Overview
Duet is a toolkit that provides near-frontier AI for applications at a flat $20/month. It is described as a small (~200-line, stdlib-only) toolkit that routes requests between local and cloud models. The system uses three tiers: Tier 1 is local and free, using a small open-weight model on the user's own Ollama. Tier 2 is cloud and flat-fee, where the same call retargets to a huge open-weight model on Ollama's GPUs through a signed-in local daemon, with no API key or code changes. Tier 3 is direct for total outages, going straight to https://ollama.com/api/chat with an API key. All tiers speak Ollama's native /api/chat, so failover is a retarget, not a rewrite. Privacy is enforced twice: tools marked cloud_safe=False are never seen by the cloud brain and are refused by dispatch even if hallucinated. The pattern is app-agnostic, suitable for any person or pipeline asking questions in plain language against live state.
Pricing
Free
Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.
Visit WebsitePrices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.
Use Cases
- Teams use Duet to power automation brains for workflow engines, handling classification, drafting, and triage steps.
- Teams use Duet to build application-specific assistants that are fed docs and private source data and embedded in the app's web window.
- Teams use Duet as an internal ops/admin copilot by wrapping dashboards' data in tools.
- Teams use Duet to create a SaaS product assistant that provides grounded answers about the customer's own account with PII fenced local.
- Teams use Duet as a developer assistant over their own tooling, with build status, deploy state, and log summaries as tools.
Features
AI assistant
Natural language interface
Workflow automation
Reviews
No reviews yet. Be the first to share your experience!