The ICPCC Talks Our Team Keeps Rewatching

We’ve all been there. You watch a brilliant conference talk, the live chat goes wild, you bookmark it for later, and then the Slack notifications swallow it whole. Most conference content has a half-life of about 48 hours. But after the latest ICP Community Conference, something different happened. Our team kept returning to the recordings—not out of obligation, but because certain sessions genuinely changed how we build on the Internet Computer Protocol. We’d find ourselves quoting specific slides in sprint planning, or pausing a fireside chat to argue about composability. These aren’t just archived streams gathering digital dust; they’re the sessions that rewired our approach to DAO governance, scalability, and on-chain identity. Here are the talks we haven’t stopped rewatching, and why they matter for anyone building in the UK ecosystem right now.

The DAO Governance Deep Dive That Shifted Our Thinking

We thought we understood DAO governance until we watched the deep dive on the Service Nervous System (SNS). This wasn’t another theoretical walkthrough of voting mechanisms. The team from the DFINITY Foundation broke down liquid democracy with such clarity that even our front-end developers, who normally avoid governance debates, leaned in. They walked through how neurons track voting history, how you can delegate specific decision types while keeping veto power on others, and what happens when a proposal hits a contentious threshold. The live demo of a DAO tooling dashboard—showing real-time proposal lifecycles for an active SNS—made the whole thing feel tangible, not aspirational.

Why on-chain voting finally feels practical

The friction with on-chain voting has always been the cognitive load. Nobody wants to evaluate thirty technical proposals a week. This session showed how followees can be tagged by expertise category, so you might delegate treasury management votes to a finance-focused neuron holder while keeping product roadmap votes under your direct control. They demoed a UI that surfaces your delegate’s voting record and flags when they deviate from their stated principles. That accountability layer is what we’ve been missing. It transforms delegation from blind trust into an informed, revocable choice. One of our engineers immediately started sketching a similar dashboard for our own SNS setup.

The one slide on neuron staking we screenshotted

There was a single slide—slide 17, we checked—that mapped out dissolve delay bonuses against voting power curves in a way that finally made the incentive structure click. It visualised how an eight-year dissolve delay compares to a six-month one, not just in raw multiplier terms but in effective influence over proposal outcomes. That screenshot lives in our team’s Notion now. Whenever someone questions why we recommend longer staking periods for core contributors, we drop the image. It’s become shorthand for a whole conversation about long-term alignment in DAOs.

Internet Computer Protocol Infrastructure: The Scaling Keynote

The technical keynote on subnet architecture was dense, but in the best way. The speakers tackled the questions European node operators have been raising for months—particularly around latency and geographic distribution. For UK-based developers and node providers, the discussion on boundary node architecture felt overdue. They addressed how requests route through boundary nodes before hitting subnet replicas, and what that means for users in London versus Singapore. The transparency about current limitations was refreshing; they didn’t pretend every metric was perfect, and that honesty made the roadmap more credible.

Breaking down the boundary node improvements

Boundary nodes act as the entry point to the Internet Computer, handling TLS termination and routing requests to the correct subnet. The keynote detailed recent improvements to the boundary node layer that reduce latency by caching routing tables more aggressively and pre-establishing connections to frequently accessed canisters. For a user in Manchester hitting a dApp deployed on a European subnet, these changes shave off milliseconds that compound across API calls. The team showed benchmarks comparing response times before and after the update, and the difference was significant enough that we immediately checked whether our own canisters were benefiting.

What the new scalability figures mean for UK developers

The keynote included updated throughput numbers that sparked a flurry of messages in our team channel. Subnets can now handle more canisters and higher transaction volumes than the figures quoted at previous conferences. For UK startups building consumer-facing applications, this changes the calculus. You’re no longer worrying about whether your canister will hit a ceiling if your user base doubles. The presenters mapped out a trajectory that suggests subnets will scale further without requiring developers to restructure their code. That stability promise matters when you’re pitching on-chain architecture to a board that remembers the gas fee spikes of other networks.

Real-World Web3: The London Use-Case Panel

This panel was the standout for sheer practicality. Three London-based on-chain startup founders sat on stage and talked honestly about what worked, what broke, and what cost less than they expected. No hype, no token talk—just operational reality. One founder described migrating their entire backend from AWS to canister smart contracts over a single weekend, then spending the following week fixing assumptions they’d carried over from Web2 architecture. The room was packed, and the Q&A ran over by twenty minutes because the audience kept asking specific questions about gas costs, uptime guarantees, and user onboarding flows.

From traditional cloud to 100% on-chain in Soho

One founder, running a fintech compliance tool out of a Soho coworking space, detailed their migration path. They’d been spending roughly £4,200 monthly on cloud infrastructure for a product that processed regulatory checks. After moving fully on-chain, their infrastructure costs dropped to the cost of cycles—less than £300 per month. The trade-off was rethinking their database architecture entirely. Instead of a PostgreSQL instance, they used stable memory in a canister. Instead of a message queue, they used inter-canister calls. The founder admitted the mental shift was harder than the technical one, but the numbers spoke for themselves.

The regulatory clarity discussion everyone applauded

Midway through the panel, the conversation turned to the UK’s regulatory environment for blockchain-based businesses. One panellist, who had recently completed a sandbox programme with the FCA, explained how running fully on-chain actually simplified their compliance reporting. Because every transaction is auditable and the canister code is verifiable, they could demonstrate data integrity without building a separate audit trail. The audience reaction was telling—a round of applause broke out when the founder said regulators weren’t the obstacle; opaque off-chain systems were. That moment crystallised something we’d been feeling but hadn’t articulated.

The Decentralized Identity Talk We Didn’t Expect to Love

Identity sessions tend to be dry. This one wasn’t. The Internet Identity presentation opened with a live biometric authentication demo—FaceID on a volunteer’s phone, generating a session key that authenticated against a canister on stage. It worked first time, which anyone who’s done live demos knows is half miracle. But the real energy came from what followed: a discussion about applying this technology to public services, specifically the NHS. The room split into camps, and the Q&A became a proper debate rather than polite questions.

How passkeys beat passwords in a live stress test

The demo wasn’t just a happy path walkthrough. The presenter deliberately used a device they’d never authenticated with before, showing how passkeys eliminate the need to remember or store credentials. They registered a new anchor, signed a transaction, and then revoked the session—all in under thirty seconds. Compared to the average password reset flow, which studies show takes over three minutes and often leads to abandonment, the efficiency gain is staggering. For UK services trying to reduce drop-off during onboarding, this matters.

The audience debate on NHS and citizen data

Things got heated when an audience member asked whether Internet Identity could underpin NHS patient records. Privacy advocates in the room raised concerns about biometric data centralisation, while others pointed out that the architecture doesn’t store biometrics on-chain—the authentication happens locally, and only a derived public key touches the network. A GP from Birmingham stood up and described the current nightmare of cross-trust record sharing, and suddenly the conversation wasn’t theoretical anymore. We left that session convinced that decentralised identity for public services is closer than most people think, and that the UK might be the proving ground.

Our Favorite Fireside Chat: Building Open Internet Services

Sometimes the unscripted sessions deliver the most value. This fireside chat paired two core ICP engineers with three community developers, and the result was a candid, occasionally blunt conversation about why the industry keeps defaulting to centralised clouds. The DFINITY Foundation roadmap came up repeatedly, not as a sales pitch but as a reference point for what’s realistically achievable in the next twelve months. The tone was relaxed but the ideas were sharp—the kind of discussion that makes you rethink assumptions you didn’t know you held.

The moment the conversation turned to true composability

One engineer described building a lending protocol that called into a separate identity canister and a separate reputation canister, neither of which their team had written. On traditional infrastructure, this would require API keys, rate limiting negotiations, and trust assumptions. On the Internet Computer, it required knowing the canister IDs. The community developer who’d built the reputation system was sitting in the audience and confirmed they had no idea their canister was being used that way. That’s real composability—not just theoretical interoperability but actual, unplanned reuse. Our team has since adopted this as a design principle: build canisters that other developers might call without asking permission.

Why we keep quoting the ‘serverless is a myth’ remark

Halfway through, one of the DFINITY engineers said, “Serverless is a myth—there’s always a server, you just don’t own it.” The line landed hard. The point wasn’t that serverless functions are useless, but that they obscure the centralisation underneath. When you deploy to a cloud function, you’re running on Amazon’s or Google’s servers, subject to their pricing changes and usage policies. On the Internet Computer, the servers belong to a decentralised network of independent node providers. We’ve quoted that remark in three client meetings since, and it consistently reframes the conversation from cost comparison to sovereignty.

The Community Lightning Sessions That Deserve a Full Rewatch

The lightning session format—seven minutes per project, no exceptions—forced presenters to cut the fluff. What emerged was a showcase of raw experimentation from independent builders who’d been tinkering with canister smart contracts in ways the core protocol designers probably never anticipated. The Manchester and Bristol developer communities were particularly well-represented, reflecting the growing concentration of ICP talent outside London.

The micropayment dApp that surprised everyone

A developer from Manchester demonstrated a micropayment system for pay-per-article journalism, built entirely in a single canister. The twist: transaction amounts as low as £0.01 were economically viable because the Internet Computer’s reverse gas model means users don’t pay per transaction. The audience audibly reacted when the presenter showed a live payment of half a penny clearing instantly. Traditional payment rails would lose money on that transaction; here, it just worked. Six months later, that dApp has processed over a hundred thousand micropayments, and the developer is in talks with a regional newspaper group.

A zero-knowledge proof experiment born in a UK hackathon

A team from Bristol presented a zero-knowledge proof implementation they’d built during a weekend hackathon. It wasn’t production-ready—they were upfront about that—but the fact that they could run ZKP verification inside a canister at all was remarkable. They used the example of proving age without revealing a birthdate, a use case directly relevant to the UK’s evolving digital identity standards. The project has since received a grant and is evolving into a privacy layer that other canisters can call. Watching that lightning talk now feels like seeing an early draft of something significant.

FAQ

What is the ICP Community Conference?

The ICP Community Conference (ICPCC) is a global event focused on the Internet Computer Protocol, bringing together developers, founders, and researchers to share technical insights, project demos, and governance discussions. It features keynotes from the DFINITY Foundation alongside independent community talks and panels.

How does the Service Nervous System (SNS) work?

The Service Nervous System is an on-chain governance framework that allows DAOs to control canister smart contracts through proposal-based voting. Token holders stake neurons with configurable dissolve delays and can vote directly or delegate voting power to trusted followees, creating a liquid democracy system.

Can UK startups realistically build fully on-chain?

Yes, and several London-based founders demonstrated this during the conference. The Internet Computer’s canister architecture can replace traditional cloud infrastructure for many use cases, with significant cost reductions reported. The main challenge is architectural rethinking rather than technical limitation.

Is Internet Identity safe for sensitive data like NHS records?

Internet Identity uses passkeys and biometric authentication that processes credentials locally on a user’s device. No biometric data is stored on-chain—only a derived public key. The architecture is designed to be privacy-preserving, though implementation for public services would require additional governance and compliance layers.

Where can I watch these ICPCC talks?

All sessions from the ICP Community Conference are available on the official ICP YouTube channel and through the conference archive. We recommend starting with the governance deep dive and the lightning sessions if you want to see the breadth of what’s being built on the Internet Computer Protocol.