v3.11.0

Released October 23, 2026

Introduces two new capabilities — an optional AI Firewall content guard (beta) and Audio Transcription — both limited‑availability, off by default and enabled per deployment.

Forward‑only upgrade. This release includes database schema additions (for Audio Transcription) and cannot be rolled back to the previous version once applied. Self‑hosted and Ask Sage Edge deployments should plan accordingly; managed (SaaS) tenants are upgraded by Ask Sage.

Confirming this release. Where enabled for your deployment, a Transcription entry appears in the side navigation, and when the AI Firewall (beta, limited availability) is on, a blocked message shows an in‑chat “blocked by policy” notice instead of a normal reply. Both capabilities are limited‑availability, off by default, and enabled per deployment — confirm availability with your Ask Sage administrator.

🛡️ AI Firewall (content guard) — Beta

Beta — limited availability. The AI Firewall is a beta feature offered to a limited set of customers. Contact Ask Sage to request access.

An optional, policy‑driven content guard can now screen prompts — and, optionally, model‑requested tool calls and model responses — through a configurable security provider before they reach a model or return to the user. When content violates policy it is blocked (or, where enabled, redacted), and the chat shows a clear “blocked by policy” notice in place of an answer.

Off by default; enabled per deployment. An administrator turns it on and configures the provider, endpoint, and API key. It supports pluggable providers, so a deployment can use a supported security vendor or its own rules.
Configurable fail posture. On a provider timeout or outage, a deployment can fail closed (block) or fail open (allow); users see a clear, distinct message when screening is temporarily unavailable.
Web‑search aware. Outbound web‑search queries and URL fetches are screened too, and a blocked search never leaves the boundary.
Audited. Enabling the firewall, configuration changes, and block/redact actions are recorded; matched content itself is never logged.

🎙️ Audio Transcription

Limited availability. Audio Transcription is offered to a limited set of customers. Contact Ask Sage to request access.

A new premium capability turns an uploaded audio recording into a speaker‑labeled, verbatim transcript.

Upload an MP3 or M4A recording (up to 1 GB and about 238 minutes); up to three jobs can run at a time.
Named speakers. Provide a speaker roster and the transcript labels speakers by name where the recording makes the match clear; unmatched voices keep a neutral label.
Verbatim formatting with optional filler‑word removal, a choice of timestamp / line‑number / no line prefix, an optional header, and optional custom formatting instructions.
Data‑control labels (PII/PHI, CUI, ITAR/EAR, and 10 U.S.C. § 1102) select a compliant model for processing and drive retention (for example, PII/PHI and § 1102 transcripts are kept for 72 hours).
Email and in‑app notification when a transcript is ready or if a job fails.
Off by default; enabled per deployment. When enabled, a Transcription entry appears in the side navigation.

🔒 Security & platform

Ask Sage services are delivered as FIPS‑hardened, digest‑pinned, and cosign‑signed images within the FedRAMP Class D authorization boundary, and are continuously rebuilt against the latest patched base images as part of Ask Sage’s continuous‑monitoring commitment.