Skip to main content
Version: 2.5

Roadmap Overview

The tSM roadmap describes where the platform is going, what is already available, and which capabilities are still planned. The current documentation set is the unreleased tSM 2.5 version. Use the documentation version selector for the frozen tSM 2.4 documentation.

Roadmap status

Items marked 2.5 target describe the intended scope of the unreleased version and may be refined before release. Production availability is confirmed by the release notes. Horizon items describe the direction beyond tSM 2.5.


Our Vision

1) Intelligence in Context

AI moves from a configurator-only assistant to a universal, context-aware capability:

  • The same assistant entry point is available across configuration and business screens.
  • The active screen, entity, selection, role, and configured scenario determine what the assistant can read and which actions it may propose.
  • Answers identify the evidence used; changes are proposed as typed, reviewable actions.
  • Long-running or high-impact work is handed to the Process Engine for approval, retry, audit, and recovery.

2) Builder Freedom, Enterprise Control

  • Work in the application or an IDE and publish configuration through review and versioning.
  • Reuse Assistant Profiles, context providers, task templates, tools, and process patterns.
  • Keep authorization, data access, confirmations, and audit deterministic even when AI is involved.

3) Open by Design

  • Public APIs, BPMN, REST, Kafka, MCP, and governed A2A bindings provide explicit integration boundaries.
  • The Process Engine remains the durable control plane; AI runtimes and model providers stay replaceable behind contracts.
  • TM Forum alignment guides domain interoperability without coupling tSM to one vendor runtime.

Version Boundary

AreaAvailable in tSM 2.4tSM 2.5 target
AI shellTop-bar AI chat, streamed responses, conversation threads, image attachments, and feedbackOne universal shell that resolves a controlled assistant scenario from the active screen
Configuration assistanceDocumentation and SpEL assistance, including contextual script reviewForm, TQL, and BPMN assistance using the same draft → validate → diff → accept workflow
ExtensibilityPublic API, REST script bindings, MCP script bindings, and project-specific assistantsVersioned Assistant Profiles, server-side context providers, typed actions, evaluation sets, and release controls
Business workInitial chatbot integrations and pilotsContextual assistance on customer, ticket, order, listing, and other selected business screens
Process automationBPMN, user/service/external/Kafka tasks, task templates, events, transactions, versioning, and auditAgent Tasks, Agentic Subprocesses, outbound A2A Agent Tasks through tsm-ai with configurable state mappings/listeners, inbound A2A Script Bindings through Gateway, approval-gated actions, and correlated AI/process operations
Engine foundationtSM-maintained Process Engine embedded in each participating domain serviceContinue the embedded architecture and integrate qualified maintenance, security, observability, and agentic-automation improvements

The 2.5 roadmap preserves the 2.4 baseline and extends it through shared configuration, runtime services, and governance.

What 2.5 Changes

  • For business users: answers and next-best actions appear in the record or listing being used.
  • For configurators: a scenario is assembled from reusable instructions, data providers, tools, permissions, limits, and evaluation cases instead of custom backend code for every screen.
  • For operators: AI and process activities expose evidence, tool calls, approvals, retries, incidents, costs, and final business impact in one correlated view.
  • For architects: the tsm-ai Agent Runtime owns internal Agent Runs; its Outbound A2A Client Runtime owns remote-agent transport and the local invocation projection; tSM Gateway owns the ledger only for inbound A2A Script Binding Tasks; and the remote agent owns its authoritative Task. The embedded Process Engine coordinates them through durable wait states, state mappings, and response listeners.

The Themes

  • AI Smart Operations — universal contextual assistance and explainable actions on business screens.
  • AI Assisted Development — generation and validation of configuration artifacts with human control.
  • Process Automation & Agentic Orchestration — durable, observable automation across people, systems, and bounded AI agents.
  • Reporting & Data Management — from static outputs to conversations with governed data.
  • Security & Governance — least privilege, explicit approvals, versioning, audit, and evaluation.
  • VS Code Extension — IDE workflows connected to the same platform capabilities.
  • TM Forum Alignment — open domain and interoperability principles.

Want the nuts-and-bolts? See the roadmap pages in this section.


How We Ship

tSM releases balance stability with momentum:

  • Patch~monthly / as needed (security immediate)
    Bug/security fixes and UX polish for the current version; tiny add-ons only if needed to support existing capability.
    Included in Product Maintenance Premium.

  • Minorbi-annual
    New features and enhancements within existing modules, designed for backward-compatible, low-friction upgrades.
    Included in Product Maintenance Premium.

  • Majorno fixed cadence (strategic & customer-driven)
    New modules and step-change capabilities; may require deeper upgrade planning. Developed in parallel with minors; selective backports possible ahead of e.g. tSM 2.x.


Horizon

Beyond 2.5, tSM will evaluate predictive process intelligence, bounded closed-loop remediation, cross-domain agent collaboration, and telco-oriented A2A interoperability. Autonomy expands per proven scenario, backed by measured data quality, scenario-specific safety, and business outcomes.

A Word on Flexibility

This roadmap shares our intent and direction. Timelines and content may change with customer priorities and product learning. For purchasing or planning, treat this as guidance — future items can be reprioritized, delayed, or adjusted at DataLite’s sole discretion.