UiPath TypeSafe connector
Preview UiPath Integration Service connector for Jev Choice, Score, Noul, and Evaluate with your own TypeSafe API key.
Overview
UiPath announced a preview TypeSafe connector on October 20, 2026. It adds Jev activities to UiPath Integration Service so automations can call List Models, Choice, Score, Noul (yes/no), and Evaluate (System One) on text or JSON state. Search Integration Service for Jev, TypeSafe, or System One to install it on Maestro Automate, Maestro Orchestrate, and other UiPath surfaces that support connectors.
Problem: RPA and enterprise automation teams need typed Jev gates inside Maestro and Orchestrator workflows without routing every decision through a generative LLM judge.
Built for: UiPath developers building Maestro Automate or Maestro Orchestrate flows who want List Models, Choice, Score, Noul, and Evaluate activities against their own TypeSafe API key.
First indexed on Jev Directory: 2026-10-20
Creator and team
- Name
- Sandeep Panda
- Organization
- UiPath
How Jev is used
- Role in the product flow
- Typed decision activities inside UiPath workflows: classification, scoring, extraction gates, and multi-question Evaluate on one state blob
- Primitives
- ChoiceScoreNoul
- State in
- State as text or JSON plus Instructions, Options or Levels, and an optional Model field (defaults to jev-latest; output shows the versioned id such as jev-1.13.0). State plus the longest question can be up to 32k tokens per UiPath docs.
- Decision out
- Choice labels, Score level with probability-weighted score and per-level probabilities plus confidence, Noul yes/no with confidence, or Evaluate answers for any mix of questions about one state in a single request.
- Bring your own TypeSafe API key (bearer token, including from an Orchestrator credential asset)
- Configure State, Instructions, and primitive-specific options in a TypeSafe activity
- Read confidence on outputs and branch to human review or a reasoning agent when below threshold
- Optional two-step extraction: one Jev call locates document spans per field, a second Choice picks exact values
This is the first UiPath connector on Jev Directory: Jev runs as Integration Service activities instead of a community n8n node or a one-off script. UiPath's October 20, 2026 product blog (https://www.uipath.com/blog/product-and-updates/typesafe-connector-bring-jev-models-to-uipath?ref=jev.aitools.fyi) includes remarks from Erik Gafni, TypeSafe CTO. Requests use your TypeSafe agreement directly; that post states they are not routed through the UiPath AI Trust Layer. That matches box-jev-incident-triage and workday-sana-jev-generative-ui, where governed enterprise systems call Jev for typed gates, but UiPath targets general automation builders on Maestro and Orchestrator. Score accepts 2 to 10 ordered levels and returns probabilities; output tokens are not billed per the UiPath Score activity documentation (https://docs.uipath.com/activities/other/latest/integration-service/uipath-typesafe-typesafe-score?ref=jev.aitools.fyi). UiPath highlights grading agent output, document field extraction, classification, and re-ranking search results, with a confidence threshold pattern before human review or a reasoning agent (UiPath calls this a two-brain-system). The connector is preview only: UiPath's speed and cost figures are early internal experiments, not independent benchmarks, and there is no named customer deployment in public materials yet. dommonte-n8n-nodes-typesafe-ai covers community workflow nodes; openlayer-ai-jevals fits teams comparing Jev gates to LLM judges. Cite that blog and https://docs.uipath.com/integration-service/automation-cloud/latest/user-guide/uipath-typesafe-typesafe?ref=jev.aitools.fyi.
Sourced performance claims
- UiPath early internal experiments on eval datasets reported 94% agreement with an LLM-as-a-judge system.Source: UiPath blog, Oct 20 2026 (UiPath-measured, not independent)
- Those same UiPath internal experiments reported about 7.7x lower cost and 4x to 20x faster than the equivalent LLM judge.Source: UiPath blog, Oct 20 2026 (UiPath-measured, not independent)
Features and stack
Features
- Preview TypeSafe connector in UiPath Integration Service
- Activities: List Models, Choice, Score, Noul, Evaluate (System One)
- Bring-your-own TypeSafe API key with Orchestrator credential support
- Windows and Cross-platform UiPath projects per activities docs
- Confidence thresholds for human review or downstream reasoning agents
Stack
- UiPath Integration Service
- UiPath Maestro Automate and Orchestrate
- TypeSafe Jev (jev-latest default)
Links
FAQ
- Is the UiPath TypeSafe connector generally available?
- UiPath documents the connector as preview in Integration Service as of October 20, 2026. Confirm status and entitlements in your UiPath tenant before production use.
- Does UiPath route Jev calls through the AI Trust Layer?
- UiPath's blog states requests fall under your own agreement with TypeSafe and are not routed through the UiPath AI Trust Layer when you bring your own TypeSafe API key.
- How do UiPath's benchmark numbers compare to independent tests?
- UiPath's blog reports early internal experiments (94% agreement with an LLM judge, lower cost, faster latency). Those are UiPath-measured on their eval datasets, not independent verification, and there is no named customer deployment cited yet.
- How is this different from the n8n TypeSafe node?
- dommonte-n8n-nodes-typesafe-ai is a community n8n integration for a different hosting model. This listing covers UiPath's first-party Integration Service connector for Maestro and Orchestrator automations.
Related learn guides
Original Jev guidance that pairs with this product pattern.
- Jev use cases
The patterns builders actually search for: moderation, routing, triage, RAG verify, and agent gates.
- Jev vs LLM classification
When to gate with System One probabilities instead of asking a chat model to label things.
Related products
Hand-picked neighbors with rich profiles or overlapping tags.