Changelog

A running history of what's shipped in HELM OS, in plain language. For the full technical detail behind any release, contact us.

Latest

Decision Engine, Actions Engine & Execution Modes

Every AI-proposed action now goes through a Decision Engine that classifies risk and decides whether it needs your approval.
New Execution Modes — Guided, Assisted, and Autonomous — let you choose how much HELM does on its own. Customer-facing actions always require approval regardless of mode.
Standing Instructions: set your own rules (e.g. "always ask before emailing deals over €50k") that layer on top of the defaults.
New admin "Decisions & Actions" view shows a full history of what HELM's AI has decided and done.
Mission Control now surfaces real, data-backed recommendations — not just AI chat answers.

Earlier

Describe a goal in plain language (e.g. "find me 500 ideal customers") and HELM builds a reviewable, step-by-step plan to get there.
Every plan is checked for duplicate contacts, tone, and compliance before anything is sent.
Unverified data is now clearly labelled in the interface — HELM never presents an inferred detail as a confirmed fact.

Earlier

New "AI Connections" centre: see every connected integration, its permissions, and its health in one place.
Foundational work to support connecting more tools over time without one-off engineering for each.

Earlier

Upload your own documents (pricing sheets, SOPs, product info) so HELM's AI can reference your business specifically, not just generic sales knowledge.

Earlier

Merged five separate admin screens into one Admin Home with tabs, so admins have a single place to manage users, workspaces, and platform health.

Earlier

Home is now Mission Control: today's priorities, pipeline, meetings, and follow-ups in one view, instead of a generic dashboard.
Simplified navigation from ~40 flat menu items down to 7 clear areas.