For decades, the “moat” around large enterprises was built on two things: massive human capital and proprietary software infrastructure. If you were a small business, you “borrowed” your tech through SaaS subscriptions, paying for tools that fit your workflow. But according to Cameron Wall, technologist and founder of BlockScribe, we are entering an era of “organisational compression” where the advantage has flipped.

The traditional SaaS model is hitting a wall. Wall, who recently built a custom budgeting application for his wife in a matter of days using Claude, argues that the “build, borrow, or buy” framework is changing. “The ‘borrow’ is becoming less of a requirement,” Wall notes, referring to the era of renting generic software. Many of the AI tools are still only available via subscription, albeit at far lower cost than some Enterprise SaaS fees.

For a small business, the ability to install a Local AI Build, software that lives on your own hardware rather than a vendor’s cloud, could be a game changer for many small businesses. It eliminates many of the fees that have previously put off SMEs investing in the creation of their own tech. More importantly, it allows for hyper-customisation. In 2026, the competitive edge belongs to the founder or business owner who can “vibe code” a solution that fits their business 100%, rather than settling for a product that does 80%.

No More Nigel from IT

Large corporations are currently hamstrung by what Wall calls legacy thinking. They are managed by “Nigel,” a pseudonym for the IT director who has been there for 12 years and views AI as a threat to his established systems. While “Nigel” spends 18 months in security reviews, a nimble SME can use the Model Context Protocol (MCP) to snap together specialised AI agents like Lego blocks.

This “Microservices” approach allows small teams to deploy enterprise-grade infrastructure, from automated legal document review to autonomous supply chain management, at a fraction of the cost. The security risk, once a major deterrent for small firms, is also being solved by AI itself. Wall highlights that agents like Claude Sonnet 4.6 can now audit their own codebases, finding and fixing hundreds of vulnerabilities in real-time, effectively providing a “security team” for a company of one.

However, ‘vibe coding’ a solution is the easy part; the challenge lies in the Day 2 operations. When you build a custom agentic workflow, you aren’t just the creator, you are the lead engineer, the security officer, and the maintenance crew. AI models are not static; they suffer from ‘model drift’ where performance degrades over time and a single update to an underlying API can break a custom-built ‘Lego’ chain overnight. For many SMEs, the dream of a ‘company of one’ can quickly turn into a nightmare of constant troubleshooting, where the time saved by automation is swallowed whole by the technical debt of maintaining a bespoke system.

So while AI present some huge opportunities, whether or not they are right for every small business, is another question.

Agentic Commerce: The New Middleman

The most radical shift, however, is the move toward Agentic Commerce. As MD of Cloud Ventures, Wall is at the forefront of “programmable money”, the integration of digital wallets directly into AI agents.

“Each agent will have a wallet… it’ll just pay it and consume it and off it goes,” Wall explains. This creates a new “Machine-to-Machine” economy. For small businesses, this means the expensive game of human-centric marketing (SEO, brand awareness, UI/UX) is being bypassed. If your business provides the most efficient API endpoint for an AI agent to fulfill a task, whether it’s booking a flight or ordering HVAC parts, the agent will find you and pay you instantly using stablecoins.

Experiment and ask questions

“Lean into AI and use AI at every possible opportunity,” implores Ben Thompson, CEO of Employment Hero. “Every time you use it, you learn something. You may feel like you’re deep in the pool and don’t know how to swim – that’s fine, you’re learning, you’re not going to drown. Just get your hands on all the available tools and play with them, inside of work and outside of work. It’s really important that everyone becomes AI-proficient

In this new landscape, the structure of the company appears to be compressing. For many, the goal is no longer to be big; the goal is to be dense, a small, high-powered nucleus of humans directing a fleet of autonomous agents. As Wall puts it, “I don’t think it’ll be as bad as people think, but it’s happening so fast.” The “David” of 2026 doesn’t need a sling; they may just need a local build and a well-structured database.

Yet, while local builds promise a ‘fortress,’ they also remove the safety net provided by enterprise-grade platforms. In a world of tightening AI regulations and autonomous agents handling real money, the ‘hallucination risk’ becomes a liability risk. If a custom-built agent makes a hiring error or an autonomous wallet executes a flawed contract, the small business owner stands alone.

This is where the ‘build vs. buy’ debate reaches its final hurdle: is the 100% custom fit worth 100% of the risk, or is there a middle ground in platforms that offer the power of agents with the guardrails of an established ecosystem? Ultimately, that question could prove to be the most important one facing SME operators today.

5 Places SMEs Can Deploy Agents Today

If you are ready to move from observation to execution, these five platforms offer the most practical “entry points” for agentic workflows in 2026:

• Employment Hero (Recruitment): Use their Hero AI to automate the most time-consuming parts of hiring. It can autonomously screen resumes, conduct initial video interviews and score candidates based on your specific metrics, acting as a tireless digital talent scout.
• Relevance AI (The “AI Workforce”): This platform allows you to build an entire “digital workforce” by snapping agents together. You can deploy a “Sales Agent” to qualify leads, which then hands off data to a “Research Agent” to prep your meeting notes, effectively creating a multi-departmental team with zero additional headcount.
• Gumloop (Operations): An AI-first automation builder that lets you drag and drop “nodes” to create complex workflows. It’s effectively “Zapier with a brain,” capable of scraping competitor sites, analyzing market trends, and executing multi-step tasks autonomously.
• Intercom Fin 2.0 (Customer Support): A customer service agent that goes beyond simple chatbots. It is trained specifically on your company’s data and can resolve complex queries, process returns, and manage bookings without human intervention.
• V0 / Lovable (Product Development): For the “vibe coding” Cameron mentions, these platforms allow you to describe a web application in plain English and watch the agent build the application. Great for newbies.