UnSaaSed.
Own the stack that makes revenue.

Count the logos in accounts payable. Then ask a harder question: which of those subscriptions would a sophisticated buyer pay a multiple for?

Most operators already know the answer. The stack is bloated. Data is trapped behind APIs. Demand dies between tools that do not talk. AI got added as another seat, another bot, another prompt pack. Revenue did not move. That is not a technology failure. It is an ownership failure.

UnSaaSing is the name we give the correction. Keep what is rented only where it is cheap and replaceable. Own the workflows that capture demand, convert it, and retain it. Those are the rails AI can actually run on. Those are the assets that show up at exit.

1. The leaky bucket

Plateaued revenue is usually a capture problem, not a demand problem. Leads arrive. Response is slow. Follow-up is uneven. Reactivation never happens. Handoffs between CRM, quoting, and billing drop records on the floor. Each tool is "working." The funnel is not.

Rented software hides this because nobody owns the full path. The CRM vendor is not accountable for revenue. The chatbot vendor is not accountable for close rate. The operator is accountable for both and cannot see either without a stack map.

We score stacks as Owned, Patched, or Leaky. Leaky means the revenue path depends on memory, Slack, and whoever is at their desk. Patched means zaps and workarounds that would break if a founder went on vacation. Owned means the sequence is readable, exportable, and transferable.

2. AI changed the math. It did not change the ownership question.

Agents are cheap. Coordination is not. Private equity is already acting on that. In August 2026, Clearlake Capital partnered with Google Cloud to put full-stack AI (infrastructure, data, agents, security) across portfolio companies. That is not ChatGPT seats. That is production-grade stack.

Mid-market operators are still on the other side of the trade: another SaaS logo, another bot, no process they own. Capability got cheaper. The bottleneck moved to orchestration: who decides what happens next when a lead arrives, a job finishes, or an exception fires.

If the workflow lives in a vendor, swapping the model breaks the company. If the workflow lives in a process you own, the model is a component. That is the UnSaaSed line. AI is leverage on businesses that own their rails, and a solvent on businesses that only rented the work.

3. Three layers (not a rip-and-replace)

Serious UnSaaSing does not start by throwing out the CRM or the industry system the team actually uses. It sits above them.

Co-existence

Keep systems of record that work. Add an orchestration layer that coordinates CRM, billing, ops, and agents. Humans enter at defined gates, with an audit trail.

Resilience

Ask: if this vendor triples price or kills an API, does revenue still run? If the founder is out for two weeks, does follow-up still happen? If the answer is no, you do not own an asset. You own a job with overhead.

Deployment

Put agents on owned workflows first: speed-to-lead, follow-up coverage, quoting, exception routing. Do not deploy agents on a leaky funnel and call it transformation.

	Rented stack	Owned stack
Logic	Trapped in zaps and vendor settings	Readable sequence above the tools
Data	Export is a project	Export is a button
AI	Point solutions, prompt packs	Agents on owned workflows
Exit	Expense line	Underwriteable infrastructure

4. The 90-day build

Traditional PE treated the first 18 months after close as a construction site. That is too slow when agents can be on a real workflow in days. The 90-day standard we use with operators:

Days 1-30. Map the revenue path. Score Owned / Patched / Leaky. Rank leaks in dollars, not opinions. Pick the first workflow to own (usually speed-to-lead or follow-up).

Days 31-60. Stand up the owned layer. Connect systems of record. Put one agent on one workflow with a human gate. Measure response time and coverage, not vanity AI metrics.

Days 61-90. Expand to the next leak. Kill or freeze redundant seats. Write the operating sequence so it survives the founder. You should be able to show a buyer what is owned.

This is not a six-month strategy offsite. If the first 30 days cannot name a dollar leak, stop. Do not buy more tools.

5. Build vs buy vs neither

The mid-market playbook is not "build everything." It is: own the sequence, rent the commodity.

• Buy when the category is mature, switching costs are low, and the tool does not hold your revenue logic.
• Build (or tightly control) when the workflow is how you make money, how you quote, how you retain, or how you underwrite.
• Neither when the real problem is process, pricing, or people. Software will not fix a mandate nobody follows.

Exit math is blunt. Subscriptions you rent are worth zero to a buyer. Workflows you own can be diligence, not a cleanup project. That is why UnSaaSing shows up in M&A conversations even when the original brief was "we need AI."

6. What we find in stack audits (composite)

Patterns repeat. Redundant seats. Core workflows living in a founder's inbox. Customer history split across three systems with no owner. API limits that make "AI" a demo. Zero documentation of what happens after a lead comes in. The $1,000 UnSaaSing audit exists because operators will not start on a retainer until this is in writing: stack map, leak list, top ROI moves, 90-day plan, readout. They keep the work product either way.

7. What this paper is not

It is not a claim that every company should fire its vendors. It is not a layoff thesis. Agents add capacity on owned rails. Headcount stories without process ownership just move the leak. It is not investment advice, a performance forecast, or an offer of securities.

The next decade will not be won by the operator with the most logos in the stack. It will be won by the operator who can show, in a data room, that demand still converts when the founder is not in the room.