System of Action — Part 2: Replacing Cousin Richie

Recap: System of Action — Part 1: Hero Users

The race to become the System of Action is on, and the stakes couldn’t be higher. Vertical SaaS incumbents are facing a serious challenge from native AI upstarts. As I laid out in Part 1, the VSaaS winners of the future won’t simply be tools to run the business; they’ll do the work. E.g., A legaltech won’t just help run the practice, but it will also help practice the law. (See how our portfolio company Clio is leading the way on this.)

To become the System of Action, a Native AI needs a wedge. In Part 1, I described creating a wedge by targeting the Hero User who has agency to choose their own software. To win hearts and minds by helping the Hero User do their good work, and automating away the administrative work. And to engage that Hero User with PLG tactics — easy to try, easy to buy, to gain enough user adoption to force integration with the Control Point.

‍

But what if there is no Hero User who can choose their own software? What if the person doing the work is stuck using tools chosen by someone else? 

‍

Most workers don’t get to choose their tools. No matter how good the chef, if he doesn’t own the restaurant, he’s not picking the POS. The superstar insurance rep crushing her quotas every month is not picking the CRM.

‍
So, how does a Native AI dislodge the Control Point incumbent when there is no Hero User? You’re back to selling to the business owner, solving “Stick Jobs, handled by Cousin Richie”.

‍

Let’s dig in.

‍

Start with good old-fashioned product management: Find the Sticky Jobs

When the buyer is the business owner (or function owner in an enterprise), the bar is high. Automating admin work with whizbang tech is interesting, but it won’t rise to the top of their to-do list.

‍

Start with good old-fashioned product management — know your customer and their problems. Your best shot at winning a spot in the business owner’s budget is by solving their hero problems: high-ROI, high-impact problems that cause ulcers and sleepless nights. They kill deals, make customers mad, or land you in hot water with regulators. And the closer the pain is to revenue, the better. 

‍

(Want to go deeper on this kind of product management thinking? Start with Tidemark friend Gokul Rajaram, who is one of the smartest minds in the space and has it well-covered.)  

‍

Sticky Jobs

Many Sticky Jobs are must-dos that get done late, badly, or not at all. Think: answering the phone during a chaotic lunch rush; scrambling to find a last-minute factory worker to keep a $100M line running; or filing compliance reports on time. 

‍

Why the breakdown? Because Sticky Jobs are manual, complex, and co-dependent. Solving them requires coordination across systems, people, and external stakeholders, as well as making tough decisions under pressure. One missed call, one blank field, one misfield report, and everything falls apart. 

‍

Sometimes there’s a lack of expertise. Sometimes the right information just isn’t there. And especially for small businesses, there just isn’t the time. These are high-hassle, low-reward jobs that no one wants to do. The commission is too low, and the juice is rarely worth the squeeze.

‍

Look for “Cousin Richie”

Some of the best problems to target are what I call “Cousin Richie” problems. Every owner has an idiot Cousin Richie, who works at the shop and they can take a number of forms. The offshore call center, the freelance bookkeeper, or that local marketing agency turning out low-value newsletters. 

If any of the tasks you identified above — setting sales appointments, renewing accounts with tiny customers — are in Cousin Richie’s hands, guess what? You’re in luck!  

‍

Why? Because AI runs circles around Cousin Richie.

AI never shows up late, forgets his login, or spends half the day vaping out back. AI works 24/7, speaks 40 languages, and is infinitely coachable. It never forgets, never loses its cool, and is always 100% focused on whatever you want it to be. 

‍

And because your AI runs circles around Cousin Richie, you should be able to charge a price that is a fraction of the cost of Cousin Richie, but still 2-5x the cost of the Control Point software.

Tap into where Gen-AI shines (circa 2025)

A lot has been written about the transformational capabilities of AI. As of July 2025 (and this will expand every month), here are the three I’d focus on.

‍

The first is the rise of the conversational user interface. Most apps assume users already know what they want. They’re built around keywords, not conversations. But a Native AI introduces an iterative UI that lets users explore naturally, without needing to know exactly how to ask. And because it remembers what came before, each interaction builds on the last. The result is a more fluid and collaborative dynamic between human and software. It’s less about finding and more about discovering.

‍

Second is the ability to enrich and enhance the backchannels of workplace communication: phone calls, email, PDFs, and faxes (yes, still). They’re messy, opaque, and non-structured. But a lot of real work still flows through them. They defy tracking, resist organization, and overwhelm most systems that try. GenAI doesn’t. It can parse them, enrich them, and turn them into structured, actionable inputs. No more digging through endless email threads or combing PDFs for the right number. AI finds it, then nudges you toward your next task or quietly does it for you. 

‍

Third and perhaps the most profound shift is the ability of GenAI to take non-deterministic actions. It doesn’t need every rule hard-coded in advance. That’s a sharp break from decades of brittle enterprise software, where workflows follow logic trees and pristine input leads to predictable output. Real businesses don’t work that way. The edge case is the case. And GenAI thrives in this unpredictable, ambiguous mess. It can navigate toward the right outcome even when the path isn’t mapped out.

‍

Yes, this introduces risk. AI will get things wrong, which is why we build guardrails. But the upside is speed. When the clock is ticking and the options are many, the ability to act on a best guess, learn fast, and improve every time may be the only way forward.

‍

Sticky Jobs in the World

Once you start looking for Sticky Jobs — Cousin Richie or no —  you’ll see them everywhere. These are the tasks that don’t live neatly in one role or system. They require coordination, negotiation, iteration, and constant judgment calls. You’ll find them buried in invoices, contracts, design sketches, inboxes, voicemails, and yes, still occasional faxes. They’re the problems that get “talked out” on the phone. They’re just hard to automate because they cross teams, departments, and even companies. They’re always moving, always messy, and never resolved the same way twice. 

‍

Here are some obvious places to look:

‍

• Phone-based communication: They suck and should be reserved for critical, time-sensitive matters. It’s a great place to start, and our friends at Euclid have laid out a great thesis here. Any phone-initiated process warrants a hard look. 
• Complex orders, RFPs, RFQs, renewals, change orders, or multi-party quoting.  This requires producing, reading, understanding, redlining, and managing version control on 100-page documents. Humans are bad at this; AIs are purpose-built. 
• Fragmented workflows or supply chains with unpredictable exceptions or multiple layers of approvals. 
• Mo’ people, mo’ hassle: Any endeavour that involves multiple people or stakeholders.

‍

Native AI can transform the experience. It can interpret messy inputs, track context across conversations, and tailor (or automate) responses based on who’s asking and the kind of answer they need. It drafts messages, flags gaps, and routes approvals, not by following rules, but by adapting to changing conditions. Set parameters, and it can negotiate on your behalf, in full compliance to deal size and customer tier, while staying in sync with internal signoff workflows.

‍

AI doesn’t replace the person, at least not at first. It rides shotgun, nudging, guiding, and unblocking. All the while, it’s learning context, cadence, and the informal real-world rules concerning how work gets done. 

‍

Then there’s also work that never gets done at all. This is GenAI’s stealth power: tackling jobs too expensive or too tedious for humans. In smaller businesses, these critical tasks are the first to fall through the cracks. Calls go unanswered, pricing gets stale, and supplier decisions default to “whatever’s easiest.” When the pain becomes too great, they get outsourced to marketing agencies, procurement firms, lawyers, accountants, and various consultants, all billing $400 an hour. They’re perfect targets for agentic solutions. In this world, agents become the new agencies.

‍

Value Road Map

On the path to supremacy, expect a fight. The incumbent won’t go quietly. They’ll launch a competing product at a fraction of the price, mimic your UI, clone your workflow, and poach your employees. It’s not bad behavior. The stakes are survival, and it’s exactly what they should do. 

‍

Customers will be tempted by their offer. Who wouldn’t choose to stick with their existing control point that is fully integrated with all their workflow, data, and account ownership gravity? When inertia feels seamless, switching seems risky. 

‍

That’s why you need a Value Roadmap, one that puts real distance between you and the incumbent. While they’re busy copying your first product, you’re shipping the second and building a third. While they match your features, you’re expanding into new capabilities, new personas, and deepening your strategic relevance.

‍

This isn’t just a product roadmap: it’s a value strategy tied to your customer’s org chart. Here’s how you do it.  On one axis, lay out the “hero jobs” that are adjacent to the job your initial product serves. Likewise, on the other axis, put adjacent personas to the target persona that your initial product serves. 

‍

‍

Each new product either does an adjacent job, serves an adjacent persona, or ideally both. When you consider the next new product, think through the frameworks we’ve discussed in the past (see Control Point Patterns 2024, Platforms of Compounding Greatness).  

‍

What you’re laying out is the roadmap to how you’re going to extend your product portfolio to do more jobs and to add more value to more people in the organization. That’s the key to locking yourself in against upcoming counterattacks by the Control Points. This is particularly important in enterprises, where success depends on serving many stakeholders and touching many roles. 

‍

From Wedge to Supremacy

When you solve a Stick Job, you win the business owner. That’s your wedge. You’ve earned the right to integrate and surround. You start by tying into workflows, gathering context, and making yourself indispensable. That’s how a Native AI becomes the System of Action. The real power comes next.  

‍

As the System of Action, users naturally want to feed you more context, more data, more touchpoints, because it leads to better outcomes. You now have usurped the Control Point’s data gravity by “supersetting” its data.

‍

With more and more data, it’s natural that the Native AI challenger can trigger more and more actions and become an orchestrator of workflows.

‍

‍

A true System of Action is also a learning machine. It watches inputs, actions, and outcomes, and how people adapt. If you own the engagement layer, you see it all: The formal workflows and the informal stuff buried in Slack threads, hallway chats, and one-off spreadsheets. And once you see that, you can start to codify it, optimize it, and even automate it. A System of Action can affect:

‍

• Process discovery: You can capture and codify digitally how work gets done.
• Process understanding: You start to identify and understand bottlenecks, root causes, and recurring exceptions. 
• Process conformance: You see how the actual process works versus the way you think it should be done, and decide if you want to make changes.
• Continuous learning/Process enhancement: Your outcomes improve via suggestions, nudges, or automation.
• Autonomous optimization and execution: The system becomes an intelligent action engine.  

‍

Coordinate action across external stakeholders, and an agentic System of Action can “extend through the value chain.” First, the agents may coordinate communications between stakeholders — buyers, distributors, and suppliers. Then it may collect, reconcile, and aggregate data from different stakeholders and become a de facto industry ledger. Once these agents start triggering workflows across stakeholders, your system becomes the platform on which the industry runs. 

‍

Part 3 System of Action - the incumbent Control Point’s counter strategy - is up next.