Hello, {{hc_sub_firstname | friend}}! Param here.

Every week brings a new AI announcement and new "This Changes Everything!" headlines. If you own or operate a small business, the noise is genuinely overwhelming, and the temptation is to either ignore it entirely or try to solve everything at once.

At Lynnfield, we have been doing something more boring and more useful. We built a systematic framework for identifying where AI creates real, measurable value inside the businesses we own. It is not about the latest model or bolting a chatbot onto a website. It is about understanding the operational DNA of a business and finding the specific places where AI can eliminate work that nobody is being paid to do.

This framework is still evolving. The technology moves fast and we update our approach constantly, but the underlying methodology is stable, and it works.

Before we touch any AI tool, we map the business. When we acquire a company, or seriously evaluate one, the first thing we do is walk the entire operation end-to-end. We go on-site, following the work from the moment a lead comes in to the moment a customer pays and the job is closed out.

We document everything: every handoff, every form, every phone call, every manual process, every bottleneck. We are looking at the business from the floor, and not from a P&L statement. What do people actually do all day? Where does time go? Where do things slow down? Where do errors happen?

Most small businesses have never documented their own processes end-to-end. The act of mapping almost always surfaces redundancies and legacy steps that exist simply because “that is how it has always been done”. This map is valuable regardless of whether AI ever enters the picture.

Once we have the map, we apply a principle adapted from Elon Musk's manufacturing algorithm. The algorithm has five steps: question every requirement; delete any part or process you can; simplify and optimize; accelerate cycle time; and only then, automate. That sequence is deliberate. The argument, and one we agree with, is that the worst outcome is automating a process that should have been deleted. Our framework draws primarily from the first three steps.

For every step on the map, we ask one question: does the customer actively pay for this? The customer pays for expertise, quality, and outcomes. Internal paperwork, manual data entry, re-keying information between systems, generating reports nobody reads – none of that is what they hired us for. 

We circle every step that fails the test. Then we stack rank those items across three dimensions: how much human effort the step consumes, how much calendar time it adds to the overall process, and what it costs in people, tools, or direct spend.

The steps scoring highest across all three become the AI roadmap.

We are not asking "where can we use AI?" We are asking "where is the most waste?" Those are different questions and they lead to different answers. The first starts with the technology. The second starts with the operation. And only one of them consistently finds the right problems to solve.

This approach also prevents the two most expensive AI mistakes we see in the small business space.

16. Surveys consistently find that small business owners spend roughly 36% of their working week on administrative tasks – invoicing, data entry, scheduling, document management, chasing payments. Multiply that number by the number of employees in an organization and… Ouch!

Continue reading the main story below ⬇️

We have been working on-site with an industrial Heating, Ventilation, and Air Conditioning (HVAC) company. Their estimators carry 20 to 30 years of experience. They know materials, labor rates, equipment specifications, local code requirements, and every judgment call that goes into pricing a complex installation.

A single bid was taking roughly four hours. A client sends over architectural drawings. Then, an estimator manually calculates material quantities, references current pricing, applies labor rates, factors in equipment costs, and builds the bid by hand. Four hours per bid.

When we walked through the process and applied the test, the issue was immediate. The customer pays for the expertise: what equipment to specify, how to route ductwork through a difficult space, what will actually work in their building. They are not paying for looking up material prices, calculating quantities from a drawing, or filling in a bid template. That is an expert's time spent on clerical work.

We took all the structured inputs that go into a bid – labor rates, material prices, equipment catalogs, historical bid data, margin targets – and built an AI system around them. The AI ingests the client's architectural drawings and generates a draft: material takeoffs, pricing, labor estimates, total project cost.

The result: In 30 minutes, the AI-generated bid came within 5% of a manual estimate produced by their most experienced estimator. On a process that was taking four hours!

The new workflow has the estimator reviewing the AI-generated draft, applying judgment to the 10 to 15% that requires real experience – equipment selection, complexity adjustments, risk factors, customer-specific considerations – then finalizing in under 30 minutes. Eight times faster. 

The same team can now respond to four to five times more Requests for Proposal (RFPs) without adding headcount. In construction and mechanical contracting, the average contractor wins roughly one in four competitive bids. Submitting four to five times as many RFPs, at the same win rate, has a direct and compounding effect on revenue – without hiring a single additional person. And in a competitive bidding environment, getting to customers first matters too. Faster turnaround often determines who wins the job before the field is fully assembled.

The estimators and the human review step are still there. But, now, the AI handles the quantitative, repeatable work. The expert handles everything that requires judgment. The expertise is amplified, not replaced. That is the right model for a skilled trade business.

The HVAC example is one instance of a pattern we are seeing across the portfolio. The highest-value AI applications in small businesses consistently cluster around the same areas:

What these have in common is not that they are simple. Many are genuinely complex. What they share is that the customer is not paying for them. They consume disproportionate amounts of skilled people's time while contributing nothing to what customers actually value.

We are not using AI anywhere near customer-facing expertise, skilled trade work, or the relationships that keep customers coming back. In fact, those are the moats that protect our business. This framework is specifically designed to leave them alone while focusing only on the waste.

At Lynnfield, we acquire cash-flowing businesses in the $500K–$5M EBITDA range and offer co-investment opportunities to qualified investors. Join our investor list to receive deal flow as we evaluate new acquisitions. 

Continue reading the main story below ⬇️

AI is not magic. It is a tool. And like any tool, it creates value only when applied to the right problem with the right approach.

Our AI capabilities across the portfolio are still being built. We are testing new tools, updating our approach as the technology changes, and learning in real time from what works and what doesn’t. The HVAC estimating system is one proof point, and it is a strong one. But this is a practice, not a finished product, and I think it is worth saying that plainly.

What will not change is the methodology: 

The result is practical, measurable value creation. Four hours to thirty minutes. Five percent accuracy with expert judgment preserved. As the technology evolves, so will our playbook. That is how you create lasting value, not just headlines.

Thanks for reading!

If you own or invest in a business where estimating, scheduling, or back-office work is a real bottleneck, reply to this email. We are always happy to share what we are learning.

Talk soon,
Param

P.S: In case you’re joining us late, check out the previous editions of this newsletter.