Software Awareness: A SaaS Founder’s Guide to AI Disruption

AI changes both the demand for your software and the supply of alternatives. Your job is to work out what customers will still pay for, why they will buy it from you, and how to strengthen that position given this new reality.

Increasingly capable AI and much cheaper and faster software development have become available to everyone - you, your existing competitors, customers, partners, and a new kind of market entrant.

Just because your customers “don’t even know AI exists” or “wouldn’t ever vibe code their own solution” does not mean you can pretend nothing has changed. Nor will simply adding AI features establish a lasting advantage.

This guide works through four questions, one chapter at a time:

• Demand: Will customers still need what you sell—and will they buy it in the same way?
• Supply: What happens when competitors, customers, and the platforms around you can all build more cheaply?
• Moats: If someone could reproduce your software tomorrow, what would still make your business difficult to replace?
• Offense: What can you use your existing advantages to become—and what should you start doing now?

Each chapter develops its question through examples and ends with five assessment questions and a concrete output for your own business. We follow a hypothetical company selling scheduling and dispatch software to home-service businesses, showing how the same business can move from diagnosis to action.

By the end, you should have identified your biggest exposure, your strongest surviving advantages, and an opportunity to pursue. Answer for a specific customer segment and revenue stream; a company-wide average can conceal the problem.

Named company examples refer to documented product features; the strategic implications are analysis, not claims about those companies’ internal performance or ultimate defensibility.

Demand: What happens to the work customers pay for?

The first risk to examine is whether AI changes the customer’s need for what you sell. You can win against every direct competitor and still lose revenue if customers need fewer seats, stop needing the reports, documents or other artifacts you manage, or buy the result in a different form.

Start by describing the customer’s objective without naming your product. A hiring team wants to evaluate candidates. A property manager wants rent collected and repairs completed. A service business wants profitable jobs booked and fulfilled. Coding tests, dashboards, reminder sequences, and dispatch screens are ways of getting there - not the end demand.

This distinction matters because AI can preserve the objective while changing almost everything around it.

Follow the money from the customer to your product

First, consider the customer’s own business. Software for a translation agency depends partly on the agency’s revenue, staffing, and way of working. Even if demand for multilingual communication expands, the agency may capture less of the spending if the need for human translators evaporates. Its project-management supplier could suffer while the underlying activity grows.

Next, consider the work itself. If AI makes a task much cheaper, customers may do more of it. They could test more campaigns, monitor more systems, or serve customers who were previously uneconomic. But more activity does not guarantee more software spending: the price per unit can fall, or a different provider can collect the revenue.

Then examine the role and workflow your product serves. A team of ten operators could become two supervisors handling exceptions. A weekly report could become an ongoing process that detects a problem and acts. The outcome survives; the seats or reports customers pay for change.

Finally, ask how the customer buys and pays. The person who enjoys your interface may stop being the person who selects the vendor. An operations manager might buy a service that uses your software behind the scenes. Your product can remain in use while losing the customer relationship, expansion opportunity and renewal certainty.

A scheduling business with two different futures

Our hypothetical vendor helps office staff schedule technicians and dispatch jobs. Its founders might initially conclude that their business is safe, because AI cannot (currently at least) fix a leaking pipe nor repair a broken dishwasher.

However, continued demand for the services your clients provide says little about their demand for administrative software. If AI means your customers need fewer dispatchers, a per-dispatcher subscription could contract. If an outside agent handles calls and operates the scheduling system, the vendor may retain the account while losing the direct customer relationship and becoming easier to replace later.

There is an opportunity in the change too. A contractor that could never justify an after-hours receptionist might now pay to capture bookings it previously missed. Cheaper execution can expand the market for completed bookings even as demand for dispatcher seats shrinks.

The demand assessment should distinguish the threat to existing dispatcher seats from the opportunity to serve more businesses or perform more work.

Supply: What becomes the customer’s best alternative?

Assume customers still want the outcome. The next question is how many credible ways they will have to obtain it and how those alternatives change your economics.

“Everyone gets AI” is a useful stress test. Your own gains in product quality and efficiency happen alongside gains available to competitors, customers, and partners. The important comparison is the customer’s choice after everyone has improved.

Look in four places

Customers building internally. A customer may replace a narrow report, approval tool, or integration with something built around its own data and procedures. The scope can be much smaller than your entire product. If a customer only uses three features of your product, it only needs an adequate substitute for those three.

However, internal ownership also brings maintenance, support, and operational responsibility.

Those can preserve the case for buying. Test that case against the customer’s actual requirements and capabilities, rather than assuming either that every customer will build or that no customer will want the burden.

Other software providers. Established competitors and new entrants can use cheaper development to close feature gaps, serve narrower niches, or offer broader products. An implementation advantage can buy you time, but its value depends on what you do with that time. Use it to acquire customers, improve delivery, and deepen advantages beyond the feature itself.

The incumbent ecosystem. Your integration partner, system of record, or implementation consultancy already has context and customer access. It may offer an adequate substitute as part of a broader relationship. “The platform’s version is worse” is weak protection when customers find it sufficient and cheaper or easier to procure.

AI providers selling the work. A new AI agent or service can answer the customer’s needs without replacing every underlying application. It may operate your product on the customer’s behalf, take over the relationship, and decide later which tools remain necessary.

Intercom offers a concrete illustration: its customer service agent, Fin, can operate with an existing helpdesk, while Intercom also sells its own combined helpdesk and agent offering. An automation provider can enter through the work being performed while leaving another vendor’s infrastructure in place. The commercial consequences depend on which provider retains the relationship and spending.

Copying the code is only part of replacement

For our scheduling vendor, generating the code to show a calendar and a dispatch interface might be straightforward. Running tomorrow’s jobs requires accurate availability, service areas, technician skills, travel assumptions, customer commitments, and a way to resolve conflicts.

Those requirements matter. But compare them with the right attacker. A new entrant may lack the necessary context. The field-service platform may already hold it. A service provider might obtain it by operating the existing system with the customer’s permission.

An advantage is always relative to a particular alternative. “We have integrations” means little until you establish whether the most credible competitor already has equivalent access.

Compare total cost at an acceptable quality level: subscription or service fees, implementation, ongoing human work, supervision, maintenance, and the cost of mistakes.

Expect pressure before replacement

A competing option can reduce your pricing power before it takes a customer. Buyers may negotiate harder, resist expansion, or decide that a feature no longer warrants a separate contract. Existing customers can remain sticky while new customers choose something else.

That is why retention alone is insufficient evidence. Review lost deals, discounting, smaller initial purchases, missing seat expansion, and new objections at renewal. Ask sales and customer success to record the alternative the buyer actually considered. “Too expensive” often needs another question: expensive compared with what?

For the scheduling vendor, the most credible threat might be the field-service platform bundling adequate dispatch. Or it might be a new AI startup selling an agent that provides completed bookings. Those imply different priorities. The first calls for a clear reason to buy beyond the bundle; the second calls for deciding whether to provide the agent (i.e., move into the labor budget), serve it (through something like an MCP interface), or risk losing control of the relationship.

Moats: What do you have that remains valuable?

Imagine a competitor reproduced your features tomorrow. Why would customers still choose you?

They might know and trust you. Their business might already run on your records. You might hold an approval, a network, or experience that a competitor cannot get by generating the same code. Those are advantages worth building on.

Here are eight places to look. You do not need all eight. Find the ones that matter to your customers and that your most credible competitors would struggle to match.

Why customers choose you	What a competitor has to earn or rebuild
Buyers know and trust you	Relationships, recommendations, and a route to the buyer
You’re a system of record	The records and connections the business runs on
You’re allowed to act	Permission to book, pay, approve, or otherwise do the work
They don’t check your work	Confidence earned through reliable delivery
Someone official already said yes	Required approvals, licenses, or accreditation
Everyone’s already in it	Active adoption and coordination between participants
You get better as you do the work	Useful experience that improves results and takes time to acquire
It’s hard to keep running	The expertise and infrastructure to keep delivering

Buyers know and trust you

An accountant recommends your product. An agency builds its services around it. Someone inside the customer’s business champions you. These relationships bring you into buying decisions that a competitor may never hear about.

QuickBooks trains and certifies accountants through its ProAdvisor program. Those accountants earn money helping clients use the product. Their recommendation can give a business owner a reason to choose before comparing features. A challenger has to win over the adviser too.

Shopify has a partner ecosystem of agencies and specialists who build stores, customize them, and help merchants sell. Those partners have expertise and services built around the platform. They give Shopify a route to buyers that a better storefront editor alone cannot reproduce.

Your existing customers also give you a place to start when you launch something new. Look for relationships that produce introductions, sales, renewals, or willingness to try your next offer.

Ask yourself: Who already recommends you, brings you into deals, or helps you win them?

You’re a system of record

Your software holds the schedules, balances, or approvals the business runs on. People have configured roles and rules, connected other systems, and organized their daily work around it. Replacing you means getting that operation working somewhere else.

Zendesk records customer conversations and the history of ticket updates. A replacement needs to preserve which issues remain open, what has already been tried, and the context the next support agent needs. Exporting the records is one part of rebuilding that working operation.

Gymdesk connects memberships, payment history, class bookings, and attendance. Some gyms also use it for door access. Replacing it means getting those connections working again so the right members can attend tomorrow’s classes and enter the building.

The strength is in those working dependencies. A database with easy-to-recreate connections offers less protection. An agent could also operate your system while taking over the customer relationship.

Ask yourself: If customers left, what would break or need to be rebuilt?

You’re allowed to act

A tool that recommends a payment has a different job from one authorized to send it. The same applies to committing an appointment, approving a refund, or changing access to a building. Customers have given you permission to make something happen.

Stripe Connect helps platforms onboard accounts, accept payments, and manage payouts. It sits inside the process that moves money. A replacement needs the customer’s authorization and the connections to carry out that work.

Reimbi collects interview expenses, routes them for approval, and sends reimbursements directly to candidates. The customer has entrusted it with completing the payment. You do not have to be Stripe to earn permission to act.

Permission can help you take on more work for the customer. Its strength depends on how readily another provider could obtain the same authority; an API key alone is easy to replace. Ask who controls that access. If another platform can withdraw it, redirect the customer, or replace your role, that limits the protection it gives you.

Ask yourself: What can you do on the customer’s behalf, and what would another provider need before it could do the same?

They don’t check your work

Customers let your work drive the next step without rechecking everything. They have learned that you get it right, flag exceptions, and help put things right when something fails.

Gusto calculates payroll taxes, deducts them, and files the relevant forms. Customers still supply and approve information, but they buy the service so they do not have to recreate every calculation themselves. A replacement has to earn their confidence before the next payroll run.

Avalara calculates sales tax using the transaction’s location, product type, and applicable exemptions. Its value includes taking that detailed work off the customer’s hands. The trust to rely on those calculations has to be earned through delivery.

Trust can also mean that the people who rely on your records already accept them—for example, the evidence an e-sign service preserves when an agreement is challenged.

Some work will always need review. Making that review faster and more dependable can be valuable too. The advantage is the confidence you have earned in a specific job, backed by how you handle failures.

Ask yourself: What do customers trust you to get right without checking it all themselves?

Someone official already said yes

Some businesses need a license, accreditation, or other outside approval before they can deliver what the customer is buying. Faster software development does not automatically shorten that process.

Sidekick Learning is an authorized continuing-education provider through the Behavior Analyst Certification Board. Its customers can earn continuing-education credits. Building a similar training portal does not give a competitor that provider authorization.

Avalara is an approved Certified Service Provider under the Streamlined Sales Tax program. That status qualifies it to provide services funded by participating states for eligible sellers. Copying its software would not reproduce the approval or access to that program.

Be specific about what you hold. Selling to regulated customers does not mean you own a hard-to-get approval. The useful advantage is an approval customers need and competitors would have to spend time and effort earning.

Ask yourself: What approval do you hold that a competitor would have to earn?

Everyone’s already in it

One person can change tools fairly easily. A team needs to agree and move its work. A school or community may need to bring along people who do not even work together.

Slack connects colleagues and, through Slack Connect, people at customers and partners. One person preferring a different app does not move those conversations. Switching can require agreement and participation from people across several organizations.

Commit Swimming connects coaches, swimmers, and parents around registrations, billing, rosters, and team messages. A club changing systems has to bring its working community along. The same applies to a school app with students and parents already enrolled and using it.

Sometimes the other participants are part of the value: buyers join because the suppliers they need are there, and suppliers join to reach those buyers.

An active customer base can still be valuable when customers use the product independently: adoption, habits, and relationships take time to build. Focus on actual use and how much effort a move requires.

Ask yourself: Who would have to agree, move, or reconnect for a customer to replace you?

You get better as you do the work

Doing the work can teach you things competitors cannot see. You take an action, observe the result, and use what happened to do a better job next time. That learning can help at three levels.

One person. An assistant learns someone’s priorities, preferences, and recurring corrections. It becomes more useful and harder for that person to replace.

One company. A scheduling worker learns which technicians suit which jobs, how dispatchers handle exceptions, and what particular customers need. Those lessons improve work across the account.

Across customers. Experience across businesses improves the shared service. Existing customers benefit from each other’s experience, and new customers start ahead.

The more widely useful lessons travel, the more customers benefit. But detailed knowledge of one company can be worth more than generic lessons shared widely. The advantage comes from better results and experience a replacement would need to acquire.

Stripe Radar evaluates payments using patterns across Stripe’s network and feedback on transactions identified as fraudulent. A merchant benefits from experience far beyond its own payments. A newcomer can reproduce a risk-scoring interface without having that history to draw on.

Churnkey says its Intelligence products are trained on accumulated cancellation sessions and billing events. Its Adaptive Offers tests which offers persuade customers to stay. The opportunity comes from seeing how people respond and using that experience to improve retention decisions.

Learning within an account can help you keep it. Learning across accounts can also help you win and serve the next one. Where you have permission to learn across customers, look for useful lessons you can retain even when one leaves. You do not need to train a foundation model: better questions, rules, and decisions can be enough. The Offense chapter shows how to build this loop into the work.

Useful data can also be valuable before there is a loop. DealForma maintains an analyst-curated database of biopharma deals, terms, and company histories for benchmarking and research. Recreating the search screen does not recreate the researched dataset. Exclusive access, careful collection, or historical depth can matter in their own right. You still need the right to use the information and evidence that it helps.

Ask yourself: What does experience teach you to do better for a person or company, and how much of that could help other customers?

It’s hard to keep running

Some services work in a world that keeps changing. Websites block scrapers. Email providers change how they filter messages. New attacks appear. Customers pay for someone to keep dealing with that change.

ScrapingBee handles browser rendering, rotating proxies, and anti-bot problems for customers. Building a scraper that works today is one task. Keeping requests working as websites and blocking systems change is an ongoing service.

Drip handles email authentication, sending infrastructure, and bounced messages. A campaign editor is only useful if messages reach people. Customers are also buying the work of maintaining deliverability as inbox providers and sending conditions change.

Some of that ongoing work can be shared across many customers. Spreading the cost of deliverability expertise or proxy infrastructure can let you offer a dependable service for less than customers would spend maintaining it themselves.

The advantage grows when experience lets you solve recurring problems faster, more reliably, or at lower cost than a newcomer. Maintenance being expensive is not enough; you need to be good at the work customers want off their hands.

Ask yourself: Is your product hard to build once, or hard to keep working—and what makes you better at the latter?

Choose what to strengthen

Look for advantages you can point to: partners who send business, permissions customers have granted, outcomes your experience improves. The categories overlap, so the aim is to find your strongest assets rather than collect eight labels.

Compare those assets with the alternatives from the Supply chapter. A new startup might need years to build your relationships; an existing platform might already have them. Ask whether the advantage helps you win customers, keep them, charge more, or deliver better results at lower cost.

Existing customers can give a new service a head start. What will make them keep choosing it once competitors can deliver similar work?

Then choose what to build next. Could you turn trusted advice into permission to act? Introduce another service to customers who already rely on you? Capture the outcome data you currently miss? Use cheaper development to make those advantages stronger while doing more for customers.

Offense: What can you profitably become?

AI expands what you can profitably deliver to customers. Work that once needed a dedicated employee or an expensive custom implementation may now fit inside your offering. Your current product and pricing do not define the limits of the business you can build.

Use the demand, supply, and asset assessments to spot those opportunities. You already know customers, understand parts of their operations, and connect to systems they use. What more could you do for them now?

Find the work customers want done

Look around your product before looking at your feature backlog.

What do customers still pay people to do? Follow the work before and after someone uses your software. A dashboard of overdue invoices sits inside a larger job: contacting customers, handling replies, and getting paid. Some of that work could become part of your offer.

What valuable work gets skipped? Missed calls, unanswered quotes, and neglected accounts can represent demand nobody is serving. Helping customers capture that value can create a new budget, even where there is no employee to replace.

Who has been too expensive to serve? Customers who need extensive setup, customization, or ongoing help may become viable as delivery costs fall. Revisit the deals you declined and the segments you ruled out.

Ask customers to show you the work. Find where they spend money, lose revenue, or give up on getting something done.

Consider three practical moves

Improve the existing business. Use AI to make the product more useful and easier to adopt. Faster configuration and better support could make smaller customers economical to serve. Customer-specific functionality could help you win a vertical you previously could not support. Tie the improvement to something concrete: a new customer segment, a larger account, or lower delivery costs.

Provide something agents need. Make a valuable capability available wherever agents do work. An e-sign provider could become the signing service customers authorize their agents to use across purchasing, onboarding, and other workflows. That creates additional routes for transactions through the business. Existing approvals and integrations can help win that use. An API or an MCP server delivers the capability; the opportunity comes from what customers need it for.

Become the worker. Take responsibility for an outcome your software currently helps people produce. Instead of selling a tool for following up on invoices, offer to run collections: contact customers, handle replies, follow up, and track payment. This can open spending currently allocated to employees or outside providers. Start with work you understand, with authority to act and a way to handle exceptions. Expand the work you take on as you learn to deliver it reliably.

These moves can reinforce one another. Choose the one that gives you the strongest first offer.

Give customers more to buy

Our scheduling vendor could begin with after-hours booking and grow into other work that keeps technicians productively booked:

Customer problem	Offer	What delivery could teach you
Calls go unanswered after hours	Answer, qualify, and book those jobs	Which questions and eligibility rules prevent failed bookings
Cancellations leave technicians idle	Fill openings from a suitable waitlist	Which customers respond to an opening and how much notice they need
Quotes never receive follow-up	Follow up and convert outstanding quotes into bookings	Which timing and messages work for different kinds of jobs

Each is a specific offer a customer can buy. Together, they suggest a larger business than scheduling software alone. A small contractor that could never afford a receptionist might now buy booking coverage; an existing customer might pay for several services that bring in more work.

If you become the worker, build the outcome loop

You already have a learning loop: watch customers use the product, learn what needs improving, build it, and ship it to everyone. AI coding tools help you move through that loop faster.

When your product does the work, you can also build learning into how it operates: take an action, observe the result, and use what happened to improve the next decision. That is the outcome loop.

For the scheduling vendor, a calendar entry is only the beginning. Did the customer keep the appointment? Was the right technician sent? Was the job completed? If a dispatcher intervened, what information or rule was missing?

Suppose failed visits reveal that certain jobs require equipment the service never asks about. Add that question before booking and use the answer to choose a suitable technician. Check whether repeat visits fall. The lesson has changed how the work gets done. Parts of the loop can run automatically: the service could adjust whom it contacts first as it observes which waitlisted customers accept and keep short-notice appointments.

Learning can change when the service follows up, which work it prioritizes, or what it escalates. It does not require the product to rewrite itself or train a foundation model. Start by capturing outcomes and corrections; use them to improve decisions, then automate the parts of that process that work reliably.

Choose work where you can see what happens afterward. If another system records the completed job or payment, connect that feedback. Real outcomes reveal things a demonstration cannot, and some take time to arrive. More activity only becomes useful experience when it changes what you do and improves results.

Build from what you learn

Start with learning that improves the work for the person or company you already serve. A scheduling worker might learn which customers accept short-notice appointments, or which jobs consistently need more time with a particular team. That knowledge can make the service more useful and harder to replace within the account.

Then ask how much of that learning could help elsewhere. If asking an equipment question prevents wasted visits for one plumbing company, a similar company could benefit from that question from its first booking. Keep customer-specific details local and use shared lessons where you have permission and evidence that they help: a practice that works for emergency plumbing may not work for planned installations.

When useful learning travels across customers, growth can strengthen delivery. New customers bring more experience, and later customers get a better starting point. At any level, the advantage depends on outcomes and corrections that improve results and that competitors cannot readily obtain. Better models can help you use that experience and take on more work.

Whichever move you choose, build the moats

Use cheaper development to turn a more useful offering into a stronger business. When code and interface design are easy to reproduce, favor improvements that also establish customer relationships, active adoption, operational connections, or permission to do more.

A school app could let its existing network of parents pay for clubs and school trips. That solves another problem, creates transaction revenue, and connects the app to more of the school's daily operations. An e-sign provider can earn authorization for additional workflows. Neither needs to become a worker to grow or strengthen its position.

Use the asset map from the previous chapter to choose which advantages the new offer will build. For a worker, valuable outcome history may be one of them. For any of the three moves, ask what serving more customers will help you establish beyond additional features.

Get into the work

Pick one offer and put it in front of customers who already feel the problem. For the scheduling vendor, that could be paid after-hours booking for existing accounts. Name an owner and get it into use within 30 days. Handle exceptions manually where necessary and capture what they teach you.

Use delivery to find the next opportunity: another task customers want off their hands, another source of missed revenue, or a customer segment you can now serve. Build from what customers buy and what you learn doing the work.

Put the four answers together

The purpose of this assessment is to improve a decision about the business. You should now be able to explain the effect on demand, the strongest alternatives, the assets that matter, and the opportunity to pursue.

Resist collapsing those answers into a comforting average. Strong switching costs can coexist with declining demand. A large automation opportunity can coexist with weak economics. Current retention can look healthy while new sales become harder. These combinations call for different actions.

Write a short decision memo that covers five things:

1. Our biggest exposure: the revenue stream or economic assumption most vulnerable to change.
2. Our strongest advantage: the asset that helps us win against a named alternative, supported by evidence.
3. Our best opportunity: the offer we will put in customers’ hands and the advantage delivering it will build.
4. Our next action: the first customers, the person responsible, and a launch date.
5. What we will learn: the outcomes and corrections we will capture and how they will shape the next move.

For the hypothetical scheduling vendor, the memo might read:

Our exposure is declining demand for dispatcher seats, particularly where a platform bundles adequate scheduling. Our starting assets are customer relationships, configured service rules, and access to live schedules. The product lead will launch paid after-hours booking with existing customers within 30 days. Customers will authorize the service to book eligible jobs, with our team handling exceptions. We will follow bookings through job completion and capture cancellations, technician mismatches, and dispatcher corrections. That evidence will improve eligibility, scheduling, and escalation. We aim to earn responsibility for keeping technicians productively booked and accumulate experience a new entrant would need to acquire. Customer uptake and delivery results will guide which work we take on next.

That memo names the opportunity, the owner, and the next move. It also explains how doing the work will strengthen the business.

The right response will vary. A business with healthy demand and clear advantages may focus on execution and incremental automation. A business facing a credible substitute may need to change its offer or pricing. Some businesses are facing existential threats and need to pivot hard. Better to know that now than have the market show you.

You do not need certainty about AI’s eventual capabilities to act. You need to understand the assumptions supporting your current business, notice when those assumptions change, and use today’s assets to earn tomorrow’s revenue.

Build around something customers will continue to value. Know why they will buy it from you.