Claude is now the dominant AI spend for small business. A new Capital on Tap survey found that 1-in-8 UK small businesses now pay for AI tools, up from just 1.1% in 2023 to 12.8% today. Of that spend, Anthropic's Claude captures 54%, versus OpenAI's 27%. That's a striking reversal from the narrative that ChatGPT owns the small business market. Average quarterly spend sits at £75.60, which is not nothing, but also not ruinous. Publishing businesses lead adoption at 26%, followed by consulting and media. The real number that should make you pause: over 28% of SMEs are running two or more platforms. That's where duplication creeps in and your actual cost quietly doubles. The CEO's advice in the report is worth taking seriously. Audit each tool honestly, and keep sensitive data off any platform you haven't properly vetted. Full story here.
Three owners told Business Insider exactly where AI marketing breaks down. The honest part: Meta's built-in AI tools have a conflict of interest baked in. They push toward more ad spend because that's what Meta wants, not what your margin wants. ChatGPT and Claude work better when you bring real brand context with you, not when you ask a generic question and expect a tailored answer. The consensus from all three owners: use AI to move fast on diagnostics and copy variations, but never let the platform grade its own homework on budget or positioning decisions. That last line is the one to write on a sticky note. Read the owners' breakdown here.
Agentic AI has arrived for small business. That's good news, mostly. Agentic AI means systems that can plan and carry out multi-step tasks on their own, without you babysitting every click. Think of it like hiring an assistant who can handle a whole project, not just answer one question at a time. It's no longer just for enterprises with IT departments. But here's the warning that actually matters: an agent amplifies whatever process you hand it. Broken workflow in, faster broken workflow out. Before you deploy anything agentic, you need clean data, documented workflows, and staff who understand what the agent is supposed to accomplish. The technology is ready. The question is whether your operations are. Full breakdown here.
Tools Worth a Look
NitroStack: One environment to build, test, debug, and deploy AI agent workflows and MCP servers. For technical founders and developers building production AI apps.
Craft Agents: Open-source, locally-run AI agent setup connecting Claude, GPT, Gemini, or local models to your tools via MCP. Your data stays on your machine. For privacy-conscious owners.
Caldera: Turns your Obsidian notes into a public-facing profile with cross-linked backlinks between collaborators. For consultants and content creators.
Emochi: Voice journal that transcribes your spoken thoughts and surfaces emotional patterns through a generative character. Built by a doctor and AI engineers. For solopreneurs who want reflection without typing.
Train with Pasta: Apple Intelligence-powered personal trainer that builds custom workouts on request and coaches you through them with voice guidance. For busy owners who want AI coaching without a gym membership.
Trending on GitHub
MadsLorentzen/ai-job-search: A job application system that runs on your own machine, built on Claude Code. It evaluates postings, tailors your CV, writes cover letters, and preps you for interviews. (1,365 stars today)
stablyai/orca: A desktop, mobile, and remote app that lets you run a whole fleet of AI coding agents (Claude Code, Grok, and others) using your own subscriptions, not theirs. (830 stars today)
affaan-m/ECC: A performance layer for AI coding agents that adds skills, memory, security, and smarter research behavior on top of tools like Claude Code, Codex, and Cursor. (1,145 stars today)
anthropics/claude-code: Claude's own agentic coding tool that lives in your terminal, understands your codebase, and handles routine tasks and git workflows through plain English commands. (179 stars today)
akitaonrails/ai-memory: A long-term memory layer for AI coding assistants, so context doesn't vanish between sessions or when you switch between different AI tools. (178 stars today)
More Headlines
Small business AI adoption is uneven: Early adopters report real time savings, but privacy concerns are the top reason holdouts haven't moved, even as AI spending plans are up across the board.
WordPress and web dev job postings are down over 50%: Small business owners are turning to AI site builders and freelance web dev work is collapsing, a practical signal if you're still paying agency rates for basic site work.
AI is changing how small business lending works: Faster credit decisions, cash-flow-based underwriting, and alternative data scoring mean some owners may now qualify for credit that traditional models would reject.
The next AI edge for entrepreneurs isn't productivity, it's execution speed: Moving from idea to action faster is becoming the real competitive gap, not just doing existing tasks more efficiently.
Buildy.ai lets you build a business by talking through it: Voice-to-company AI coordinates agents doing the actual work, so founders can think out loud and watch tasks get executed.
Meta launched Muse for Small Business: Zuckerberg is pushing beyond consumer AI, and small business is now a named target market.
Google.org committed $15M to build AI tools for Indian SMEs: The grant to The/Nudge covers credit access, market reach, and compliance support for small businesses in India.
What small businesses actually want from bank AI: Stax research breaks down how owners are using AI in payments and banking, and where they want their banks to do more.
AI governance debates are worth watching now, not later: Who sets the rules for AI will affect how you can use it in your business, and entrepreneurs currently have more influence on that conversation than they realize.
-Scott
SmartOwner is published (almost) daily by the team at DigitalTreehouse. Want AI consulting or automations for your business? Reply to this email.