'There Is No Product' — Why the Best Companies Sell Outcomes
Sidu Ponnappa's 342K-view thesis: AI is converting software from asset to inventory. A product is crystallized difficulty. Remove the difficulty, and you remove the product. Where is the line for yours?
Sidu Ponnappa (ex-SVP Engineering at Gojek, now co-founder of RealFast AI) published an essay in February 2026 with a 342K-view title: "There Is No Product."
His core thesis: The "SaaS is dead" debate is about the wrong thing. Everyone's arguing about pricing models and seat counts. The real question is simpler and more fundamental.
Is software still an asset? Or has it become inventory?
Asset vs Inventory
An asset is expensive and risky to build. Its output is scarce. It can be amortized across customers. It has a moat.
Inventory is cheap and trivial to build. Its output is abundant. It can't be amortized because anyone can manufacture their own. No moat.
AI is converting software from asset to inventory, class by class.
The Line
There's a line between inventory and assets. The line only moves up.
| Year | Below the Line (Inventory) | Above the Line (Asset) |
|---|---|---|
| 2024 | Boilerplate, autocomplete | Everything requiring judgment |
| 2025 | Components, integrations, auth, dashboards | Multi-system integration, domain expertise |
| 2026 | Full CRUD apps, HRMS, project trackers, analytics, internal tools | Compilers, OS, novel algorithms, domain data systems |
| 2027+ | ? | Shrinking |
What stays above the line:
- Compilers and operating systems
- Software encoding genuinely novel algorithms
- Software requiring years of accumulated domain data
- Cross-border taxation engines
- Real-time trading systems
- Climate modeling platforms
The difficulty isn't in the code. It's in the understanding the code encodes.
The TCO Crossover
Sidu's most practical argument: at some company size, building custom software is cheaper than paying SaaS per-seat fees.
SaaS economics: Per-seat cost scales with headcount. Add an employee, pay another license. The vendor's costs don't scale linearly with your headcount, but your bill does.
Custom build economics: Fixed annual maintenance cost. Adding users costs nothing. TCO per employee drops as you grow.
At some headcount, the curves cross. After that, custom wins and the gap widens.
His Gojek example: a $400K/year HRMS vendor quote. They built custom in 8 months (pre-AI, 2021). Today, with AI tools, the same build takes weeks at 1/5 the cost.
The crossover point is dropping fast. Companies that previously couldn't justify custom builds can now afford them. The SaaS vendor's per-seat pricing model accelerates its own replacement.
Thomson Reuters: The Overnight Cliff
When Anthropic shipped a legal research plugin for Claude in February 2026, Thomson Reuters, RELX, and Wolters Kluwer lost billions in market cap overnight.
Legal research crossed below the line in a single product launch.
The difficulty of searching legal databases was the product. When Claude can search, summarize, and cite legal precedent natively, the difficulty evaporates. The filing cabinet that was worth billions becomes a feature of a general-purpose AI tool.
The Bias Check
Sidu runs RealFast AI—a services company backed by Peak XV (Sequoia India) that sells Salesforce implementation and AI agent deployment. His essay conveniently concludes that companies should hire services firms instead of buying SaaS.
"SaaS is dead, build custom" is his sales pitch with academic framing. Read with that lens.
What matters regardless of bias: The framework is useful even if the conclusion is self-serving. The line is real. It does move up. Software below the line loses product status. Software above the line retains it. The question of where the line is for your specific product is the strategic decision.
Connecting the Frameworks
| Framework | What Sidu Adds |
|---|---|
| 8-to-80 | Domain expertise keeps software ABOVE the line |
| Deployment gap | Below-the-line software can be deployed by anyone; above-the-line requires expertise |
| Boris Cherny: "coding is solved" | Yes. That's why code is no longer the asset. The UNDERSTANDING encoded in the code is. |
| Gap-thinking | Below the line = crowded (everyone can build it). Above the line = gap (few can build it). |
What This Means for Builders
1. Test Your Product Against the Line
Can someone with Claude + n8n + a YouTube tutorial replicate your product in a weekend? If yes, you're below the line. You don't have a product. You have a feature that's about to be free.
2. The Understanding IS the Product
Code is inventory. Architecture decisions are assets. Domain knowledge encoded in working systems is an asset. A real-time trading system with 2,500 analyzed trading days of proprietary signal data is an asset. A chatbot connected to a CRM is inventory.
3. The Line Only Moves Up
What's above the line today may be below it in 18 months. Plan accordingly. Don't build a business assuming a permanent advantage when the water level rises on schedule.
Every model generation pushes the line higher. GPT-4 pushed it. Claude 3 pushed it. Claude 4 pushed it further. Each generation makes more software trivial to build, moving more products below the line.
4. Services Survive (For Now)
Even if SaaS products die, the need for custom implementation doesn't. Someone has to configure, integrate, customize, and maintain the custom builds. That's the services play.
Based on Sidu Ponnappa's "After AI, There Is No Product" (February 2026, 342K views). Bias-checked against RealFast AI's business model.