At some point in the last decade, someone explained to you why leads and contacts need to be different objects.
You adopted it. Not because it matched how your business works — but because Salesforce required it. And then you spent the next five years working around it. Deduplicating records. Building mirrored custom fields. Paying consultants to manage complexity that the tool itself created.
I wrote recently about why the entire CRM category is architecturally wrong for small service businesses. But the problem isn’t unique to CRM. It’s the defining pattern of the SaaS era.
You didn’t buy a tool that fit your business. You bought a tool and reshaped your business to fit it. HubSpot’s pipeline stages. Marketo’s program structures. Monday.com’s board logic. Each one came with a philosophy about how work should flow, and you adopted that philosophy — not because it matched yours, but because the tool required it.
We had a name for this in corporate America: vendor lock-in. And for a long time, we accepted it as a reasonable trade-off. Custom software was expensive. Building your own CRM or marketing automation platform meant hiring developers, managing infrastructure, maintaining code. The SaaS bargain — adopt our workflow, and we’ll handle the rest — was genuinely better than the alternative.
That bargain expired. Most people haven’t noticed.
The Pattern Repeating, in New Packaging
$285 billion in SaaS market value evaporated in early 2026. Analysts called it the “SaaSpocalypse.” Investors realized that when AI agents can execute workflows autonomously, you don’t need as many software seats. Atlassian reported its first-ever decline in enterprise seat counts. Salesforce shares dropped 26%.
The industry narrative says: AI agents are disrupting SaaS.
Here’s what I think is actually happening: AI agent products are becoming the next generation of SaaS. Same conformity, new label.
Look at how these products are marketed. “$3K/month — replaces a $15K/month SDR hire.” “An AI marketer that works 24/7.” The framing is deliberately against headcount costs, positioning the product as an employee. But what you’re actually buying is a playbook. The vendor’s playbook. Their prospecting sequence. Their email cadence. Their definition of what “qualified” means.
The moment your business needs something the playbook doesn’t cover — a prospect who responded on a channel the AI doesn’t monitor, a deal structure that doesn’t fit the standard scoring model, a follow-up sequence that requires context from a conversation that happened offline — you hit a wall. The “employee” can’t adapt. The template can’t flex. You’re back to doing it yourself, except now you’re also paying $3K/month for the template.
Sound familiar? It should. It’s the same wall you hit when your CRM couldn’t handle the way your team actually tracks relationships, or when your marketing automation couldn’t accommodate the approval process your clients required. The tool works beautifully — for the workflow it was designed for. Your workflow is always slightly different.
The Conformity Tax — And Why It Used to Be Worth Paying
I lived in this era. I spent years in corporate America implementing enterprise technology — Eloqua, Salesforce, Wrike. I know the conformity tax intimately because I helped companies pay it. And for a long time, it was a rational trade-off: custom software was expensive, building your own was prohibitive, and adopting someone else’s workflow was genuinely better than the alternative.
That trade-off made economic sense in 2018. Here’s why it doesn’t anymore.
Last year, a client needed to connect millions in annual ad spend to actual revenue — multi-touch attribution with real-time reporting. The obvious move: customize the existing platform further. More fields. More reports. More consultants. More complexity layered onto a data model already straining under years of workarounds.
Instead, we built an event-based data architecture from scratch. Every customer touchpoint stored as a transaction, chronologically, designed around how the business actually works. Not the append-only, pipeline-shaped data model that requires gymnastics to answer basic questions. Just events, in order, queryable without workarounds.
Total development team: zero traditional developers.
Two years ago, that project would have required three developers and six months. Today, a solo operator with domain expertise can ship production-grade, custom systems in weeks. The bottleneck moved. It’s no longer “can we build it?” It’s “do we know what to build?”
That question is where the real advantage lives. Not in tools. Not in vendor templates. In the years you’ve spent learning how your customers actually behave, how your team actually works, and what your data actually needs to look like.
So Why Are We Still Buying Conformity?
Everyone’s debating whether AI agents will replace SaaS. That’s the wrong conversation.
The right one: now that custom is accessible, why is anyone still paying the conformity tax?
When an AI agent product says “I’ll be your SDR for $3K/month,” it assumes you need to buy this capability pre-packaged. That assumption was valid when building your own was impossible. It’s increasingly absurd when building your own — tailored to your exact sales process, connected to your exact data sources, integrated with your exact workflow — costs less than the annual subscription and actually does what you need.
This isn’t an argument against all software. It’s an argument against accepting someone else’s workflow as the price of admission — when the price of building your own just dropped by an order of magnitude.
What This Means for You This Week
Before you evaluate any AI agent product — the AI SDR, the AI marketer, the AI customer service rep — ask one question:
Am I buying a capability, or am I buying someone else’s idea of how that capability should work?
If the product can only execute the vendor’s playbook, you’re paying a conformity tax. And unlike 2018, you’re paying it voluntarily.
I spent a decade helping companies adopt vendor platforms. I watched smart operators reshape their businesses to fit tools that weren’t designed for them — and then hire consultants to manage the gap. I’m not doing that to my clients. Not when the same decade of business experience that made those workarounds necessary is now the thing that makes custom builds possible.
Where This Gets Practical
If you’re evaluating AI agent products and aren’t sure whether you’re buying capability or conformity — or if you’ve already bought one and are starting to feel the walls close in — that’s the work I do. No pitches, just an honest assessment of whether off-the-shelf actually fits, or whether the build-vs-buy math has changed for your specific situation.
Want more like this? I write about AI adoption for people who run things — not hype, not doom, just what’s actually working. Subscribe here.
Sources:
- Forbes, “SaaS-Pocalypse Has Begun” (February 2026) — $285B+ market cap evaporation from SaaS stocks
- TechCrunch, “What’s Driving the SaaSpocalypse” (March 2026)
- Deloitte, “SaaS Meets AI Agents” (2026 TMT Predictions) — Predicted shift to outcome-based pricing models
- Pallas Advisory, “You Don’t Have a CRM Problem. You Have a Client Memory Problem.“
- Pallas Advisory, “When Everyone Has the Same AI, Your Business Experience Becomes the Edge“
- Pallas Advisory, “PwC Says Technology Is Only 20% of AI’s Value“
