The Signal
57% of small businesses in the US are now investing in AI, up from 36% in 2023. And 85% of those that adopted AI expect positive returns. But here is the part that gets missed: small businesses have a structural advantage. No legacy systems, no committees, no 18-month approval cycles. The question is not budget — it is knowing where to start.
The Story: David vs. Goliath, AI edition
There is an assumption floating around that AI is an enterprise game. That you need a data team, a six-figure budget, and a "Chief AI Officer" to get meaningful results. The data tells a different story.
Salesforce surveyed over 3,300 SMB leaders and found that 91% of small businesses using AI reported revenue growth. Even more telling: 83% of growing businesses are already using AI, compared to 55% of declining ones. The gap is not about resources. It is about willingness to move.
Why small wins big
Large companies have a problem that small companies do not: complexity. A 5,000-person enterprise that wants to automate invoice processing has to coordinate across procurement, finance, IT, legal, and compliance. That process takes months — sometimes over a year — before anything goes live.
A 20-person company? The founder talks to the bookkeeper on Monday, sets up an AI invoice capture tool on Tuesday, and has it running by Friday. No committee. No 40-page RFI document. No six-month pilot.
McKinsey's 2025 State of AI report confirmed this gap. While nearly half of companies with $5 billion+ in revenue had reached the scaling phase with AI, only 29% of smaller companies had. But here is the nuance: that is not because small businesses are behind — it is because they do not need "scaling." They need one workflow that works.
The "start with one workflow" framework
The businesses I have seen get the most value from AI do not try to transform everything at once. They pick the one task that eats more than two hours of someone's week, and they automate that first.
Not the most complex process. Not the one that sounds most impressive. The most annoying one. The one someone on your team quietly dreads every week.
Once that first workflow is running, something interesting happens. The team sees what is possible. The second automation is faster, because the mental model already exists. The third is faster still. It compounds.
Five high-impact starting points
If you are wondering where to begin, these five workflows consistently deliver the fastest returns for teams of 10–50 people:
Support triage — AI reads incoming messages, categorises them by urgency and topic, and routes them to the right person. Saves 5–10 hours per week for teams handling 50+ enquiries.
CRM updates — Instead of manually logging calls and emails, AI captures key details from conversations and updates your CRM automatically. Research shows this kind of email automation saves 5–10 hours per week.
Meeting notes — AI transcribes meetings, extracts action items, and distributes summaries. No more "can someone send the notes?" follow-ups.
Invoice capture — AI reads invoices, extracts line items, and pre-fills your accounting software. Businesses handling 100+ monthly transactions report 30–50% reductions in accounting time.
Content briefs — AI drafts outlines, pulls in relevant data, and creates first-pass briefs for blogs, social posts, or proposals. Teams report 3–5x content output with the same headcount.
What it actually costs
Let us talk real numbers. A typical small business AI stack — covering writing, scheduling, meeting notes, and basic automation — runs between $100 and $300 per month. That is roughly the cost of a single team lunch.
Against that, the hours saved are significant. Even a conservative estimate of 10 hours per week at $40/hour comes to $1,600 per month in recovered time. Industry data from Swfte AI suggests that SMBs achieve positive ROI within six weeks of AI implementation, with 27% productivity increases and 23% cost reductions on average.
The maths works. But only if you start.
At LF Labs, we work with companies this size every day, and the pattern is always the same: the biggest barrier is not cost or complexity — it is the decision to begin.
The Operator's Toolkit: The $0 AI starter stack
You do not need a budget to get started. These tools all offer free or very low-cost tiers:
ChatGPT or Claude — Drafting, summarising, brainstorming. Both have free tiers that are surprisingly capable for everyday business tasks.
Otter.ai — Meeting transcription and automatic action items. Connects to Zoom, Google Meet, and Teams. Free tier includes 300 minutes per month.
Notion AI — Internal knowledge management and SOPs. Keeps your team's processes documented and searchable, with AI that can answer questions about your own docs.
Zapier or Make — Connect your tools and automate workflows without code. "When this happens in Tool A, do this in Tool B." Free tiers cover basic automations.
Calendly with AI routing — Scheduling without the back-and-forth. AI-powered routing sends the right meeting to the right person based on the enquiry.
Start with one. Get it working. Then add the next.
The Radar: Three things worth knowing this week
AI goes local. Edge AI chips are advancing rapidly, with the market forecast to exceed $80 billion by 2036. The trajectory points toward sophisticated AI running on standard business hardware — no cloud dependency, no ongoing compute costs, and better data privacy by default.
Outcome-based pricing is here. The old model of paying per software seat is giving way to pay-per-result. Zendesk now charges per resolved ticket. Intercom charges $0.99 per AI resolution. Salesforce Agentforce costs $2 per conversation. For small businesses, this means you pay when AI delivers value — not before.
30% of small business employees now use AI daily. That is up from near-zero two years ago, according to Business.com's 2026 Small Business AI Outlook. Chatbots are the most common entry point, but adoption of advanced tools like predictive analytics is climbing fast.
From the Field
I would love to know: what is the one task that eats the most time in your business each week? The repetitive one. The one someone on your team quietly dreads. Hit reply and tell me — I read every response, and it helps me write better editions for you.
Until next time,
Francois