Alex McIsaac here!

In Q1 we sourced and screened over 1,600 companies globally started by a Canadian founder and had 100+ unique conversations. We usually speak to founders very early, often before they’ve raised their first round, and sometimes before they’ve even built an alpha product. Our conversations stretched across B2B software, vertical AI, infrastructure, fintech, industrial SaaS, and hard tech. While each founder’s story is unique, some clear patterns have emerged that say a lot about where the market is headed and where early opportunity lives.

Here are the 10 trends I’m seeing at the earliest stages of company formation:

1. Vertical AI is becoming autonomous and workflow-native

AI copilots are quickly giving way to full-stack agentic systems. These aren’t just assistants suggesting content — they’re autonomous workers embedded directly in daily workflows. From litigation agents drafting multi-step legal documents, to agents navigating real estate CRM pipelines, founders are building tools that not only think, but do — and more interestingly, they operate where the user already is: Excel, Outlook, Slack, Salesforce.

We are finding the most exciting products feel like a new teammate, not a new tab.

2. Infrastructure is being rewritten for AI from the ground up

Founders are rethinking the building blocks of software to meet the demands of modern AI workloads. This isn’t about bolting LLMs onto existing tools — it’s about creating foundational systems optimized for streaming, orchestration, and real-time responsiveness. Legacy stacks — built for static data and human-triggered workflows — are showing their age, and technical founders are rebuilding them with first principles in mind. We’re seeing new primitives emerge that prioritize latency, modularity, and deep integration with ML pipelines. Many of these teams have roots at large infra-heavy companies, and they’re taking the lessons (and frustrations) from those environments to architect cleaner, more composable systems from scratch. The result is a quiet but powerful replatforming of core software infrastructure to support the next generation of intelligent applications.

3. Hard tech is back, and grounded in deployment

There’s a ‘not so quiet’ resurgence of full-stack hardware-software companies — many of them AI-native from day one. From vision-based drones that operate in GPS-denied warzones, to secure OS layers for industrial energy storage systems, these teams are shipping code and circuitry at the same time.

Unlike the moonshot era of ‘sci-fi decks,’ these startups are deeply operational. They’re building to deploy in constrained environments — battlefields, industrial plants, energy grids — not just lab demos.

4. Agents are moving from “chat” to execution

The chat UX is quickly becoming table stakes. Additionally, we are now seeing agents take more action and voice play a role as the new interface for many applications.

We spoke with multiple founders building systems that use cloud browsers, credential vaults, and API stitching to let agents take action autonomously — submitting forms, scheduling meetings, conducting research, or triggering approvals in enterprise workflows. These systems combine voice, text, and structured data to act across multiple systems in parallel, not sequentially.

5. Data is the new wedge

Across industries, founders are turning “data exhaust” into structured, monetizable assets. In real estate, home inspection reports are becoming quality-assurance datasets. In manufacturing, visual defect logs are fueling new QA models. In industrial procurement teams are leveraging AI to review and process 1000+ page documents and accelerate response timelines by weeks

We’re seeing a trend toward turning once-dead-end workflows into upstream data collection engines — often with second-order business models around re-sale, insight layers, and new applications.

6. Go-to-market is the new moat

Founders are putting as much creativity into distribution as they are into product. Rather than chasing crowded segments, the most effective teams are targeting underserved, often unsexy verticals — where customer pain is high, switching costs are low, and incumbents are slow-moving. What’s changed is the level of precision: go-to-market is no longer a secondary consideration, it’s a core part of the wedge. Founders are designing their initial GTM around trust, context, and embedded workflows — winning not by being louder, but by being earlier, sharper, and closer to the problem.

Wedge-market fit is beating product-market fit.

7. Canada is a proving ground for deep tech

Some of the most technically ambitious startups I spoke with this quarter are quietly being built and tested in Canada. Founders are using Canada’s regulatory frameworks, utility partnerships, and R&D incentives to validate hard tech in areas like defense, energy, manufacturing, and public infrastructure.

These aren’t academic projects; they’re applied bets on where industrial systems are going. Whether its AI-powered chip design or satellites, Canada is offering a uniquely supportive launchpad for capital-intensive, IP-heavy companies that need space to experiment before scaling globally.

8. Lean teams are shipping faster than ever

A new wave of solo founders and lean duos — often without deep technical backgrounds — are using tools like Cursor, Replit, and off-the-shelf models to build and ship revenue-ready products in weeks, not months.

These teams are skipping the traditional pre-seed product motions and going straight to pilots, early revenue, and with faster product iterations. With access to AI-native dev environments and composable APIs, execution velocity has become the clearest signal.

The gap between idea and ARR has never been smaller.

9. Fintech is reawakening in underserved markets

There’s renewed activity in underbanked fintech: from credit products for non-prime consumers in Canada, to embedded lending infrastructure for car purchases, to agent-led underwriting systems for SMB insurance. The angle isn’t neobank-style branding — it’s software-first CAC reduction, combined with domain-specific GTM.

Instead of building another Stripe or Plaid, these teams are applying AI to edge-case underwriting, workflow automation, and distribution bottlenecks. They might be unsexy — but could be big opportunities.

10. AI enterprise adoption is still early but warming up, quickly

Large buyers — especially in infra, productivity, and internal ops — are beginning to move from paid POCs to paid contracts. From Zoom to Microsoft to Ramp, teams are running paid pilots and joining design partnerships with startups that solve real friction. That said, buyers are cautious. Trust, data access, observability, and failure modes remain the gating issues. But the early signs are promising — and the founders who nail enterprise onboarding and reliability early are getting a head start.

—

Q1 2025 has made one thing clear: we’re past the novelty phase of AI. Founders are now building for outcomes, not demos. And the most interesting ideas aren’t about replacing humans — they’re about re-architecting workflows to make humans exponentially more effective.

If you’re building at the earliest stages and thinking deeply about vertical AI applications, infrastructure for the AI-native stack, and software applications for legacy industries please reach out!