No one likes their ERP, CRM or HR software - picture something like Salesforce, Workday or SAP. Yet these “Systems of Record” (SOR) are incredibly sticky because they manage important customer, employee and financial data. As a result, they’ve been hard places for startups to compete. But with AI, I’ve been seeing an opportunity for new SORs to emerge. The opportunity is with SMBs in services sectors - which were historically poor places to build startups. It’s a textbook example of Clay Christensen’s theory of disruption. The playbook is as follows.
The playbook
1) Identify a segment poorly served by an incumbent System of Record
In many industries this means SMBs, since SORs tend to focus their product and sales strategy on enterprises. In some cases, SMB segments are unintentionally underserved as SORs continue expanding their offering across more and more verticals (see the image below for ServiceTitan’s breadth of coverage in home services). Generally speaking, the further down market you go, the more fragmented the tech stacks become and the higher likelihood of encountering weaker SORs (or simply spreadsheets + brittle automations).
ServiceTitan’s website lists 24 different segments covered by their software. Whilst these sectors have some similarities, there are also important differences between them (e.g. impact of seasonality, one-off versus recurring in nature) which impacts how well one-size-fits-all software can serve them. Each segment also has a different mix of enterprise vs. mid-market vs. SMB providers
2) Use AI to build a high ROI wedge product
So far, the advice may sound quizzical. Focus on an SMB segment in a services industry? With a low historical willingness to pay?
But it’s precisely because of these (historical) disadvantages that it’s crucial to focus here. AI has unlocked new value propositions that are 100x better than the status quo. The best value - irrespective of sector - is generating net new revenue for businesses (even more so in a tariffed world). That, coupled with weaker competition from incumbents, can help a nimble startup make quick inroads.
I have seen voice AI serve as a powerful (yet easily replicable) wedge across a number of industries. An example is 24/7 front-desk call answering for vet clinics, law firms and car dealerships. In these industries, inbound callers typically have a high intent to purchase, which means answering every call can make the business more money.
3) Focus the value proposition on labor automation to juice your ACVs
The other big disruption with AI has been the building of software that automates jobs (such as front office intake). This can justify ACVs that are substantially higher than regular SMB software, turning a previously low-value segment into an attractive one. If done right, it can benefit from SMB-like sales cycles but charge mid-market size ACVs. This isn’t just a cute theory, but is something we are seeing play out across our portfolio, with startups charging SMBs $20 - 50K ACVs.
4) Build functionality that helps users migrate away from the SOR
So you’ve found an overlooked market segment, built a high value wedge and found a job to automate. From here, it’s important to architect your product in a way that discreetly maps the SOR, allowing users to seamlessly migrate away from it. There are three approaches I’ve seen work:
- Connect directly to the SOR —> in the case of the AI Vet Receptionist shown in section 2) above, the product has more value when it connects to the SOR because the AI can reference patient-specific details (e.g. medical history) when answering questions. This allows for both a mapping of various SOR fields as well as an understanding of where it is lacking. [Note that API access can be turned off when the SOR begins to see you as competitive]
- Mimic the SOR’s schema —> this is best conveyed through an example. Consider the company Soff (shown below), whose wedge product automates data extraction for manufacturing companies. Soff automates data extraction from PDFs, emails and spreadsheets and pushes it into SAP, saving customers meaningful time on data entry. But more importantly, Soff’s system ends up also recording the same information that gets populated into SAP, shadow-building SAP’s data schema
- Build export functionality into the wedge —> building easy export functionality can also provide an unfair advantage when competing with an SOR. In its early days, Klaviyo (an example marketing company) built a tool to seamlessly export data from Mailchimp in seconds, giving them a huge customer acquisition advantage when selling to those same customers
5) Go multi-product rapidly
With a wedge firmly in place, it’s essential to rapidly become a multi-product offering. Time is of the essence with this approach because of the risks associated with being connected to the SOR per 4). This is also why focusing on a market segment that is less core to the incumbent SOR can provide important air cover.
When it comes to going multi-product, the focus should be on getting to core SOR feature parity as quickly as possible. Note that if you are targeting an underserved segment like SMBs, you do have the opportunity to rethink what the core feature set offering is. Nonetheless, you will likely also need to build some basic ERP, CRM and HRIS functionality to entice customers to switch. (This is obviously no easy feat, but thanks to AI code gen tools like Cursor/Cline/Windsurf, building software has become radically easier, shrinking development time by an order of magnitude. We are seeing this multi-product buildout take fast moving startups a few months).
Pricing as a differentiator
It’s not always possible to find a market segment untouched by incumbent SORs. In these cases, there is still the opportunity for AI startups to compete in the way outlined above. In these cases, pricing can be the best vector of competition - specifically, pricing on the basis of outcomes rather than seats. If done right, this can make it structurally impossible for incumbents to compete without cannibalizing their core offering (see below for an example).
Imagine an SMB plumbing company that has a front office rep, a dispatcher and a scheduler (a common setup). An incumbent SOR may price on the basis of seats, say 3x $12K/seat. If the startup is able to automate a substantial portion of these three roles with AI, the plumbing company may be able to consolidate the three roles into one. If the salary for each role was $50K, the company can save $100K in salaries, pay the same $36K ACV to the startup and still save $100K! The startup could likely justify an even higher ACV than that given the cost differential.
Industries in which SORs are vulnerable
I believe many services industries are vulnerable to the playbook above. Some sectors I’m monitoring include:
- Manufacturing
- Construction
- Collections
- Marketing and advertising agencies
- Customer service
- Home services
- Accounting and audit
- Recruiting
If you are building something that resembles an SOR and the above resonates with you, I’d love to hear from you at samit@1984.vc.