When One New Zealand announced that its order provisioning cycle had collapsed from roughly ten days to under ten minutes, the telecommunications industry took notice. This dramatic compression was not merely a speed boost; it signalled a deeper rethinking of how legacy processes can be reshaped through intelligent automation. Built on UiPath’s Maestro orchestration engine, the initiative sits inside a broader AI-first transformation programme that the telco has been rolling out across its customer‑facing and back‑office operations. The achievement shows that even highly fragmented environments—spanning digital storefronts, Salesforce CRM, and Oracle ERP—can be woven together into a seamless, auditable flow. For decision‑makers watching similar digital‑transformation journeys, the case offers a concrete benchmark: a process that once lived in the realm of manual hand‑offs and tribal knowledge now runs with machine‑precision, yet still retains human oversight where it matters most. The result is a template for how automation can move beyond isolated task bots to become an enterprise‑wide nervous system. Moreover, the project highlights the importance of treating automation as a strategic capability rather than a tactical shortcut, encouraging leaders to invest in platforms that provide end‑to‑end visibility, orchestration flexibility, and the ability to evolve as business needs change. By sharing the lessons learned from this undertaking, One New Zealand hopes to inspire other carriers to reconsider where they apply automation and to aim for transformations that deliver both operational efficiency and strategic insight.
The original provisioning workflow was a manual relay that passed an order through dozens of touchpoints spread across an offsite team, with little visibility into what happened at each step. Because the process touched multiple systems—Salesforce for customer capture, Oracle for billing and inventory, and various digital portals for device configuration—any change required a cascade of emails, spreadsheets, and phone calls. This fragmentation meant that even a simple SIM swap could linger for days while analysts chased down missing data or waited for a colleague in another time zone to finish their part. By contrast, the automated flow built with UiPath Maestro pulls data from each source in real time, validates business rules orchestrated across the heterogeneous landscape, and pushes the final provisioning command back to the network inventory without human intervention. The shift eliminates latency caused by hand‑offs and replaces guesswork with deterministic logic, giving the telco a predictable, repeatable engine that can scale with order volume. In addition, the automation creates a rich audit trail that captures every system interaction, enabling managers to analyse performance trends, identify bottlenecks, and simulate the impact of future changes before they are introduced into production. This data‑driven foundation not only improves day‑to‑day operations but also supports longer‑term planning initiatives such as capacity forecasting, promotional effectiveness studies, and regulatory compliance reporting.
Cy Wright, General Manager at One New Zealand, emphasized that the headline‑grabbing speed improvement was actually a secondary benefit. According to Wright, the true value lay in the visibility that the automation exposed. When a process is reduced to a set of executable steps that can be monitored, measured, and analysed, managers gain a diagnostic lens that was previously unavailable. In the old world, understanding why a particular order stalled required interviewing the offsite team, piecing together anecdotal evidence, and hoping that someone remembered the exact nuance of a past case. The new system logs every decision point, every system call, and every exception, turning tacit know‑how into explicit data. This transparency enables leaders to pinpoint training gaps, assess the impact of configuration changes, and simulate the throughput effect of process tweaks before they are rolled out. Moreover, the visibility supports proactive customer service: agents can see exactly where an order is in the pipeline and provide accurate status updates, reducing inbound inquiries and improving satisfaction. By converting hidden workflow logic into observable metrics, One New Zealand has created a feedback loop that continuously refines both the automation and the underlying business processes.
One of the biggest obstacles the project faced was the absence of documented procedures. The provisioning process had evolved organically over years, sustained largely by tribal knowledge that lived in the heads of a few veteran operators. When the automation team tried to map the workflow, they discovered that key decision rules were never written down; they were inferred from past examples or communicated through informal chats. This lack of articulation made it difficult to design accurate bots, as any assumption risked being wrong. To overcome this, the team adopted an iterative approach: they would prototype a fragment of the flow, run it against live orders, observe where the bot behavior diverged from human handling, and then refine the rules. Each cycle revealed another hidden nuance—perhaps a specific promo code that triggered a different inventory allocation, or a regional regulation that required an extra validation step. By treating the live environment as a learning laboratory, the project gradually codified the invisible logic that had kept the process running, albeit inefficiently. This approach also produced a living documentation set that accompanies the automation, making future updates easier and reducing reliance on individual memory.
The reliance on live orders for validation extended the testing timeline far beyond the initial plan. Originally scheduled for one week of user acceptance testing in a sandbox, the effort stretched to three weeks because the test environments could not reproduce the full variety of real‑world scenarios. Network promotions, device‑specific firmware quirks, regional regulatory constraints, and even seasonal spikes in order volume introduced variables that simply did not appear in the synthetic data sets. Consequently, the team had to wait for actual customer orders to surface these edge cases, observe how the automation responded, and adjust the orchestration accordingly. Although this approach lengthened the calendar, it produced a far more robust solution: the final workflow could handle the full spectrum of conditions that the telco encounters in production, reducing the risk of post‑deployment surprises and building confidence among stakeholders that the automation would not break under pressure. The experience underscores a common pitfall in automation projects—over‑reliance on idealised test data—and offers a practical remedy: incorporate real‑world variability early, even if it means extending timelines, to ensure that the resulting automation is resilient and adaptable.
Choosing to start with the most complex provisioning flow instead of a low‑hanging‑fruit task was a deliberate strategic bet. Wright explained that the team’s logic was simple: if they could make automation work in a setting riddled with system heterogeneity, manual hand‑offs, and undocumented rules, then replicating that success elsewhere would be far easier. The initial proof‑of‑concept, while promising, was deemed insufficient until it proved it could operate across the actual Salesforce‑Oracle‑digital landscape without breaking. This mindset flies in the face of conventional automation advice, which often recommends beginning with repetitive, rule‑based tasks to win quick wins and build momentum. By tackling the hard problem first, One New Zealand not only de‑risked future initiatives but also forced the organization to confront its data‑quality and integration gaps early, turning those weaknesses into foundations for a more resilient automation platform. The experience also demonstrated that investing in a challenging use case can generate broader organisational learning, as teams develop skills in cross‑system mapping, exception handling, and stakeholder management that prove valuable for subsequent automation efforts.
Feiran Hao, Vice President of Pricing and Product Strategy at UiPath, noted that the One New Zealand project exemplifies a broader shift in how automation vendors are positioning their platforms. Traditionally, the selling point of robotic process automation (RPA) centered on automating high‑volume, low‑complexity tasks to deliver rapid cost savings. Hao observed that the market is now moving toward orchestrating end‑to‑end, dynamic workflows that can adapt to changing conditions and span multiple systems. The availability of coding agents—AI‑assisted development tools that generate or suggest code snippets—has lowered the barrier to building sophisticated automations quickly. As a result, what once was considered a niche capability for achieving quick wins has become a viable strategy for large‑scale transformation. This evolution means that enterprises can now pursue ambitious automation goals without first having to exhaust a catalogue of simple tasks, thereby accelerating the timeline for realizing strategic benefits. Moreover, the emphasis on orchestration aligns with the growing demand for composable architectures, where businesses can mix and match services to create tailored solutions that evolve with market demands.
One New Zealand’s approach to automation is anchored in a formal responsible AI policy complemented by an internal upskilling programme called Elevate. Every software agent that is deployed must undergo a privacy and compliance review, irrespective of the business function it serves. This gatekeeping ensures that the automation does not inadvertently expose customer data, violate regulatory mandates, or introduce bias into decision‑making processes. Wright stressed that the governance framework is non‑negotiable: the potential harm of a misrouted device or an incorrect service activation outweighs any efficiency gain. To enforce this, the organization employs a model where small, focused teams build the automation components while a larger pool of subject‑matter experts validates the outputs. Any deviation from the predefined guardrails triggers an immediate halt and routes the case back to a human reviewer, preserving the safety net that protects both the customer and the brand. Additionally, the Elevate programme provides continuous learning opportunities, ensuring that developers stay current with best practices in AI ethics, data security, and change management, which further strengthens the organisation’s ability to scale automation responsibly.
Accountability is another cornerstone of One New Zealand’s automation philosophy. Wright was explicit in stating that AI is merely a tool, and ultimate responsibility for any automated outcome rests with the humans who design, deploy, and monitor it. This mindset eliminates the temptation to blame the algorithm when something goes wrong and instead encourages a culture of continuous improvement. Complementing this, Hao highlighted UiPath’s belief that certain decisions—especially those that directly affect customer experience—should remain in human hands. The platform can analyse data, recommend actions, and even execute routine steps, but the final approval must come from a person who can see the full context. Crucially, that human decision is backed by full traceability: the system logs why the AI made its recommendation, which data points influenced it, and what alternative paths were considered. This audit trail not only satisfies regulators but also empowers teams to learn from each incident and refine the underlying models. By maintaining a clear line of responsibility and providing transparent insights into automated decisions, One New Zealand builds trust with customers, regulators, and internal stakeholders alike.
Looking ahead, Wright revealed that the provisioning automation is just the first of many initiatives. His team maintains a backlog of roughly thirty additional processes spread across the order‑to‑cash lifecycle, with new use cases being added daily as operational pain points surface. The immediate priority, alongside delivering these new flows, is to build internal capability so that the telco’s own staff can own, operate, and enhance the automations rather than relying on external consultants or vendors. This capability uplift involves training developers on UiPath’s Maestro, establishing communities of practice that share reusable components, and instituting metrics that measure both efficiency gains and quality outcomes. By embedding automation expertise inside the organization, One New Zealand aims to create a self‑sustaining engine of innovation that can adapt to market shifts, regulatory changes, and emerging technologies without constantly seeking outside help. Furthermore, the organization plans to leverage the data generated by these automations to feed predictive analytics models, enabling proactive inventory management, churn reduction, and personalized offer generation.
The impact of the provisioning project has already rippled beyond One New Zealand’s walls. Wright shared that other telecommunications carriers and enterprises facing similar legacy‑system sprawl have begun reaching out for guidance, viewing the automation as a live demonstration of end‑to‑end agentic orchestration at production scale. When he presented the flow to his own board, the ability to watch an order traverse Salesforce, Oracle, and digital channels in real time, with every step visible and auditable, proved to be a compelling narrative that secured further investment and enthusiasm. External observers are particularly interested in how the telco balanced speed with governance, how it dealt with the lack of test environments, and how it cultivated human‑in‑the‑loop controls. As a result, One New Zealand is positioning itself not just as a beneficiary of automation but as a thought leader that can help peers navigate the complexities of modernizing entrenched business processes. This knowledge‑sharing role also opens avenues for potential partnerships, joint‑venture pilots, and industry‑wide standards that could accelerate automation adoption across the sector.
For organizations contemplating a similar leap, several actionable insights emerge from One New Zealand’s experience. First, identify a process that is both strategically important and notoriously opaque—automating it will force you to surface hidden rules and data quality issues that, once resolved, benefit the entire enterprise. Second, invest in building realistic test scenarios or be prepared to iterate with live traffic; synthetic data often fails to capture the nuance that drives exceptions. Third, establish a clear governance model from the outset, combining automated guardrails with mandatory human review for high‑impact decisions, and ensure every action is traceable. Fourth, treat the automation initiative as a vehicle for upskilling: create internal centres of excellence, document learnings, and encourage cross‑team collaboration so that knowledge does not remain siloed. Fifth, communicate the visibility gains to stakeholders; the ability to monitor and analyse the process in real time often delivers more value than the raw speed improvement. Finally, start small enough to learn, but think big enough to transform—choose a challenging use case that proves the platform’s mettle, then scale the approach across the organization. By following these steps, companies can transform fragile, manual workflows into resilient, automated engines that drive both efficiency and insight.