Human-Powered Trust is the Real Moat: A Case for Tech-Enabled Services

Everyone is talking about AI. It is completely changing the way businesses – and individual consumers – leverage technology and perform day-to-day tasks.

But I’ve also lost count of the number of times I’ve heard someone use the term “SaaSpocalypse” in meetings during the past couple of months. 

Meanwhile, the most durable companies may just be the quiet ones. Especially those that sit on the sidelines of this SaaSpocalypse, churning out cash while the rest of the tech world is practically lighting it on fire.

I think it is very possible that the most durable companies will not be flashy, sparkly unicorns raising capital at 20x revenue multiples. These less flashy businesses may never be AI-native. But, that certainly doesn’t mean they can’t be valuable companies that turn into life-changing opportunities for founders and deliver good returns for investors. Does it?

‍

AI Tools and Due Diligence in Investing

At Straylight, we had a full execution and value creation team meeting this week and spent two hours planning how we should best leverage AI tools, how we want to incorporate the assessment of AI-related risks and moats into our due diligence, how we plan to assess our portfolio of nearly 25 active investments on whether they are creating value from AI, all alongside their business model, risk-profiles. 

We’ve listened to advice from our auditors and stayed away from notetaking tools in order to keep company information confidential. We are standardizing an enterprise Claude plan because of its Microsoft integrations; we love our Excel models and PowerPoint decks. 

My colleagues have also found immense time savings and efficiency using Claude and ChatGPT, among other tools. One of our partners vibe coded a branded, proprietary database: of exit opportunities for our portfolio companies and new investment opportunities coming in the door, all generated from data our team sourced first-hand from hundreds if not thousands of calls and meetings we’ve held with venture and private equity firms. Another one of our partners created two GPTs for our deal team to leverage: one that assesses specific acquisition opportunities we feed it for our portfolio, as if it were a corporate development director, the other creates pre-meeting briefs on new investment opportunities. 

Additionally, given the suppression of public software comps, we are creating a framework to use on two live deals that will help our investment committee understand in detail the areas of risk and opportunity for a deal as it relates to AI, helping inform us whether we are investing at the right valuation or not.

Yes, I am fully bragging about my team – they are all way smarter than I am when it comes to AI use cases, and much of this piece has been influenced and inspired by these colleagues. I know that I need to do the work to get smart on AI tools and my team is helping to push me into the deep end, which I appreciate as I dive in. 

I want to add a disclaimer: we are not currently allowing AI to fully replace any of our analyses as investors. We still expect our team to do their own work and check any and all outputs from AI. Our team is expected to think critically and perform both the quantitative and qualitative analysis, much of which requires nuanced understanding and actual industry experience. Yes, we can probably create an AI tool that delivers a bottoms-up financial operating model, but I believe that in order to leverage the tool you still need to be able to feed it the right inputs. You need to be able to understand what goes into COGS for a SaaS company, how to create efficiencies in spend that increase gross margins, and what historical ACVs and deal size + count per AE on your sales team will allow you to reach $25MM in revenue in three years while operating at a positive EBITDA profile. 

And in order to take an output from an AI tool, such as a new investment opportunity deal summary, and do anything useful with it – like advocate for the opportunity to an investment committee – I believe I still need an understanding of the SaaS industry, informed by years if not decades of experience as either an investor, operator, or both, as well as where the firm sits in the investment landscape, and how we should value opportunities in order to achieve the return to which we are underwriting. 

‍

What’s next?

Meanwhile, I keep wondering about the defensibility of our investments. Amid this continued excitement around the adoption of AI, there is also fear and pause among software investors because of slumping public software multiples.

A few months ago, I wrote about the risks that come with having overly strict investment mandates and how it could turn entrepreneurship into a commodity. Now, I keep asking myself: what business models are the most durable in a world where coding and software generation has nearly become a commodity? The answer to me, at least for now, increasingly points to tech-enabled services business: companies that leverage software to improve internal workflows and the productivity of humans, while still relying on outcomes delivered by humans.

At Straylight, we aim to invest in durable technology companies that are overlooked by the traditional systems of venture capital and private equity. We seek out vertical SaaS and tech-enabled services companies that have natural product moats, often with industry-specific value propositions. We look for durability (i.e. capital efficiency) over high-burn, so that if the sh*t hits the fan, the company – and its investors – are not faced with a capital impairment reality and go to zero, but instead can exist on low or no cash burn. 

These traits alone, however, don’t necessarily equate to strong defensibility against AI disruption. I believe that having a business model that inherently depends on the trust established during a human-to-human relationship is software’s strongest defense against AI. 

Straylight isn’t the type of firm to jump at the newest, hottest, most sparkly and shiny new company. We’ve never invested in a blockchain deal. We’ve yet to do an AI-native or AI-agent deal, although nearly all of our companies have adopted and leaned into AI over the past several years. And to be honest, this lack of investing in an AI-native company is largely because any of the AI companies we’ve evaluated have asked for comparatively high valuations, and we didn’t think it prudent to pay a higher multiple for a company than we expect to be able to sell it for in three to five years. 

We don’t chase the next unicorn or  rocketship, and while tech-enabled services companies are far from sexy, we’re okay with that. They may have lower gross margins and thus the overall unit economics may be less exciting. But, I’ll take my mid 60s or low 70s gross margin business that is breakeven with line of sight to 80% gross margin, over your 95% gross margin, burning-a-million-dollars-a-month business. It’s likely it will run out of cash in less than two years and be forced to raise another 20 million bucks, kicking the can on any possible near-term liquidity event while diluting down everyone on the cap table… including the founder.

‍

Trust: beyond the “human in the loop”

Tech-enabled services companies by their very nature involve human effort, with someone involved in the delivery of the offering or service the customer is purchasing. If the tech-enabled service is human-forward, and the customer interacts directly with that human, there is an intrinsic layer of trust built into the delivery of the service. I believe that the trust and reliability that comes from an interaction with a person is going to be nearly impossible for AI to replicate and displace. Yes, I know and acknowledge humans can lie, but that is not what we are talking about here; in a business and delivery of goods-and-services setting in a marketplace, I believe you can trust the person or expert you are relying on.

But can’t AI agents offer the same information or service that humans do in many business scenarios, you might ask? 

Sometimes yes, but often, no, and that’s because we as humans crave truth, alongside being able to trust someone or something. It’s also despite my shocked delight at the local Taco Bell drive-thru recently, when I placed a cheesy fiesta potato order with an agent. My first ever AI drive-thru ordering experience. 

Conversely, just this week I was trying to resolve a billing question on an invoice and utilized the issuing company’s website ‘AI-chatbot’. After asking my question, the bot routed me to a real human. Or at least, made it seem that way. I had expected the ‘AI-chatbot’ to do more to try to help me. And this from a venture-backed software company which has raised nearly $150M. 

In another example, because of a change in our personal tax situation, my family and I decided to work with a tax advisory firm for filing our taxes. While the firm uses a tax practice management software (one backed by Accel-KKR), to collect and organize my tax documents, I still opted into a one-on-one appointment with the tax advisor to be able to ask and get answers from a real person I could trust to get it right. 

When you think about industries like health care (telehealth, medical staffing, clinical trials), accounting and bookkeeping, security and risk mitigation, insurance, recruiting, and other high-touch advisory services like management consulting, all involve some level of person-to-person trust. Personally, I just don’t think AI is going to be able to disrupt or replace that trust. Ordering fast food is a perfect example of where trust doesn’t matter: the stakes are low, the workflow can easily be standardized and automated, and the outcome is predictable and easy to achieve with technology. But when the stakes increase – tax filings, complicated billing questions, patient education – accountability, responsibility, and trust become crucial. 

That’s why I believe the most durable companies - including tech-enabled services businesses - are going to be those that combine software with human expertise. In an era obsessed with doing more with less, this means that irreplaceable human-to-human trust… might just end up being the most durable moat of all.

‍

‍Michelle Erikson is Vice President at Straylight Capital. Her role involves assessing the health and scalability of potential investments, and supporting the growth of portfolio companies. Prior to Plymouth, she led marketing for the University of Michigan’s Athletics Department and worked for a venture-backed sports accelerator. While not a Midwesterner by birth, she has adopted us lovingly as her home: she loves Euchre, the Upper Peninsula, and driving by corn fields on her commute.