When Ng Tian Chong first started experimenting with Microsoft 365 Copilot, he expected a productivity helper; what he found was an idea accelerator and a
sparring partner that helped reshape strategy, speed up decision-making, and — in a memorable personal aside — even recommend the bait that landed him a 40‑pound cobia.
Background
Singtel Singapore’s CEO Ng Tian Chong joined the company to lead the consolidation of its consumer and enterprise businesses and to drive greater agility across a workforce of roughly 5,000 people. As part of that transformation, Singtel has publicly framed a move to become an
AI‑first telco, embedding generative AI into both internal workflows and customer‑facing services. Ng’s account of using Microsoft 365 Copilot — first as a fast search/summarization tool, and later as an integrated research and idea‑generation assistant — illustrates how C‑suite leaders are beginning to fold AI into strategic work, not just tactical tasks. Microsoft’s own Source Asia profile of Ng captures a specifically modern executive workflow: Copilot for quick briefs and email summaries, and Copilot’s Researcher agent for deeper, multi‑step investigations that combine organizational data with trusted external sources. That change in role — from inbox triage to strategic ideation partner — is central to the Singtel story and echoes larger enterprise trends around Copilot adoption.
What Ng actually did with Microsoft 365 Copilot
From search and summaries to strategy
Ng’s journey followed a predictable adoption curve: curiosity, skepticism, cautious optimism, then strategic use. Initially he used Copilot as a smarter search tool — to summarize email threads and the day’s headlines, and to reduce the cognitive load of preparing for back‑to‑back meetings. Over time, he began to use the Copilot Researcher agent for structured research: feeding the agent internal and public data about the telco market and asking it to propose strategic themes. The Researcher output helped him frame options for a refreshed corporate vision and to “unlock more insights” in a compressed time window.
A memorable test: the 40‑pound cobia
Ng also recounted an anecdote that underlines a different kind of value — serendipity and contextual advice. He asked Copilot about bait choices for deep‑sea fishing and adjusted his approach as suggested, with successful results. The fishing story functions as an illustrative human vignette: AI is now something executives consult in day‑to‑day life, not only in spreadsheets and decks. That anecdote originates in the Microsoft feature profile and should be read as a personal, self‑reported vignette rather than an independently verified technical outcome.
Overview: What Microsoft 365 Copilot and Researcher actually are
Copilot as a productivity layer
Microsoft 365 Copilot is Microsoft’s generative AI assistant embedded across Word, Excel, PowerPoint, Outlook, Teams and other Microsoft 365 apps. It’s designed to automate routine tasks, generate drafts, produce summaries, and surface contextual insights from files and communications your account can access. For many users, the immediate benefit is time saved on repetitive tasks and faster access to a synthesis of information spread across calendars, email, documents and chats. Independent reporting and Microsoft’s product pages describe similar usage benefits: faster drafting, meeting recaps, and email summarization among the more common, everyday gains.
Researcher agent: deep, source‑cited research inside your tenant
Researcher agent is a named Copilot capability engineered for multi‑step research tasks. Unlike the standard, quick-response Copilot chat, Researcher is designed to:
- Pull from both internal work content (files, emails, meeting transcripts) and the web
- Spend more time reasoning and synthesizing
- Produce structured, source‑cited reports with headings, bullets, and visuals
- Ask clarifying questions and iterate toward a final deliverable
Microsoft’s Learn documentation and support pages lay out Researcher’s purpose and workflow: it’s meant for complex, decision‑oriented tasks where citations and traceability matter. That is the function Ng describes when he uses Copilot to test strategic themes for Singtel.
Why this matters for enterprise IT and business leaders
Productivity gains and role amplification
- Executives like Ng are applying Copilot to reduce preparation time before meetings and to avoid burdening their senior teams with briefing work. This amplifies leadership capacity by preserving senior time for judgment and stakeholder engagement rather than information digestion.
- Across industries, case studies and enterprise surveys show similar patterns: users report faster draft creation, better meeting summaries, and quicker catch‑up on missed discussions. Those benefits have been central to Microsoft’s pitch and independent reporting on Copilot’s impact.
Faster insight cycles = faster decisions
Researcher’s ability to combine internal signals with vetted external sources shortens the intelligence cycle. For telcos — where market shifts (regulatory, competitive, technological) produce rapid strategic inflections — compressing insight cycles to hours or minutes can create measurable advantage in bid responses, partner negotiations, and product timing. Ng’s use of Copilot to surface strategic themes is a concrete example of that capability in practice.
The wider telco context: Singtel’s AI push
Singtel has publicly signalled an AI‑forward strategy — investing in internal training, agentic approaches to network operations, AI‑driven customer care, and partnerships that accelerate enterprise AI adoption. Those systemic investments increase the value of a CEO who can use AI tools natively in daily work: the organization aligns around the same models and expectations. Independent trade reporting on Singtel shows the company investing in AI, training and agentic automation — all consistent with Ng’s emphasis on an AI‑first posture.
Strengths demonstrated by Ng’s Copilot usage
- Integration with real workflows: Ng shows how Copilot moves beyond novelty to become part of day‑to‑day leadership work: email briefs, meeting preparation, and strategic research. That integration is a decisive step beyond one‑off productivity hacks.
- Researcher’s traceability: For strategic decision support, traceability and source citation matter. Researcher’s design for citation and structured output improves trustworthiness for executive usage.
- Speed + creativity: Copilot surfaces alternative angles and themes quickly, enabling leaders to iterate on vision and strategy in compressed timeframes — an advantage when competitors are also racing to adapt.
- Adoption as cultural signal: A CEO using Copilot visibly signals that the organization takes AI seriously; this can accelerate internal adoption, reskilling, and a shift toward data + model‑driven operating practices.
- Practical, cross‑context value: From board level strategy to personal tasks (the fishing anecdote), Copilot’s utility is broad, which helps normalize the tool across different employee levels and use cases.
Risks, limitations, and governance considerations
While the gains are real, the Singtel example also exposes critical operational risks that every IT leader must manage before broad Copilot adoption.
1) Hallucination and factual errors
Generative AI can produce fluent but incorrect assertions. Even with Researcher’s citations, outputs can conflate sources or cite low‑quality web content. Leaders must treat Copilot outputs as
drafts for human review, not finished decisions. Microsoft’s product docs explicitly advise human validation for Researcher outputs.
2) Overreliance and skill atrophy
A risk of treating Copilot as a cognitive crutch is that teams may progressively offload critical thinking tasks, losing deep domain expertise over time. The benefit is time; the tradeoff is potential loss of institutional judgement if the organization does not pair tools with robust learning and oversight.
3) Data privacy, compliance and data residency
Copilot accesses content within users’ Microsoft 365 environment; enterprises must ensure that Copilot’s workspace access, logging, and data handling meet regulatory and contractual obligations. Telcos, in particular, operate under strict data and national security frameworks; governance controls (data connectors, tenant policies, Entra configurations) are non‑negotiable. Microsoft provides enterprise controls and commercial data protections, but implementing them is an IT responsibility.
4) Security and supply‑chain exposure
Using Copilot integrates yet another cloud service and model surface into the organization’s attack surface. Model choice, API routing and third‑party model integrations warrant careful supplier risk assessment and contract terms around liability and model use. Recent product developments that allow multiple model backends underscore the need for governance of which models and vendors your tenant may route queries to.
5) Cultural and ethical risks
Deploying Copilot without structured training can amplify bias, produce tone or content misalignments, or expose staff to ethically questionable outputs. A policy framework for acceptable Copilot usage and a training program in prompt engineering and critical review are essential.
Practical roadmap for enterprise rollout (recommended steps)
- Define the scope and business objectives
- Identify use cases — executive briefs, knowledge‑worker drafting, customer support augmentation — and define success metrics (time saved, number of drafts produced, reduced turnaround).
- Pilot with a controlled group
- Start with leadership and a few cross‑functional squads; measure outputs, user feedback, and error rates.
- Implement governance and configuration
- Set tenant‑level policies for data connectors, logging, and model selection; enable Microsoft’s enterprise controls. Ensure legal signoff for sensitive workflows.
- Training and “prompt engineering” enablement
- Run focused workshops to teach employees how to craft precise prompts and how to validate Copilot outputs; create prompt templates for recurring tasks.
- Embed human review and audit trails
- Make human sign‑off part of decision pipelines for high‑impact outputs; capture outputs and revisions for audit and continuous improvement.
- Scale gradually and monitor ROI
- Use adoption dashboards and usage metrics to guide expansion; reassign time saved to higher‑value work and reskilling programs.
- Iterate on the model mix and agent configuration
- As your needs evolve, evaluate multi‑model routing choices and agent configurations (e.g., Researcher vs. standard Copilot) to optimize accuracy and cost.
Governance checklist for telcos and regulated enterprises
- Data residency and sovereignty: confirm where model inference occurs and whether data leaves required jurisdictions.
- Access controls: apply role‑based permissions to limit Copilot’s access to sensitive repositories.
- Logging and traceability: capture prompts, responses and any downstream edits for compliance and incident response.
- Vendor risk: review model provider service terms, uptime SLAs, and incident reporting.
- Human‑in‑the‑loop: mandate human review for regulatory, contractual or safety‑critical decisions.
- Training and culture: invest in training programs (prompting, model literacy, red‑team testing).
Measuring success and ROI
- Time saved per task: track before/after baselines for common workflows (email triage, meeting prep, first draft creation).
- Quality improvement: measure error reduction or draft quality improvement via human review panels.
- Adoption rates: monitor daily/weekly active users and the diversity of use cases.
- Business outcomes: correlate Copilot‑enabled activities with tangible business results (faster RFP responses, reduced time‑to‑market, improved NPS).
- Risk metrics: monitor incidence of hallucination‑related errors, data leakage events, and compliance exceptions.
These metrics convert anecdotal wins — like Ng’s faster strategic iterations — into quantifiable operational improvements that justify scale.
The strategic tradeoffs: speed vs. certainty
Ng’s story encapsulates the essential tradeoff organizations face: Copilot speeds the work of thinking and drafting, but speed does not automatically equal quality. The right organizational approach treats Copilot outputs as
amplifiers of human judgment, not replacements. That means pairing agentic productivity tools with strengthened governance, clearer accountability, and continuous training — especially in sectors like telecommunications where customer trust and regulatory compliance are core.
What to watch next (near‑term signals)
- Model diversity and routing: Microsoft has been expanding model options within Copilot (including third‑party backends), which will affect latency, accuracy and vendor exposure. IT leaders should watch how model routing features evolve and whether they can select preferred vendors at the tenant level.
- Agentic automation: Copilot’s agent mode and Researcher workflows are moving toward persistent, multi‑step automation that can execute parts of a process inside Office files. Expect governance and auditing features to mature alongside agent capabilities.
- Regulatory focus: expect regulators and auditors to increase scrutiny on model use in regulated sectors; proactive compliance posture will be a competitive advantage.
Conclusion
Ng Tian Chong’s account of Microsoft 365 Copilot at Singtel Singapore is a compact exemplar of how senior leaders are converting generative AI from a productivity gimmick into a strategic instrument. The trajectory — from using Copilot to summarize email to employing Researcher as a strategic ideation partner — shows how AI can materially shorten insight cycles and multiply leadership capacity. That said, the gains are conditional on responsible deployment: governance, human‑in‑the‑loop review, model‑choice discipline, and continuous upskilling.
For IT and business leaders, the practical takeaway is clear: treat Copilot as a platform that requires orchestration. Do not let the allure of rapid gains eclipse the need for controls and validation. With structured pilots, strong governance, and a culture of verified usage, Copilot and Researcher can be powerful enablers of speed, confidence and creativity — not replacements for judgment. Singtel’s public move to an AI‑first posture and Ng’s personal use cases provide a real‑world case study in what’s possible when C‑suite leadership embraces and governs generative AI with intention.
Source: Microsoft Source
Singtel Singapore CEO uses Microsoft 365 Copilot to unlock ideas – and land big fish - Source Asia