‍

Figma IPO went 🍌 — obviously, product-led companies win bigger - and suddenly everyone’s saying, “SaaS is back. Long live software!” But the truth is, SaaS never died because of AI. Instead, it’s changing its skin through AI, it's finally getting smarter and delivering real automation.

One of the key drivers behind this blockbuster success is Figma’s compelling AI narrative and strategic positioning. By layering AI-powered features on top of its collaborative infrastructure, Figma has positioned itself as a core enabler for the future of workplace and design automation, with huge potential for continued AI-first growth.

Reflecting on the magic combo of AI plus software, I’ve developed a few thoughts about what this means for early-stage companies: both the new wave of native agentic startups and the emerging breed of SaaS challengers who are embracing AI.

AI is the new UI

At Connect, we are excited about investing in native, agentic-first startups. These companies solve real business problems in fundamentally new ways, defining a new emerging paradigm for software: interfaces are disappearing.

As agentic AI becomes ubiquitous in enterprise software, users no longer read and write, browse, or search — they simply will prompt an outcome, and the system delivers. Ewake.ai is a good example of it. Its production and observability agents show up where the work is happening and where the developers are. No dashboards. No extra tools. UI-less

AI x infra

We’re equally excited by how the pre-LLM most innovative product companies are embracing Gen AI to sharpen their edge and offer customers intelligence and real automation.

AI is the automation layer SaaS always promised — but never truly delivered.

AI may be the face, but software infra is the leverage—essential for scale, differentiation, and control, especially in mission-critical contexts. You often need a transaction layer, and a control system: compliance, permissioning and some form of human-in-the-loop for checks. And collaboration is key. Today, software powers human collaboration — tomorrow, it may coordinate AI agents too. Software infra is a natural canvas for AI—enabling feedback loops where models learn from the platform’s own data.

AI x infra cases: Plain and Aikido

We’re already seeing this AI x infra flywheel come to life and spin successfully across our portfolio. Plain and Aikido Security are replacing large incumbents with traditional systems of record or engagement through the combined effect of superior SaaS capabilities augmented by AI capabilities.

• Plain’s infrastructure layer unifies support channels with email and Slack at the core. It acts as the system of record and anchor for team workflows. Layered on top is a next-gen AI experience: smart triage, AI-generated replies, and autonomous support agents. Plain being the source of truth means that it has context on the entire app, which is critical to getting AI to function well on those individual components, and, eventually, do everything itself.
• Aikido is applying AI across its end-to-end security platform to boost automation, precision, and developer velocity. Its AI layer includes AutoTriage: AI-powered prioritisation of alerts; AI threat intelligence & cloud search; AI autofix: automated remediation for faster, safer code shipping.

Steep, Intropic and Purple Dot are the next AI x infra to watch.

Intelligent software systems

Faster and better than the incumbents, the “pre-LLM” software challengers are layering AI onto proven infrastructure. They’re now AI-powered. On the flip side, today’s native agentic players will also need to build and rely on infrastructure: governance, permissioning, collaboration, observability, and security.

The result is the same yummy cake: intelligent systems run atop solid software foundations, and users simply ask for what they want.

And the best product founders know the opportunity isn’t just to bolt on AI. It’s to rethink the product from first principles: what it does, how it learns, how it communicates, and how it governs.

‍

Wherever you put the needle on the AI x infra spectrum — whether you start AI-native or AI-powered — it’s a great time to build, and we’d love to meet you.

‍