. The versioned filesystem can preserve history and support rollbacks, which helps reduce destructive risk when agents modify code, process files, or access external data. Outbound call inspection and logging are especially important in production environments, as they can be used to trace how an agent interacts with external networks. Integrations explicitly mentioned include GitHub, S3, and Drive, while the Enterprise plan also includes SAML/OIDC SSO. The site has SDK and Docs entry points, but the main copy does not provide details on supported languages, frameworks, or APIs.
Tilde uses a credits-based pricing model. The free plan includes 1,000 credits per month, requires no credit card, and has a hard cap to prevent unexpected bills. Pro is $49/month and Team is $99/month, adding more credits, seats, concurrent sandboxes, egress, and higher support tiers. The usage catalog is fairly transparent: sandboxes cost 0.25 cr/min, storage is 10 cr/GB-month, object operations are 1 cr/10k ops, and egress is 20 cr/GB. Enterprise is custom-priced and mentions BYOC, multi-tenancy, SSO, and SLA, but it is still marked as coming soon.
The main strength is that Tilde addresses key pain points for running AI agents in production: rollback, auditability, isolation, and network control. The free trial has a low barrier to entry, and the paid tiers do not appear to have obvious feature restrictions. The downsides are that it is still in private preview, enterprise capabilities are not fully available yet, and there is limited information about SDKs, language support, real documentation quality, and operational maturity. It is best suited for small teams or independent developers building production-grade agents that need access to real codebases and object storage.
The main content does not disclose access from mainland China, supported payment methods, or compliance-oriented deployment options, so its China accessibility can only be considered unknown. If access or payment is restricted, alternatives such as Modal, E2B, Daytona, GitHub Codespaces, and Replit may be worth evaluating depending on the use case.
⚠ This review is compiled from public sources and does not constitute a purchase recommendation. Verify all facts on the vendor's official site. Verify on tilde.run official site.
tilde.run is an Unknown AI Apps provider. TG4G tracks its product information, an overall rating of 8.0/10, and a China-accessibility score of Workable. Click "Visit Official Site" to reach tilde.run directly.