Skip to main content

Release Notes — v1.2.0

June 2026

New Features​

Skills (formerly Prompt Templates)​

Prompt Templates are now called Skills, and the Prompt Library is now the Skill Library. This is a naming change only — all of your existing templates, inputs, data elements, model settings, and tools carry over automatically, and nothing needs to be migrated.

Alongside the rename, we've clarified the difference between a prompt (the raw text you type into chat in the moment) and a Skill (a saved, reusable, structured version of that instruction).

Learn more: Prompts vs. Skills

Unified Chat Tools Menu​

A single "+" tools menu in chat replaces the previous per-tool toggles. From one popover you can open:

  • A dedicated Integrations modal — a searchable card grid with Personal and Shared tabs for managing your connected services.
  • A Knowledge settings modal — configure Knowledge tools, collection filters, and document-type scoping for the conversation.

This consolidates tool and integration management into one consistent place across chat.

New Integrations​

Three new MCP integrations are available this release:

  • Amplitude — Query product analytics, charts, and event data directly from chat.
  • Slack — Search messages, channels, and files, and reference Slack conversations in chat.
  • Zapier — Reach thousands of apps through Zapier's MCP endpoint.

Each connects per-user via OAuth and is gated by a feature flag — contact your Revelry Ai team to enable them for your organization.

Learn more: Integrations Overview

Artifacts as a Built-in Tool​

Artifacts are now a togglable built-in tool in chat, enabled by default. Turning it off disables the entire artifact toolset — so the AI can neither create/update artifacts nor discover existing ones — which is often preferable when you want a focused research or analysis conversation. Skill-level configuration is respected in both Workflows and the v2 API.

Persistent Chat Defaults​

Your model selection and tool configuration now persist across sessions. The model and MCP tools you choose are remembered automatically, so each new chat starts with your preferred setup instead of resetting to defaults.

Structured Output (API)​

The API now supports a structured_output_schema option for generation, letting developers request JSON output that conforms to a provided schema. Supported on both OpenAI and Anthropic models.

Learn more: API Usage Guide

Super Admin — Credit Activity Dashboard​

Super admins get a new per-organization credit usage dashboard with a sortable table and direct links to the source chats, workflows, and documents that consumed credits.

Insufficient Credit Prompt​

When an organization is out of credits, a clear modal now explains the situation and routes admins to upgrade — either through in-app navigation or an external upgrade URL.

UX Improvements​

Redesigned Artifact Sharing​

Sharing an artifact now uses a streamlined two-state dropdown: share and copy the link in one place, or unshare just as easily. A "Sharing On" pill shows current status at a glance, copy actions confirm with a "✓ Copied!" indicator, and shared links open in a new tab. Sharing controls are gated behind the Manage Content permission.

Empty Collections State​

Projects with no collections yet now show a helpful empty state, including a permission-gated link to create your first collection.

Performance & Reliability​

  • Long-running generations are more resilient — the LLM request timeout was extended to 30 minutes for lengthy multi-step work.
  • Log and trace sanitization was hardened platform-wide so sensitive fields are never written to logs, error reports, or traces.

Bug Fixes​

  • Fixed rejection of Groq tool-call result messages that included metadata.
  • Guarded chat against errors when mounting without an active organization.
  • Fixed Amplitude OAuth connection for public-client authorization.