Shane Wilson's Blog

What is the opportunity for VC Investment in AI startups targeting Small and Medium Businesses (SMBs)?

On July 15, 2025 I attended Between Two Founders #007: Town Co-Founder & former Plaid CTO Jean-Denis Greze hosted by Jeff Seibert of Digits.  I RSVP’d because I wanted to learn the inside story of hyperscaling, but what stuck with me was a question: how big of an opportunity is AI for SMBs? 

Jean-Denis left Plaid in 2024, and has since launched Town.com, a tax advisory platform designed to empower small businesses with AI-enhanced expert guidance.  Town.com raised $18M in Seed funding in March 2024 with participation from First Round and Alt Capital.  During the talk, Jean-Denis explained how despite the consensus advice from VCs to avoid SMB-focused products (rooted in the challenges of distribution) the Town.com co-founding team gravitated to independent business owners.  Characterizing what may be considered “lifestyle business owners”, Jean Denis explained: “Everything that gets in the way of them running their business—stuff like tax—they just don’t like.”  Furthermore, many of these founders have been experimenting with AI for simple use-cases in their day-to-day work which has warmed them up to the idea of relying on AI in their businesses.  This, Jean-Denis sees, is an opportunity for AI-powered solutions to embed themselves in key SMB workflows.  

Thus-far, my thinking has erred towards the VC consensus: don’t focus on these small-scale applications, the opportunity is too small and delivering value to these customers will likely require a higher-degree of service.  Yet last night, I thought back to a class from my MBA on Programmatic Acquisition Strategies taught by Professor AJ Wasserstein that highlighted a myriad of multi-millionaire entrepreneurs who capitalized on fractured markets and multiples arbitrage to build impressive roll-up platforms.  In essence, paying attention to overlooked, ‘boring’, service-based companies can be extremely lucrative for those that acquire and operate efficiently.  Is the VC consensus on SMBs wrong? Is this an opportunity to be non-consensus and right?

As I started picking on this idea I envisioned one combined post.  Now that I’m down the rabbithole, I’m changing course.  In this document, I’ll structure my initial thinking on the investment opportunity for SMB-focused AI solutions.  In following posts I intend to dive deeper, exploring:

• Market sizing key sub-verticals in SMB Business services segment
• Looking for signal in history: SMB-focused startups from prior VC Waves
• What hasn’t been served by technology that could be now?
• Evidence of early leaders, 2022-2025
• Market forces and Early-stage companies to watch

If you have thoughts on this project or an interest to connect with companies highlighted here shoot me an email at shanewilson9898@gmail.com.

The Case for SMBAIVC

In the United States, SMBs (<500 FTEs) are responsible for 44% of GDP, ~50% of employment, and more than $185B in tech spending. [1] Since the end of ZIRP, SMBs have indicated inflation and staffing as critical challenges to growth. [2]  This creates an opportunity for AI-powered solutions that can reduce demand for costly external business service providers and replace expensive and hard-to-find talent with agentic solutions. 

Some levers that may be at play that could make SMB AI interesting:

• Bet on multiple, pick a winner
  SMBs are highly fractured across industry, geography, and workflow. What looks like noise to a generalist can be a structured targeting opportunity for a focused team. Founders can differentiate by segment, distribution, or channel, and success doesn’t require category dominance—just depth in a high-need niche. As investors, this fragmentation allows a multi-shot strategy with the ability to double down quickly on early traction.
• Valuations and deal terms could be more favorable
  Because many investors remain cautious on SMB-focused businesses, pricing and competition at the early stages may be more reasonable. This creates potential to secure meaningful ownership without the crowding that often accompanies enterprise-themed rounds.
• Faster traction, easier purchasing
  SMBs move faster. Without procurement drag or layers of internal buy-in, real usage can begin within days of a sale. This makes it easier to gauge product-market fit early and concentrate capital behind breakout performers.

Why Big AI won’t dominate this segment

As I think back to my days training early-career sales reps, I remember saying: 

Taking $25k from a small business is not the same as taking $25k (and likely, 4x-10x that) from a large business.  If you consume all of a company’s tech budget, they expect you to solve all of their tech problems.  The big business may forget you’re even there.

One consequence of mega-valuations in the AI space post-ChatGPT is that super-funded companies need to prioritize big-game hunting with respect to customer acquisition.  Designing products for large enterprise clients encourages optimization for broad applications that can embed themselves in multiple, mostly generic workflows across large companies.  Specificity can be achieved through expensive service-based add-ons and customization. This creates a value misalignment with SMB leaders,  as “businesses with more than 25 FTEs are approximately four times more likely than other SMBs to factor in the quality of support…when they make tech-buying decisions.” [3]  It’s reasonable to believe that aforementioned challenges for SMBs that could be served by AI are being overlooked.  Businesses that are willing to provide high-value, contextualized services can capture more value from the market.  

Rather than leaving this assumption “Big AI will over-index to Enterprise Customers” dangling, let’s take a look at some broad figures from publicly traded software companies to determine if indeed there is a bias towards large customers in established companies.  It’s challenging to define apples-to-apples comparisons here as companies provide different definitions for their small and large customer segments- some segment by customers’ annual revenue, others by which plan-level their customers buy.  With that disclaimer, let’s take a look:

Company	Market Cap	SMB Revenue %	Enterprise Revenue %	Source
Cloudflare	$66B	66%	34%	Investor Presentation – Q1 FY25
Zoom	$23B	31%	69%	SaaStr Analysis
HubSpot	$28B	85%	15%	Morningstar, 2023 Report
Shopify	$155B	68%	32%	Shopify Q1 2024 Financial Results
Atlassian	$50B	61%	39%	Atlassian FY ‘24 Investor Day
Intuit	$209B	58%	42%	Intuit 2024 10-K
Squarespace	$6.5B	59%	31%	Sitebuilder Report

In the US, 88% of businesses are represented in the 0-500 employee range.   Of the provided sample, only HubSpot is particularly close to this weighting between small and large enterprises.  Of the above companies, the average delta between the percentage of small companies (88%) and revenue derived from small companies is 28%.  This is a very crude way of saying “the focus of publicly traded software companies is 28% overweight to enterprise.”  This ignores nuance regarding the relative contract size of extra large providers, but when evaluating TAM for SMB-focused startups, could we adopt this figure as an ‘uncaptured marketshare’ adjustment to refine startup TAM? 

One other salient feature from this analysis- there is reason to believe this focus drift will worsen. Freemium business models are a core tenant of the software-as-a-service playbook. Companies ship a product that providers consumer-level or small-team value at a free or near-free price point.  Enterprise users then self-segment themselves by opting for higher-priced features that align with large-company needs.  In this way, Free users get the diet version and Enterprises get the premium.  There’s evidence legacy providers are leaning on this trope for AI product rollouts-providers are rolling out high-priced AI solutions with their largest Enterprise customers first.  This trickle-down business motion biases AI solutions for large enterprises.  There’s also reason to doubt that large enterprises will appreciate AI value-add on-par with SMBs.  Messy data structures, tech debt, and lack of organizational alignment can make AI ‘fancy toys’ that do not get played with across big companies.  Smaller, nimbler orgs where leaders have a more comprehensive view of their businesses and more control over tooling could be better users and customers for training these products. All that said, startups who adopt competitive strategies to win the smaller companies with small-company-centric solutions could be a compelling strategy for long-term value capture and the development of better tools for all.

But how do you provide quality service at scale?

Before trudging along, let’s qualify this thesis.  Building value with service is a dangerous game in venture-backed businesses.  Companies that derive 100% of revenue from low and light-touch products command double-digit revenue multiples.  Service-derived revenue typically receives a 1x-3x.  While companies may be willing to pay you more for your service, you are playing a talent-heavy short game that is inherently unscalable.  Companies that rush to market with half-baked AI solutions that are supplemented with heavy human service are also indefensible- any other vibe coder with some industry context can quickly enter your market and start competing on price. 

From a qualitative standpoint, it is critical to evaluate how well SMB-focused AI solutions are providing high-value, automated service.  During the Between Two Founders Talk, both Jean-Denis and Jeff Seibert (of Digits, an accounting and bookkeeping software) touched on this idea.  Jean-Denis described how Town.com described his aspiration for no-thinking, high-value service to SMB owners:

“My dream is if you’re [a] small business owner and you sign up for Town, then what happens is on January 15, you have an email and it says, your draft tax return is done. And you click on it, and you get a beautiful video that explains to you your taxes and how much money we’ve saved you. And then you press one button, and it’s filed.”

Doing this will require more effort than connecting to OpenAI’s API, but it doesn’t necessarily need to be manual work.  Jeff Seibert’s company Digits, a bookkeeping and accounting solution that’s raised $65M from investors including BVP, Benchmark, and GV explains that less-sophisticated Machine Learning techniques could provide specific, automated services where LLMs currently fail: 

“The big thing that we’ve learned is that book keeping is fundamentally not [a hallucination problem]. … It’s a predictive problem space. And actually classic ML is far superior to that of LLMs here…All [LLMs] appear to be accurate until they get around 70% accuracy. … We then ran the same test with human accountants … [who] show around 80% accuracy. Our predictive models trained on the individual business are now getting up to 98+ percent accuracy.”

In summation, there is likely a hole in the AI market with respect to serving SMBs.  This is a function of the high demand for quality of service within this buyer segment.  How startups solve this problem, ensuring that their business model is not beholden to human-in-the-loop service is a critical question investors should unpack during diligence.  

Citations:

[1] McKinsey, Winning the SMB tech market in a challenging economy

[2] Ibid.

[3] Ibid.