Posted on — Leave a comment

Is short covering the only reason Bitcoin moved higher?

Bitcoin gained 0.73 per cent to US$73,937.65 in 24 hours, and Ethereum gained 0.519 per cent to US$2,053.91. The total crypto market rose 0.45 per cent. I read this move as a modest relief bounce after selling eased, not the start of a confirmed new trend. The figures support that view. Both assets moved with the wider market. Neither note shows a distinct asset-specific catalyst. Trading participation fell for both.

For Bitcoin, the main story is that selling pressure eased after hotter-than-expected US inflation data pushed BTC lower on November 10. News coverage linked the November 11 rise to that easing pressure. Short liquidations suggest some traders had to close bearish positions, and that forced buying likely added to the move. Traders who had bet on further declines helped lift prices when they closed those bets. That kind of buying fades once the squeeze ends.

It does not signal that long-term investors have turned more positive. This is not the same as a new bullish driver. It is a partial recovery after a selloff. Bitcoin trading volume fell 14.3 per cent over 24 hours. That drop weakens the case for a durable move. A price gain on lighter participation can reflect reduced selling rather than strong new demand. Bitcoin’s seven-day change remained down 8.06 per cent. The daily uptick therefore sits inside a weaker short-term backdrop.

The ETF flow data adds another caution. US spot Bitcoin ETFs recorded US$95 million in net outflows on November 10. Outflows remove a source of demand. They do not create fresh buying. One linked headline notes that ETF outflows hit a fourth day. I would treat that as a headwind, not a side detail. Bitcoin can rise while ETFs lose assets, but that combination makes the move less trustworthy.

A fourth day of withdrawals points to persistent caution among ETF investors. That flow can weigh on spot prices when it continues. The market can absorb one day of outflows. A run of withdrawals sends a different message about demand. My view is that the market needs more than short covering and eased selling to call this a reversal. It needs stronger participation and steadier demand.

Also Read: CPI at 0.4% broke Bitcoin’s US$73,000 floor. What happens if the Fed goes bigger?

The technical picture gives a clear map. Bitcoin trades near US$73,938. Its seven-day average is about US$74,990. Its 200-day average is about US$73,118. The recent swing low is US$72,689. If Bitcoin holds above that low and retakes US$75,000 to US$76,000, the recovery could extend. If it breaks below US$72,689, the market risks a move toward US$70,000.

I find the first condition more important than the daily gain. Holding the low would show buyers are still defending the recent range. Reclaiming US$75,000 to US$76,000 would show more than a brief rise. It would show that Bitcoin can overcome nearby resistance. Losing US$72,689 would strengthen the downside case and make the current gain look like a brief pause.

Ethereum shows a similar but weaker picture. Ethereum gained 0.519 per cent to US$2,053.91. The note describes the move as broad-market beta. Ethereum moved in the same direction as Bitcoin and the wider crypto market. It did not show a distinct Ethereum-specific catalyst.

Ethereum’s 24-hour volume fell 33.19 per cent. That is a large drop, and it limits confirmation. The Fear reading of 27 reflects cautious sentiment, but it does not explain the price move. Ethereum is down 15.03 per cent over seven days. That weak weekly performance tempers the daily gain.

Also Read: Bitcoin is bruised, not broken: Why the crypto sector is holding its ground

The Ethereum levels matter in the same way. If Ethereum holds the US$2,005 to US$2,040 support area, it could retest US$2,119. A sustained break below US$2,005 could expose US$1,900. My view is that Ethereum’s small gain is more about the wider market than about Ethereum-specific demand. That makes it vulnerable if wider-market buying fades. A support hold would keep a recovery possible.

A support break would shift attention to US$1,900. The note calls the short-term picture neutral, and I agree. That neutral label fits the data. ETH has a nearby ceiling and a nearby floor. The distance between them is modest. A trader should wait for a break before treating either direction as a real signal. The weekly decline and lower participation keep the signal weak.

The macro calendar gives the next scheduled test. US producer price data arrives on Friday, November 13 at 8:30 a.m. ET. Bitcoin’s note flags that data as the next macro trigger. I wouldn’t overstate one report, but it can shape near-term risk appetite.

The recent CPI-driven selloff shows how macro data can move crypto prices. A softer producer price reading could help the rise. A hotter reading could revive selling pressure. The market should watch how Bitcoin and Ethereum react around their key levels after that data, not just the direction of the first move.

My overall point of view is cautious. Bitcoin’s 0.73 per cent gain and Ethereum’s 0.519 per cent gain look consistent with a broader market bounce after the CPI-driven selloff. Short covering likely amplified the move.

Lower participation and ETF outflows argue against treating it as a confirmed trend reversal. Bitcoin faces resistance at US$75,000 to US$76,000 and support at US$72,689. Ethereum faces resistance at US$2,119 and support at US$2,005 to US$2,040.

The key watch is whether Bitcoin can hold US$72,689 and reclaim US$75,000 to US$76,000 after the November 13 producer price data. Ethereum needs to hold US$2,005 for its own recovery to stay alive. Until participation improves and ETF flows stabilise, I would not make any serious moves.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Is short covering the only reason Bitcoin moved higher? appeared first on e27.

Posted on — Leave a comment

Jensen Huang quits Nvidia, and Southeast Asia’s chip hubs inherit the fallout

For 33 years, Nvidia had one face, one voice and one leather jacket. On Thursday, it lost all three.

Jensen Huang, who co-founded the chipmaker in 1993 and built it into one of the world’s most valuable companies, has resigned as president and CEO. According to a regulatory filing, Nvidia’s board accepted his resignation on Wednesday.

That came days after he admitted to a romantic affair with Alice Huang (no relation), a Chinese businesswoman whom US intelligence reportedly believes was working for China’s Ministry of State Security. A separate report in The New York Times alleged a second affair, this one with an Nvidia executive who reported to him.

Chief Financial Officer Colette Kress becomes interim CEO and keeps her finance role. The board also named Harvey Jones, a director since 1993 and former Synopsys chief, as Nvidia’s first-ever chair. The market had already reacted. Shares fell more than 11 per cent in premarket trading on Thursday, and the company has lost over US$300 billion in market value since last Friday.

Also Read: The Nvidia clampdown is a warning for Southeast Asia’s AI boom

None of the espionage allegations has been tested in court, and neither Huang nor Nvidia has been charged with anything. Still, the scandal’s centre of gravity sits more than 13,000 km from Santa Clara, in Singapore.

Singapore breaks its silence, and points at Washington

Alice Huang was a top executive at Megaspeed International, a Singapore-based cloud firm that grew into Nvidia’s largest buyer in Southeast Asia. CBS News reports that Megaspeed and its Malaysian affiliate, Speedmatrix, bought about US$4.8 billion of Nvidia equipment over three years.

On Thursday, Singapore’s Ministry of Home Affairs (MHA) gave its fullest account yet. Police had investigated Alice Huang, then Megaspeed’s CEO, chair and one of its key controllers, over possible offences. These included buying servers that were exported via Singapore to Malaysia without ever entering the local market. Police gave her a stern warning for knowingly providing inaccurate information to downplay her role, and she has been barred from re-entering Singapore since February. Investigations into Megaspeed and six other individuals continue.

The ministry was careful on two points. First, the spying allegation is not part of its probe. Two weeks ago, MHA denied investigating her over ties to Chinese intelligence or to Jensen Huang. Second, Singapore does not enforce other countries’ unilateral export controls, although it expects companies based there to take such rules into account.

Then came the sharper line. Singapore had sought information from Supermicro, the San Jose server maker and close Nvidia partner, and received only limited cooperation. The ministry says it raised the matter with US authorities in July 2025 and has since repeated the request to the Department of Justice and the FBI, with little to show for it. The message was clear: Singapore wants it on record that it has not been the weak link.

The Bangkok thread and a widening dragnet

Supermicro knows this terrain. In March, US prosecutors charged its co-founder Yih-Shyan “Wally” Liaw and two others with routing roughly US$2.5 billion of servers packed with Nvidia H200 and B200 GPUs to China through a Southeast Asian front company. Bloomberg later identified that company as OBON, a Bangkok-based AI infrastructure firm that has not been charged. Liaw has pleaded not guilty, and Supermicro says it was a victim of the scheme.

According to Bloomberg, the Department of Justice is now running several probes involving Nvidia, including ones focused on Megaspeed and Supermicro. Separately, Manhattan prosecutors are examining whether entities linked to Alice Huang helped move Supermicro hardware into China. Megaspeed denies wrongdoing and says inspections by US, Malaysian and Nvidia teams found no violations.

Also Read: Southeast Asia in the 2026-2030 world order: Trade, chips, AI, and capital

For a region that has spent two years pitching itself as the neutral ground of the AI race, the map is awkward. The story features Singapore as the billing address, Johor as the data hall, and Bangkok, the capital of a country chasing a US$80 billion semiconductor push, as the base of an alleged front. That puts three ASEAN nodes in one scandal, and Washington is paying attention.

Why a Santa Clara boardroom matters in Johor

Huang was more than a CEO. He was the industry’s loudest critic of export controls, which he once called “a failure”. He leaves just as Democrats prepare to take control of Congress in January. Senator Elizabeth Warren is likely to chair the Senate Banking Committee, with subpoena power, and House lawmakers have already requested a closed-door session with Huang on 18 November.

The likely result is tighter rules, and Southeast Asia sits in the blast radius. Malaysia has already introduced permits for Nvidia chip exports under US pressure, even as its exporters sell more chips without winning more of the market. Expect more end-user vetting, allocation caps and on-site audits as Washington targets the region’s chip-smuggling problem. Honest data-centre operators and GPU cloud startups will pay much of the compliance bill, and the costs will ripple across Southeast Asia’s chip supply chain.

Singapore’s own Nvidia bets, such as Singtel’s partnership with the SIT-NVIDIA centre, will not unwind over this. But the premium on trust has just gone up.

Rivals smell an opening, but CUDA is still a moat

AMD, Broadcom’s custom-silicon business and Google’s TPUs have all chipped away at Nvidia’s dominance. In China, Huawei’s Ascend chips are Beijing’s preferred substitute. A leadership vacuum invites all of them to push harder with hyperscalers that are already hedging. Yet Nvidia’s real lock-in is CUDA, its software ecosystem, and no interim CEO can lose that in a quarter. The more immediate test comes on 17 November, when Kress presents third-quarter results without the showman beside her.

Kress is familiar with that stage. She joined Nvidia in 2013 from Cisco, after stints at Microsoft and Texas Instruments, and has fronted earnings calls for more than a decade. What she lacks is founder authority, the kind that let Huang sketch the future from a stadium stage and have markets believe him.

A governance lesson Southeast Asian founders should not ignore

The quieter headline is that the world’s most consequential chipmaker appointed a board chair only after its founder left in scandal. An outside investigation, led by former acting SDNY US Attorney Joon Kim and the law firm Wachtell, Lipton, Rosen & Katz, will now examine what the company knew and when.

Also Read: Chips, corruption, and credibility: Malaysia’s semiconductor gamble faces a trust test

Many of the region’s most celebrated tech companies remain founder-led, often with the founder also holding the chair. Their boards might ask a blunt question: if their founder became the story, who would have the standing to act, and how quickly?

Nvidia will survive Jensen Huang’s exit. Whether Southeast Asia keeps its standing as the trusted middle ground of the AI race, as the investigations he leaves behind unfold, is a far more open question.

The post Jensen Huang quits Nvidia, and Southeast Asia’s chip hubs inherit the fallout appeared first on e27.

Posted on — Leave a comment

The ghost in the middle: How one manager upskilled himself out of the room

The grey ring of condensation on the laminate table had almost dried by the time he finished explaining his pipeline.

Outside the glass wall of the coffee shop at the edge of the financial district, the midday rush was beginning to churn: young associates balancing takeaway salads, delivery riders cutting across intersections, the persistent equatorial heat beating down on the pavement. Inside, the quiet was heavy.

Mark sat with his hands resting flat against the table. They were the hands of a man in his early fifties who had spent nearly three decades in enterprise operations, hands that had once handled thick manila folders, marked up blueprints with red ink, and signed off on customs manifests across five regional shipping hubs.

“I thought I was building an ironclad moat,” he told me, his voice carrying the dry, matter-of-fact register of someone reading a post-mortem report on his own career. “I spent two years listening to every webinar, every tech podcast, every executive mandate telling us that AI wouldn’t replace managers, but managers who knew how to use AI would replace those who didn’t. So I took the bait. I learned the toolchains. I built autonomous agents. I turned myself into the sharpest conduit in the firm.”

He paused, glancing at his quiet phone.

“Last Tuesday, my desk was cleared. Not because I was lagging behind the technology, but because I succeeded in building a machine that didn’t need me to turn the crank.”

Over 25 years of reporting on labour transformations, technological upheavals, and the quiet casualties of modern industry, I have watched this pattern repeat itself under different names. In the late 1990s, when digital typesetting swept through legacy newsrooms, veteran compositors proudly taught software how to space lead type, convinced their mastery of kerning and layout logic would preserve their seniority. Within three years, the keyboards were gone, and a single desktop publisher did the work of 40 men. In the mid-2000s, when enterprise resource planning software swept corporate back-offices, logistics coordinators meticulously mapped their spreadsheets into databases, unwittingly writing the exact rulebook that rendered their coordination roles obsolete.

Every major wave of technological disruption dangles the same narcotic promise before the ambitious middle worker: Become the multiplier.

It sounds like empowerment. It flatters the ego. It suggests that by strapping an algorithmic jetpack to your back, you will become indispensable to an executive floor that has spent its entire history looking for ways to trim overhead.

Mark was the poster child of that promise. As regional operations lead for a Singapore-based supply chain company, his entire job used to be a weekly crucible of human friction. Every Monday, country managers from Jakarta, Manila, Bangkok, and Port Klang sent through 30-page performance updates: messy, defensive, uneven dossiers filled with contradictory data and unresolved interpersonal squabbles between warehouse teams and local couriers.

Mark’s entire value had once resided in his ability to read between those lines. He knew who was bluffing about vehicle maintenance, who was hiding margin decay, and whose family crisis was bleeding into depot turnaround times. He spent his Sundays in messy, uncomfortable back-and-forth calls, asking awkward questions, steadying anxious nerves, and hammering out a realistic picture of the week ahead.

Also Read: Southeast Asia is counting what it automated, it is not counting what came back

Then came the mandate to optimise.

Mark threw himself into building an agentic workflow. He strung together document-parsing scripts, language models, and automated reporting hooks. The incoming PDFs were ingested in seconds; anomaly detection routines flagged inventory variances; auto-prompted sub-agents drafted clean, bulleted executive briefs categorised by risk severity. The cycle time collapsed from 14 exhausting hours to under four minutes.

His European managing director was ecstatic. Mark was invited to present his automated setup at an all-hands leadership offsite. He was hailed as the model operator of the AI era.

“I felt invincible,” Mark told me. “My calendar opened up. I had zero unread emails. My reports were pristine, free of typos, mathematically airtight. Every Monday at 8:00 AM sharp, the C-suite received a synthesis that looked like it had been compiled by a team of McKinsey consultants.”

What Mark did not see was that in engineering out the friction, he had quietly engineered out his own soul.

The fundamental problem facing knowledge workers and managers today is not that artificial intelligence is becoming sentient; it is that human beings are actively stripping away their humanity to mimic the speed and detachment of software.

When you spend your days orchestrating automated outputs, you inevitably anchor your worth to process and delivery. But process is precisely what code is written to swallow. By converting messy human operational dilemmas into neat, deterministic summaries, Mark wasn’t displaying rare intellect; he was merely standardising an operational pipeline.

He had turned his role into an API. And once a role becomes an API, leadership does not need a well-compensated veteran director sitting in the loop to press “Run.” A junior coordinator at a fraction of the cost, or an automated cloud trigger, can execute the identical prompt.

The cracks appeared long before his dismissal, though Mark had been too intoxicated by his dashboards to notice.

“Tell me about the last real dispute you handled,” I asked him.

He looked down at the table, his brow furrowing. “A depot head in Johor Bahru. Our throughput scores began slipping in the third quarter of last year. In the past, I would have driven up across the causeway, taken him out for bak kut teh, and figured out what had broken in the depot culture. But the agentic pipeline had already diagnosed the issue: shift delays, courier routing bottlenecks, mechanical downtime. The system auto-drafted a performance remediation notice based on regional HR protocols. I reviewed the bullet points, thought the logic was flawless, and authorised the send.”

He let out a short, bitter breath.

“He resigned 72 hours later. Took his three best shift supervisors with him. The entire depot stalled for almost a month. What the system didn’t know (what I never found out because I never picked up the phone) was that the man’s wife had been admitted to intensive care, and his team had been quietly covering his night shifts so he could sleep by her hospital bed. The automated notice landed like a slap in the face. It was completely logical, completely defensible, and utterly destructive.”

The algorithm had extracted every metric, but it was stone-deaf to the subtext of human loyalty.

Also Read: The AI told me to raise my prices by 40 percent but it had never been to a wet market

When organisations rely entirely on frictionless automation, they mistake the volume and polish of output for actual health. They produce immaculate spreadsheets while the social fabric of the company rots from within. Because an AI model operates strictly on consensus (on the average of historical patterns), it can only ever produce the standard, predictable playbook. It cannot exercise taste. It cannot sense fear in a pregnant pause during a video call. It cannot take an irrational, high-stakes bet on a struggling colleague based on an unquantifiable gut feeling.

The turning point for Mark came six weeks ago, inside the very boardroom where he had once demonstrated his agent workflows.

The managing director sat across from him with the quarterly financial projections. The tone was polite, professional, and entirely devoid of malice. The company was facing regional margin pressure. They were streamlining the operational layer. The automated pipeline Mark had built was running smoothly; it was now maintained by an entry-level technical specialist who cost one-third of Mark’s salary.

“They didn’t push me out because the machine failed,” Mark said. “They pushed me out because I had spent 18 months proving that my role was just an automated sequence of prompts and approvals. I had made myself redundant by acting like the very tool meant to serve me.”

That is the quiet crisis looming over the modern knowledge economy. If your day-to-day value can be reduced to pulling data, formatting updates, synthesising memos, and routing tasks, you are not upskilling. You are documenting your own retirement.

The real insight of the automation age is counterintuitive: the competitive moat has nothing to do with execution speed. Execution is rapidly approaching zero cost. When any intern with access to an agent stack can draft a 40-page strategy memo or generate a financial analysis in minutes, the value of the document itself plummets.

What becomes infinitely scarce, and therefore infinitely valuable, is the messy human edge that no script can synthesise:

The moral courage to reject the statistically probable scenario because your intuition tells you the ground has shifted.
The relational presence to endure an awkward, unscripted silence on a phone call until an anxious team member finally tells you the truth.
The accountability to stand before executive leadership, point to an unhedged decision, and say: “The dashboard says otherwise, but I own this outcome if it fails.”

An AI model can offer probabilities; only a human being can assume risk. An AI model can summarise speech; only a human being can listen with empathy.

Before we left the café, I asked Mark what advice he would give to the thousands of professionals currently scrambling to add autonomous agents and prompt workflows to their daily routines.

He adjusted his glasses and looked toward the sunlit street outside.

“Use the tools to sweep the floor,” he said, his voice quiet but steady. “Let them clean the data. Let them format the slides. Let them process the administrative garbage that eats your daylight. But never mistake that for your purpose.”

His parting words were a challenge to anyone currently building an automated workflow:

“Keep your hands firmly on the steering wheel. Refuse the shortcut when trust is at stake. When a decision is hard, step away from the terminal and walk into the friction. Because if you spend your career trying to out-calculate the computer, the computer will eventually take your chair. Your human edge isn’t the machine you operate. Your human edge is you.”

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post The ghost in the middle: How one manager upskilled himself out of the room appeared first on e27.

Posted on — Leave a comment

Fear is high and longs got flushed. Is US$70,000 the next line in the sand?

My point of view is straightforward. The recent Bitcoin drop looks like a macro-led repricing with a leverage amplifier, not a Bitcoin-only failure. Bitcoin fell 3.45 per cent to US$71,425.63 in 24 hours and traded near a reported 12-week low. The broader crypto market fell 2.66 per cent to US$2.39 trillion. Bitcoin underperformed the wider market and became the main drag. 

The primary force came from higher US yields. A market recap on November 12 said hotter inflation pushed the US 10-year yield above five per cent as traders priced in more Fed hikes. That recap put Bitcoin down 4.3 per cent near US$71,400. The TLDR snapshot put Bitcoin down 3.45 per cent to US$71,425.63. The 2 snapshots differ slightly, and they point in the same direction. Higher yields can make speculative assets less attractive.

Bitcoin’s outsized decline pulled the wider crypto market lower. Markets reacted to Nvidia CEO Jensen Huang’s reported departure and uncertainty about potential selling tied to Strategy’s planned MSCI index removal. Those items remain reported market concerns, not proof of a single cause. 

The secondary force came from liquidations. Bitcoin long liquidations reached US$252.84 million in 24 hours, up sharply from the prior period. Forced closures can add sell orders and deepen a decline. Liquidation data reflects and amplifies selling pressure. It does not establish the original cause. Market sentiment remained fearful, and major crypto sectors were down.

That broad weakness supports the view that selling spread beyond Bitcoin. I treat the liquidation figure as evidence of leverage stress. The root cause looks more like higher yields and a market-wide move away from speculative assets. That distinction matters for the near term. A leverage flush can end once forced positions clear. A macro repricing can last longer if yields keep pressing speculative assets.

Also Read: Is short covering the only reason Bitcoin moved higher?

The 97 per cent correlation with Gold strengthens my macro-first reading. A strong correlation between crypto and Gold indicates a shared macro repricing. I do not take that correlation as proof of causation, and I would not overstate it. It weakens the case for a Bitcoin-only narrative. If Bitcoin and Gold move together under yield pressure, the more reasonable explanation is a common macro driver, not a crypto-specific event.

That is why I see the Nvidia and Strategy headlines as reported concerns that added uncertainty. They did not serve as confirmed triggers. The market recap put Bitcoin down 4.3 per cent near US$71,400, and the broader crypto market fell 2.66 per cent, so Bitcoin’s drop was a larger force in the move. The market-wide weakness also tells me that traders reduced risk across crypto.

Near-term momentum looks weak. Bitcoin sits below the US$73,936 daily pivot, and the reported 24-hour low was around US$71,200. If Bitcoin holds near US$71,200, it could attempt a rebound toward the pivot. A break below US$70,000 risks extending losses. The total crypto market cap sits at US$2.39 trillion, which is also the stated pivot level.

If the market cap holds near US$2.39 trillion and Bitcoin stabilises, losses could ease. Continued yield pressure or a break below that level would keep downside risk elevated. I would watch whether Bitcoin selling and long liquidations cool. I would not treat oversold readings alone as proof of a bottom. Stabilisation needs demand and calmer leverage.

Also Read: CPI at 0.4% broke Bitcoin’s US$73,000 floor. What happens if the Fed goes bigger?

The next daily US spot Bitcoin ETF flow report is a useful demand check. If flows improve and the market holds US$2.39 trillion, selling pressure could fade. If flows stay weak alongside continued yield pressure, the outlook weakens. I also mark Strategy’s planned MSCI index removal on November 30, 2026 as a future watch item. The available material does not confirm a Bitcoin sale tied to that change. The potential selling impact remains uncertain.

Any confirmed flows around the November 30 index change would matter more than headlines. The Nvidia CEO departure concern and the Strategy index concern both sit in the reported-concern category. They do not sit in the confirmed-catalyst category. I would keep that distinction clear when judging the next move.

My conclusion is that bearish pressure persists. Market-wide weakness and heavy long liquidations help explain the drop. The available evidence does not establish 1 definitive catalyst. The key watch over the next 24 to 48 hours is whether Bitcoin can hold above US$71,200 and recover the US$73,936 pivot. 

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Fear is high and longs got flushed. Is US$70,000 the next line in the sand? appeared first on e27.

Posted on — Leave a comment

The AI draft is the easy part. The source check is the job

In corporate communications, the first useful thing an AI assistant does is not write a finished announcement. It gives you a place to begin. It can turn a long technical brief into an organised list of claims, flag missing stakeholders, or reconstruct a scattered email chain into a timeline that people can check.

That speed is useful. But a clean summary can still be wrong in the way that matters most: it can make an exploratory conversation sound like an approved contract, an old quote sound current, or a possible roadmap sound like a commitment. The prose may be fluent while the status is false.

I treat generative models as drafting and triage tools, never as the source of record. Before a sentence leaves my desk, I want clear answers to three questions: What exactly are we claiming? What source supports it? What is its legal and commercial status?

Start with structure, not a paragraph

A communications brief is rarely a clean set of approved facts. It may combine verified information, market commentary, open decisions, and aspirational language. If you ask a model to write an announcement immediately, it may smooth those categories into one convincing paragraph. Uncertainty disappears from the prose even when it has not disappeared from the underlying evidence.

I get a better first pass by asking the model to act as a parser. It extracts each assertion into a table with the claim, its supporting document, the document’s author or owner, and whether written clearance exists. That changes the task from “make this sound exciting” to “show me what we can prove.”

The model can help identify missing sources, inconsistent dates, and repetitive language. A person still has to decide whether a claim is accurate, approved, and appropriate for public release. A polished draft can create pressure to send; a visible list of open checks makes it easier to pause.

Also Read: Open code, unowned risk: The governance gap in business-critical email

Audit the verbs of commitment

In corporate communications, verbs such as “launched,” “partnered,” “selected,” “acquired,” and “secured” carry weight. They describe different states of affairs. A model that treats them as stylistic variations can turn an exploratory discussion into an announcement of a completed relationship.

An AI-assisted draft can draw on genuine historical correspondence and recast it as a current partnership. The source material may exist; the implication may not. The fix is not another prompt. It is to return to the primary correspondence, check the relevant agreement and approvals, and remove the unsupported assertion.

That discipline applies to Votee AI’s own communications too. The HKMA’s published list names Votee AI among the technology partners selected for the first GenA.I. Sandbox++ cohort; the SFC’s published annex records the same cohort. Similarly, our CTO is listed as a contributor to OWASP’s Agentic Skills Top 10 (AST10), a framework documenting security risks in AI agent skills.  “Selected” and “contributed” are different claims, and each should match exactly what the public record shows.

The governance question for agents

The industry is moving from scripted chatbots towards systems that can plan and use tools across multiple steps. That can make software more capable, but it also makes permissions, oversight, and accountability harder to reason about.

An agent with access to company data should not be treated as safe simply because its output sounds plausible. Systems need limits on what an agent can access and do, clear records of the actions it takes, and a way for people to review or stop consequential steps. A practical “maker-checker” approach can help: one component gathers information or prepares a draft, while a separate control process checks the action against defined rules and routes higher-risk decisions to a person.

Also Read: AI is a tool. Stop treating it like a strategy

Review should scale with risk. An internal recap email does not need the same scrutiny as a public statement, a regulatory filing, or an earnings disclosure. But each process should make clear which facts are verified, who approved them, and what remains uncertain.

Use saved time to check the record

The useful promise of AI in communications is not that it removes human judgement. It is that it can reduce the manual work of sorting, collating, and summarising information from different documents.

The important question is what we do with the time saved. I use it to trace claims to their sources, ask subject-matter colleagues what a document actually authorises, and check whether a draft reflects the decision that was made, not merely the decision someone hopes will be made.

A strong draft is not the one that sounds most certain. It is the one whose confidence matches the evidence behind it. In communications, the source check is not a final proofreading step. It is part of the work.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post The AI draft is the easy part. The source check is the job appeared first on e27.