AI tool
Sage2 by Levanto Labs
Sage2 by Levanto Labs provides APIs for developers to extract structured data like dates, amounts, and durations from natural language. It also tracks changing user intent during conversations to help applications decide what to do next.
About Sage2 by Levanto Labs
Sage2 by Levanto Labs is an API toolkit that converts natural language into structured data and tracks streaming decisions as conversations unfold. Developers can extract specific values like dates, amounts, and durations, while monitoring how user intent shifts over the course of an interaction. The system is built for applications that need to act on conversational input without resending the full dialogue for each update.
Review
Sage2 entered public availability this week, targeting developers who build AI agents, sales automation, and document-processing systems. Its core pitch centers on extracting structured data from chat and voice interactions while maintaining source links back to the original text. The API handles up to 32 concurrent questions in a single streaming session, which sets a concrete ceiling for multi-threaded conversation tracking.
Key Features
- Structured data extraction - Pulls dates, amounts, times, and durations from natural language input and returns them as typed fields.
- Source-linked results - Each extracted value maps back to its exact position in the original text, so developers can verify where a figure or date originated.
- Streaming decisions - Updates decisions incrementally as a conversation progresses, without requiring the entire transcript to be reprocessed.
- Multi-question sessions - Tracks up to 32 separate questions simultaneously within one streaming connection.
- Beta capabilities - Reasoning, web grounding, and vision features are available in beta, though their stability and latency characteristics aren't yet documented.
Pricing and Value
Pricing details for Sage2 are not yet defined in the current launch materials. Access requires a Sage API key, and the product page doesn't list usage tiers, rate limits, or per-call costs. Developers evaluating the API should request pricing directly from Levanto Labs before building production workflows around it.
Pros
- Source-linked extraction lets you trace every structured value back to the raw text, which simplifies debugging when the API misinterprets something.
- Streaming decisions update without resending full conversation history, cutting down on token usage for long-running interactions.
- 32 simultaneous questions per session give room for complex, multi-faceted intent tracking in a single connection.
- The beta reasoning and web grounding modules hint at capabilities beyond simple extraction, though they're not production-grade yet.
Cons
- Pricing is unpublished. Any team budgeting for API costs will need to contact Levanto Labs directly before committing.
- Correction handling mid-conversation isn't documented - if a user changes their order twice, it's unclear whether the API revises the original extraction or appends a new one.
- Not well suited for teams that need a fully managed, no-code conversational intelligence layer. Sage2 is an API for developers who want to wire extraction logic into their own application code.
Teams building custom sales agents or support bots that need to pull dates, amounts, and intent signals from chat logs will find the source-linking and streaming design practical. The multi-question tracking also fits document-processing pipelines where several fields need simultaneous extraction. Developers who prefer a turnkey SaaS dashboard over raw API endpoints should look elsewhere - Sage2 expects you to build the application layer yourself.
