aiversum
Status register

The aiv capability register, with dates.

Every capability from this site in one place, with a state and a date. Of the eight vendors whose sites we checked in September 2026, none marks the state of its features. Regulated-sector clients check exactly that.

As of 26 September 2026
State legend
Live at a client
Running on our first production client's machine today. Not merely in the repository: running.

Security

  • Model in the client's tenant, managed identity, no keysLive at a client
  • Machine with no public address, Poland CentralLive at a client
  • CRM access through seven named tools, read-onlyLive at a client
  • Sending only after a human approves the draft, approval in the logLive at a client
  • Client Entra ID sign-in, certificate from client PKI, DefenderLive at a client
  • Engine log (sessions, tools, model) and panel log (who, when, which file)Live at a client
  • Decision identity from authentication, rejection and logging of requests without identity
  • Human roles in the panel (e.g. manager, administrator, auditor)
  • Log review in the panel: filter, statistics, daily download
  • Hash chain in the engine log (detects changed entries)
  • “Needs attention” flag for suspicious cases
  • Draft decision logged as an intent and result pair
  • Agent behaviour alerts for the SOC
  • “Stop agent” button in the panel
  • Emergency kill switches in service configuration (autonomous work, engine service)

Engine and ways of working

  • Browser panelLive at a client
  • Desktop app
  • Command line
  • Event-driven work (a new message starts the agent)Live at a client
  • Several agents with separate scopes in one environment
  • Delegation to sub-agents with verification
  • Ready-made role set: supervisor, gathering, analysis, document
  • Agents working remotely on computers
  • Process guardian agent (autonomy within set limits)
  • Network of department agents with a supervisor
  • Board report from agents' work
  • Skills (written process instructions) and agent plugins
  • ML predictions in the agent's work
  • Confidence threshold routing a case to a person
  • Scheduled work
  • Built-in reviewer agent checks another agent's work
  • Task contract: allowed files, required evidence, budget
  • Company status reporting agent (a program calculates, the model describes, a person approves)

Models and integrations

  • Azure OpenAI in the client's tenantLive at a client
  • The client's file share as the knowledge sourceLive at a client
  • Claude in Microsoft Foundry, Anthropic, OpenRouter
  • Locally hosted model
  • Database adapters MS SQL, PostgreSQL, SQLite
  • Tools that write to client systems
  • Client server room, container
  • Log shipping to the client's Azure storage (every 5 minutes)
  • Agent events in the client's SIEM (e.g. Microsoft Sentinel)
  • Reading data from the client's ERP and data warehouse

Roles and knowledge

  • Customer service: correspondence, CRM, draftLive at a client
  • Finance
  • Back office
  • Company knowledge in agent-ready form (source, date, confidence, expiry)
  • Agent assisting SOC analysts with alert triage

Client systems (integrations section, as of 26 Sep 2026)

  • SQL Server — Database adapter · scoped read access
  • PostgreSQL — Database adapter · SQL queries
  • SQLite — Local databases and application data
  • Snowflake — Warehouse data in an agent workflow
  • Databricks — Data platform and lakehouse environment
  • Microsoft Fabric — Warehouses, lakehouse and analytics data
  • BigQuery — Analytics data in Google Cloud
  • Oracle Database — Business databases and domain applications
  • Amazon Redshift — Data warehouse in the AWS environment
  • MongoDB — Document databases for applications
  • Microsoft 365 — Mail and documents · access via Graph
  • Dynamics 365 — CRM / ERP records and business processes
  • Salesforce — Customer data and sales workflows
  • SAP — ERP workflows and operational data
  • IBM i / AS/400 — On-premises systems and legacy applications
  • Comarch ERP — Business documents and workflows
  • Files & network shares — Knowledge sources in your environment
  • API / MCP / CLI — Named tools with defined permissions

Aiversum EDU platform (edu.aiversum.ai, confirmed by Damian 26 Sep 2026)

  • Exercises in every courseLive at a client
  • An exam after every courseLive at a client
  • Reels: what you can use todayLive at a client
  • Reminders and reviewsLive at a client
  • AI mentorLive at a client
  • Progress your company can seeLive at a client

The aiv capability register, with dates.

Book a demo