Micepad x Google Cloud: Reduces Infrastructure Costs by 60%

For most people, a conference is a physical thing: a hall, a badge, a row of seats, a coffee break where the real networking happens. Micepad's founders set out to prove that a virtual event could deliver all of that, and in some ways more, without anyone needing to board a plane.

Micepad began in 2013 as an Event Check-in App for live conferences, helping organisers replace printed programs and paper handouts with material delivered to attendees' tablets. That early product gave the company something most "virtual event" startups never had: years of hands-on understanding of what actually makes a conference work, from the rhythm of a keynote to the awkward mechanics of meeting a stranger in a crowded room.

When the world shifted online in 2020, Micepad was ready to translate that understanding into a digital-first platform. Rather than chasing the trend of 3D avatars and simulated exhibition booths, the team made a deliberate bet on human interaction as the core of the experience. The platform's design prioritises one-to-one connection - the conversations, introductions, and follow-ups that turn an event from a broadcast into a community.

Today, Micepad is an enterprise event platform serving clients that range from small businesses to global organisations such as Visa and Enterprise Singapore. Headquartered in Singapore with a presence across Hong Kong, Taiwan, and Malaysia, the company has been recognised with the "Most Innovative Startup Award" and a World Summit Award in the business and commerce category. And throughout its evolution, one decision has quietly underpinned its ability to grow without burning capital: its choice of cloud infrastructure.

Running a virtual conference is deceptively demanding. A single event might host dozens of simultaneous sessions, hundreds of concurrent chat threads, live polling, Q&A, photo sharing, and on-demand replays, all of which must remain glitch-free for attendees scattered across continents and time zones.

The hardest part isn't the peak. It's the shape of the demand. A conference's traffic is spiky and unpredictable: a flagship event might draw 5,000 people for a few hours, then taper to a few hundred stragglers replaying sessions overnight. The next week, the platform might be running a 20-person executive briefing. Provisioning enough capacity to guarantee a flawless experience at peak means paying for that capacity around the clock - even when 95% of it sits idle.

For a venture-backed company, that inefficiency is an annoyance. For a self-funded startup, it's existential. Micepad needed an infrastructure model that could expand instantly to meet the demands of its largest events and contract just as quickly when the room emptied - and it needed to pay only for what it actually used.

It also needed to keep the experience seamless. In a physical conference, latency is a non-issue; everyone is in the same room. In a global virtual event, a half-second of lag in a live chat or a stuttering video stream is the difference between a vibrant session and a dead one. Micepad had to solve for scale and responsiveness simultaneously, on a startup budget.

Micepad's relationship with Google Cloud predates its pivot to virtual events. The company migrated from a rival cloud provider in 2018 after joining the Google Cloud for Startups Program and discovering the Google Developer Group (GDG), a community of engineers exchanging practical knowledge about cloud architecture.

That community proved decisive. After five years of running its platform exclusively on virtual machines, Micepad's engineers learned through GDG sessions about the autoscaling capabilities of Google Kubernetes Engine (GKE). The appeal was immediate: instead of provisioning fixed blocks of compute and hoping demand matched supply, GKE could scale the platform's resources up and down automatically in response to real load.

The migration paid off in the most direct way possible - the company's own ledger.

The numbers tell the story. Micepad's annual server spend fell from roughly $100,000 to about $40,000 - a 60% reduction - even as the platform's ambitions grew. That saving wasn't a one-time discount; it was structural. Because GKE charges for what the platform actually consumes rather than what it might theoretically need, the cost curve now follows the demand curve. Quiet weeks cost little. Busy weeks cost more, but only because they're generating value.

Today, Micepad runs roughly 90% of its platform on GKE, a dramatic shift from its all-VM origins. For a service that routinely flexes between 20 and 5,000 participants, that elasticity isn't a nice-to-have - it's the entire economic model.

Cost efficiency means nothing if the experience suffers, and this is where Cloud CDN became the second pillar of Micepad's architecture. Working alongside GKE, Cloud CDN delivers content through Google Cloud's global network of data centers, keeping latency near zero regardless of where an attendee happens to be sitting.

The payoff is concrete. Micepad keeps dozens of chats, seminars, and presentations - along with live polling, Q&A, and photo sharing - running smoothly across time zones. Attendees can view and replay sessions on demand for as long as the organizer keeps the event open, and they can schedule private meetings with other participants at a time that works for both parties, even if they're on opposite sides of the planet.

The proof came during a global medical conference that drew more than 3,000 attendees from regions including Pakistan, Nigeria, and Saudi Arabia. Over three days, the platform hosted dozens of lectures, seminars, and networking sessions without a single disruption. In a physical equivalent, those doctors would have needed to fly in from every corner of the world - a logistical and financial barrier that would have excluded many of them entirely.

That event was a clean demonstration of the two headline capabilities working in concert: GKE absorbing the demand of thousands of simultaneous users, and Cloud CDN ensuring that every one of them experienced the conference as if it were happening next door.

A small engineering team can't afford to spend its hours on repetitive operational work. Before adopting Google Cloud's CI/CD tooling, deploying and testing new features consumed a meaningful slice of every developer's day.

Cloud Build changed that equation by automating the testing and deployment pipeline. The result is roughly five hours saved per developer, per week - time redirected from the mechanical work of shipping code toward the creative work of building new interactive features. For a company competing on the richness of its event experience rather than the size of its marketing budget, that reallocation of human attention is a competitive advantage in its own right.

It also reflects a broader theme in Micepad's relationship with Google Cloud: the partnership has consistently been about doing more with less. The Startups Program provided credits that let the team focus on product rather than bills. Later, Google Cloud engineers worked directly with Micepad to refine system designs and find further cost efficiencies.

Micepad's infrastructure choices also align with a value baked into the company since its earliest days. The original 2013 content management app was, at heart, a sustainability product: it let organisers eliminate paper waste by pooling materials digitally. The virtual event platform extends that mission by reducing the carbon footprint associated with global conference travel.

Running on Google Cloud reinforces rather than undermines that goal. Google Cloud's commitment to renewable energy across its global operations gives Micepad confidence that the infrastructure beneath its sustainability-minded platform isn't quietly creating environmental harm of its own. For a company whose value proposition includes reducing the need to fly, the consistency matters.

With its compute foundation proven and its costs under control, Micepad's attention has turned to a question that GKE and Cloud CDN can't answer on their own: how do you make a virtual event genuinely smarter than a physical one?

The seed of that ambition has been present from the start. Micepad's AI matchmaker already connects attendees with shared interests more effectively than working a room, grouping participants by industry and specialty and surfacing relevant introductions. Its interface lets a participant carry on up to five chats while listening to a speaker - real-time networking layered directly onto conference content. These features run on a proprietary matchmaking algorithm that the company has steadily refined.

The next step is to deepen that intelligence using Google Cloud's data and AI stack - and this is where BigQuery, Vertex AI, and Gemini come in.

BigQuery: turning event activity into insight

Every Micepad event generates a rich stream of signals: which sessions drew the most attention, how engagement rose and fell across a multi-day agenda, which connections led to scheduled meetings, and where attendees dropped off. Consolidating that activity in BigQuery turns raw event logs into a queryable picture of how a conference actually behaved.

For organizers, that's the difference between guessing and knowing. Instead of relying on a post-event survey, they can see engagement patterns directly - which topics resonated, which time slots underperformed, which networking features drove the most meaningful interactions. BigQuery's ability to handle large volumes of data quickly makes this practical even for the platform's biggest events, where a single conference can produce millions of interaction records.

Vertex AI: smarter matchmaking and personalization

Micepad's matchmaking has always been the platform's signature feature, and Vertex AI provides the foundation to evolve it from a fixed algorithm into a continuously improving system. By training and serving recommendation models on Vertex AI, Micepad can move toward matchmaking that learns from actual outcomes - which introductions led to conversations, which led to scheduled meetings - and personalizes suggestions for each individual attendee rather than relying on broad category matching alone.

This is the natural maturation of the personalization capabilities Micepad has long aimed to deliver: a system that understands not just what an attendee said they were interested in, but how people like them actually behave at events.

Gemini: AI-assisted engagement

Generative AI opens a set of possibilities that simply weren't available when Micepad first built its platform. Using Gemini, the company is exploring features such as conference chatbots that can answer attendee questions about the agenda in natural language, automatic session summaries that let participants catch up on a talk they missed, and content assistance for organisers preparing event material.

Where Micepad once planned to build conference chatbots on earlier conversational tooling, Gemini offers a substantially more capable foundation - one that can understand context, summarise long sessions, and respond to attendees in fluent, conversational language. Layered on top of the BigQuery data foundation, these features point toward an event experience that actively helps each attendee get more out of their time.

Keeping the layers honest

It's worth being precise about how these pieces fit together. The 60% cost reduction is a property of the compute migration - GKE replacing always-on VMs - and it stands entirely on its own. The AI layer is a separate investment aimed at increasing the value of each event, not reducing its cost; in fact, AI workloads add to the compute bill. The two stories are complementary, but they shouldn't be conflated: GKE makes the platform affordable to run, while BigQuery, Vertex AI, and Gemini make each event more intelligent. Together they describe a platform that is both lean and increasingly smart.

Micepad x Google Cloud stack now spans the full lifecycle of a virtual event:

The compute and delivery layers are the proven, in-production foundation that established the platform's economics. The data and AI layers represent where Micepad is heading - building on a base that has already demonstrated it can scale globally without scaling costs out of control.

Micepad's story is, on one level, a straightforward infrastructure win: a startup cut its server bill by 60% by adopting autoscaling, and gave its engineers hours back each week through automation. Those are real, measurable outcomes that any growing technology company can appreciate.

But the deeper lesson is about what that efficiency enables. Because Micepad isn't spending $100,000 a year on idle servers, it can spend that capital on building features that make events better. Because its developers aren't spending five hours a week deploying code by hand, they can spend that time inventing new ways for people to connect. And because the platform's economics are sound, the company can now invest confidently in an AI layer that will define its next phase of growth, without venture funding, and without betting the business.

That's the throughline of Micepad's partnership with Google Cloud: infrastructure that gets out of the way so the team can focus on its actual mission. As James Gwee puts it, the company is excited about the future of enabling human connection and serendipity through virtual events - and it sees the room to keep building exactly that, wherever its attendees happen to be.

About Micepad: Micepad is an enterprise event app and virtual event platform enabling dynamic global conferences with AI-driven networking, multiple live chats, and personalised engagement in a seamless online experience. Founded in 2013, Micepad helps event organisers go digital, reduce waste, and create better experiences for attendees across Asia and beyond. Clients range from small businesses to global organisations, including Visa and Enterprise Singapore. Based in Singapore with a presence across the globe, Hong Kong and Taiwan.