From V1 to V2: how DivetIQ became the headless agentic platform mid-market actually needed

DivetIQ V1 launched as a headless mid-market platform - nine modules, one identity model, an agent layer for KPI management. V2 is the version that finally makes the analytical fabric a first-class citizen.

What shipped in V2

Module count moved from nine platforms to ten, with functionality count from roughly 600 to roughly 900. Every V1 API, event and data contract is preserved, with a 24-month deprecation policy for any future change. The additions split into three streams.

New cross-cutting capabilities that every module inherits without redefining them:

• Master Data Management Plus - golden records, survivorship rules, stewardship queues, data-quality scoring per record.
• Knowledge Graph & Semantic Layer - RDF/OWL with SPARQL endpoint plus a property-graph adapter, linking customers, suppliers, assets, contracts, KPIs and processes.
• Privacy, Consent & Trust Center - GDPR/CCPA/LGPD operational tooling, customer-facing trust center, AI transparency notices for the EU AI Act.
• Mobile Super-App Framework - one offline-first shell hosting module-specific experiences with biometric unlock and MDM enrollment.

New module extensions: Treasury & Cash Plus, Subscription & Recurring Revenue Management, Field Service Lite, Project & Portfolio Management Plus, ESG Data Hub, CDP Lite, Supplier Portal, Customer Self-Service Portal, Advanced Process Mining, AI Process Co-Pilot, Process Marketplace, Agent Marketplace.

One entirely new platform: the AI-Based Business Intelligence & Data Warehouse - lakehouse on open table formats, semantic layer, conversational analytics, prebuilt KPI library wired to the agent catalog, embedded analytics SDK, headless analytics API.

Three architectural shifts

V1 stood on two architectural commitments. V2 adds a third.

1. Headless software architecture (unchanged) - every capability exposed via REST, GraphQL and gRPC, with OpenAPI 3.1 schemas published and Kafka-grade events for every state change. Customers replace the UI, embed components, or use only the API layer.
2. Agentic AI workflow layer (unchanged) - specialist agents observe events, plan multi-step actions, and execute through the same APIs your people use. Human-in-the-loop above a configurable risk threshold per agent and per role.
3. AI-native analytical fabric (new in V2) - every transactional event flows into the unified lakehouse with field-level lineage. A semantic layer encodes business meaning once. A generative-AI assistant enables conversational analytics, automated narrative generation, and prescriptive recommendations.

The practical effect: a Cash Position agent in V2 does not query a module database directly. It queries the semantic layer, gets the same definition of "cash" the CFO would, and reasons across CRM, ERP and AP simultaneously.

What got removed (and what didn't)

V2 reaffirms every removal from V1. Mid-market does not need Pillar Two CbCR, FedRAMP, defense-grade air-gap, full digital-twin-of-organization, multi-armed-bandit ad optimization, mainframe terminal emulation or aviation MRO. These stay out.

V2 also avoids the temptation to re-add features the global mid-market does not actually use. The standard for inclusion is unchanged: surveyed adoption across Gartner Magic Quadrant, IDC MarketScape and G2 above 15%, plus genuine cross-module utility.

The net result of consolidation and additions: roughly 130 modules and 1,200+ functionalities in the source spec compressed to ten platforms and roughly 900 functionalities - without losing any capability the mid-market actually uses, and adding the ones V1 underweighted.

V2 is what V1 always intended to be. The analytical fabric was the missing layer. It is no longer missing.