7 April 2026 · 2 min read · John
There's a Forrester report doing the rounds called "Mind the Agentic Action Gap." The title says it all. Most AI agent projects stall before delivering any return — not because the technology doesn't work, but because the agents can't actually do anything useful inside the business.
Forrester calls it the "action gap." An AI agent can analyse data, summarise a document, draft a response. But when it comes to acting on that — updating a system, sending something, completing a real task end to end — it hits a wall. No access to the right tools, the right data, the right permissions. So it generates insight, and then a human has to do the work anyway.
Most AI agent platforms are built top-down. They start with the technology and try to push it into your workflows. The problem is that workflows in small businesses aren't standardised, documented, or sitting neatly in an enterprise system. They're in someone's head, in email threads, in "the way we've always done it."
Forrester's readiness test for businesses thinking about AI agents is telling: can you find formal documentation on how a specific task gets done in your business? And does that documentation reflect how it's actually done? For most companies, the answer to both is no. For most small businesses, the question doesn't even make sense — there is no documentation.
Two things.
First, be sceptical of anyone selling you an "AI agent" right now. Enterprise platforms aren't ready for firms your size, and the cheaper tools calling themselves agents are usually chatbots with a new label. Forrester's research backs this up — the gap between what's promised and what's delivered is wide.
Second, the underlying idea is sound. AI that doesn't just advise but actually does the work — that's where the value is. The problem isn't the concept. Starting with a platform and hoping it fits your business is backwards. You need to start with the task and build the tool around it.
We don't sell AI agents. We build AI tools configured around specific tasks in your business — proposals, emails, client documents — using your voice, your processes, your data. There's no action gap because the tool is built to do one thing well, inside a workflow that already exists.
Less exciting than an autonomous agent that runs your whole business. Also far less likely to end up as shelfware.
If you want AI that actually does the work, not just talks about it, let's have a conversation.
Want to see how this applies to your business?
Book a free 20-minute call →According to Forrester's 'Mind the Agentic Action Gap' report, most AI agents stall because they lack access to the right tools, data, and permissions to complete real tasks end to end. They can analyse and summarise, but can't act — so a human still has to finish the work, and the return never materialises.
Not yet, for most small firms. Enterprise AI agent platforms assume standardised, documented workflows that small businesses typically don't have. Cheaper tools marketed as agents are usually rebranded chatbots. The better approach is building a focused AI tool around a specific task rather than buying a general-purpose agent platform.
An AI agent is designed to operate autonomously across multiple systems and complete complex workflows. A task-specific AI tool is configured to do one job well — drafting proposals, writing client emails — within a workflow that already exists. For small businesses, the latter is more likely to deliver measurable ROI.
Forrester's own readiness test asks two questions: do you have formal documentation for how specific tasks get done, and does that documentation reflect reality? If the answer to either is no — which it is for most small businesses — you're not ready for an AI agent platform. Starting with a single, well-defined task is a more practical entry point.