What is Wingspan and why it matters

Onix isn't a startup that fell out of the sky yesterday. It is an 18-time Google Cloud Partner of the Year, working with 30 of the Fortune 100 companies. It has over 500 AI agents in production across telecommunications, retail, healthcare, and financial services. Its flagship — the Wingspan platform — represents what the industry calls "services-as-software": a combination of proprietary software and AI agents that replaces traditional consulting projects.

Wingspan 2.0, introduced this April at the Google Cloud Next 2026 conference, brings a fundamental innovation: the Semantic Twin. This is an automatically built knowledge model of the enterprise that maps data relationships, processes, metrics, and business context — and keeps them continuously up to date. Unlike traditional "digital twin" projects, where manually mapping an ontology takes companies months to years, Wingspan fully automates this process.

Semantic Twin: why it matters

Imagine that every time you launch a new AI project at your company, you spend 6–8 weeks rediscovering what data you actually have, where it comes from, and what it means. That's the reality for most enterprises. According to data cited by Onix, up to 85% of organizations admit their data is not ready for AI deployment. Up to half of AI projects lacking a quality data foundation are estimated to be abandoned by the end of 2026.

The Semantic Twin solves this problem by creating a living, automatically updated "enterprise map" that all AI agents can connect to. Every data relationship, every transformation, every business metric is captured and maintained. When a new project comes along — whether it's a data warehouse migration or deploying predictive analytics — agents already "know" where everything is.

What Wingspan can actually do

The platform covers three key areas:

1. Data platform modernization — Automated migration from legacy systems to the cloud (Google Cloud, but also Azure), including code conversion (Raven tool), data validation (Pelican), and synthetic test data generation (Kingfisher). Onix claims migration happens 3× faster than with traditional approaches, with a 50% reduction in manual work and 100% validation accuracy.

2. Intelligent operations — Autonomous agents monitor and optimize DataOps, AIOps, and FinOps. In practice, this means automatic anomaly detection, root-cause analysis, and remediation without human intervention. Onix reports an average 25% reduction in cloud costs thanks to automatic optimization.

3. Enterprise intelligence — Users can ask questions in natural language ("Talk2Data") and get answers in real time without waiting for the analytics team. The platform also includes pre-built agents for rapid MVPs (minimum viable products) across individual industries.

Google Cloud as the engine

The expanded partnership with Google Cloud, announced in early April 2026, involves an investment of over $500 million in Google Cloud infrastructure and several hundred million dollars in service revenue for Onix. The partnership rests on three pillars: joint go-to-market across verticals, use of Wingspan as a unified platform for data modernization, and a shift from traditional "overstaffed" project teams to AI-assisted "delivery pods" that are paid based on achieved business outcomes.

Victor Morales, Vice President of Google Cloud for Consulting Partnerships, said: "Generative AI is fundamentally changing how companies operate. With Google Cloud's generative AI capabilities, Onix can help customers uncover new insights that improve their operations and productivity."

Notably, Onix isn't "glued" to Google Cloud — the platform also supports Azure and Databricks, which is an important distinction from solutions that only work within a single cloud ecosystem.

European footprint and Czech context

In February 2026, Onix appointed Vittorio Sanvito as Managing Director for the EMEA region (Europe, Middle East, and Africa), clearly signaling its ambitions in the European market. For Czech enterprises — especially those already using Google Workspace, BigQuery, or other Google Cloud services — this means a potentially more accessible path to deploying agentic AI.

Czech companies face the same challenges as their global counterparts: fragmented data in legacy systems, a shortage of data engineers, and pressure for faster ROI on AI investments. Wingspan could be an interesting alternative to the costly consulting projects of large system integrators — provided that Onix can actually deliver EMEA localization and support in the European time zone.

From a regulatory standpoint, it's important that Wingspan runs on Google Cloud infrastructure, which has the necessary certifications for the European market (GDPR, ISO 27001). Onix itself holds ISO 27001 certification for information security management.

Competitive comparison and pricing

The market for agentic AI platforms in the enterprise space is crowded in 2026. Microsoft is pursuing Copilot integrated into Dynamics 365 and Power Platform, SAP is connecting its agents via Joule, AWS is launching its own agent tools on Bedrock. Meanwhile, Anthropic has entered into a global partnership with TCS and Fujitsu, and Deloitte is building its own agent models for wealth management.

Wingspan differentiates itself in three ways:

Automated ontology — The Semantic Twin requires no manual modeling, which is unique. Most competing solutions rely on someone (a consultant) creating the data model for the company — and then maintaining it at high cost. Wingspan bypasses this step through full automation.

Outcome-based model — Instead of billing per hour worked or per user, Onix offers payment based on achieved KPIs. If the platform doesn't meet the promised metrics, the customer doesn't pay the full price. This is still rare in the enterprise market.

Openness — Wingspan is not vendor lock-in. It supports Google Cloud, Azure, and Databricks, meaning a company can use the platform regardless of which cloud it currently runs on.

As for pricing, the Wingspan platform itself starts at $3,999 per year, which is an accessible amount by enterprise standards. Full deployment including Outcome Engineers and the service component is of course significantly more expensive — Onix talks about contracts in the range of millions of dollars for large customers. For comparison: Microsoft 365 Copilot costs $30/month per user, GitHub Copilot Max $100/month. Wingspan doesn't target individuals, but enterprise-wide transformation, and its pricing model reflects that.

Real-world customer results

Onix can point to concrete numbers. For one of the world's largest retail chains, it migrated data pipelines from Snowflake and Azure to Google Cloud with 40% faster runtime and 30% cost savings over five years. Canadian telecom operator TELUS, with Onix's help, unified over 100 data sources, migrated 14 petabytes of data, and reduced the volume of stale data by 30%. Swedish healthcare company Humana moved 13,000 employees to Google Workspace. And tech company Metricwire launched an AI planning system with 90% accuracy and sub-one-second latency for complex requests.

According to Onix's latest data, 70% of managers in organizations that deployed generative AI through Wingspan report improved productivity. In the telecom sector, there was a 44% increase in first-contact resolution, and in retail, a 60% reduction in query time.

Is Wingspan right for your company?

Wingspan makes the most sense for mid-sized and large enterprises that:

• Already run on Google Cloud (or are considering migration)
• Have complex data architecture with many legacy systems
• Repeatedly hit the wall when trying to move AI pilots to production
• Want to pay for results, not for consultants' billable hours

For small businesses and startups, Wingspan is likely "overkill" — there are lighter tools like n8n, Make, or specialized AI agents from smaller vendors. On the other hand: if you're a Czech company with ambitions to scale AI across the entire organization and have a budget in the range of millions of CZK, Wingspan represents an interesting alternative to large system integrators.

The platform is not yet localized into Czech — all interfaces and documentation are in English. But that's to be expected for enterprise tools in this category and shouldn't be an obstacle for technical teams. Support in the European time zone is likely in the pipeline given the EMEA expansion.