Buy or Build Software
Who owns your data — and what happens when you need to leave?
In the age of AI coding, the build option just got cheaper. But the real question isn't cost—it's sovereignty.
The Decision Matrix
Two axes determine your choice:
DATA SOVEREIGNTY
Low ───────────► High
┌────────────────┬────────────────┐
Commodity │ BUY │ BUILD LATER │
│ (SaaS OK) │ (Own the data)│
STRATEGIC ├────────────────┼────────────────┤
VALUE │ HYBRID │ BUILD │
Core │ (Buy + Bridge) │ (Own it all) │
└────────────────┴────────────────┘
| Quadrant | When | Examples |
|---|---|---|
| BUY | Commodity function, low data sensitivity | Email, calendar, basic CRM |
| BUILD LATER | Commodity now, but data becomes strategic | Analytics, customer insights |
| HYBRID | Core function, need speed, plan to migrate | MVP with SaaS, bridge to owned |
| BUILD | Strategic function, high data sensitivity | Proprietary workflows, AI training data |
The AI Impact
AI coding changes the calculus:
| Factor | Before AI | With AI |
|---|---|---|
| Build cost | High (months of dev time) | Low (days to weeks) |
| Maintenance | Full team required | AI-assisted updates |
| Speed to market | Buy wins | Build competitive |
| Data ownership | Locked in SaaS | Yours by default |
But: AI doesn't eliminate complexity—it shifts it. Maintenance, security, and integration still require judgment.
The Crypto-Enabled Option
What if there's a third path—neither pure SaaS nor pure build?
The Trust Stack Changes Everything
Traditional SaaS: You generate data → they store it → you pay to access it → switching costs lock you in.
Crypto-enabled infrastructure inverts this:
| Primitive | Traditional SaaS | Crypto-Enabled |
|---|---|---|
| Data Sovereignty | Vendor owns your data | You own objects explicitly |
| Truth | Trust the vendor | Verify on-chain |
| Identity | Platform credentials | Portable identity (zkLogin) |
| Trust | Institutional reputation | Cryptographic proof |
| Portability | Export friction, lock-in | Open by default |
The insight: You can BUY the workflow (SaaS UX) while OWNING the data (on-chain objects). Best of both worlds.
Data Flow Properties
Before buying any tool, apply the Clean, Fast, Open test:
| Property | Question | Red Flag |
|---|---|---|
| Clean | Single source of truth? | Copy-paste between systems |
| Fast | Real-time sync? | Batch jobs, overnight sync |
| Open | Can you export ALL data? | Proprietary formats, no API |
If you can't check all boxes, you're accepting lock-in risk.
The Framework
Step 0: Tech Review
Before running this framework, verify the tool deserves evaluation at all.
Tech review anchors the decision to a feature-matrix gap. Without a named feature ID and a JTBD statement, buy-or-build analysis is premature — you're scoring a solution before confirming the problem.
Run /tech-review [tool-name] to generate the gap anchor, candidate scores, and a preliminary verdict. Bring those outputs into Step 1.
Step 1: JTBD Analysis
What job is the software doing for your business?
- Functionality — What must it do?
- Data Management — What data does it touch? Who needs access?
- Simplicity — How easy to use/maintain?
- Speed — How fast must it respond?
See Products for JTBD methodology.
Step 2: Evaluate Options
| Question | Buy | Build | Crypto-Enabled |
|---|---|---|---|
| Features match needs? | Partial (generic) | Exact (custom) | Exact + portable |
| Total cost of ownership? | Recurring + switching | Upfront + maintenance | Upfront + low switching |
| Time to value? | Fast | Slow (faster with AI) | Medium |
| Data ownership? | Vendor | You | You (verified) |
| Integration flexibility? | Limited | Full | Protocol-based |
Step 3: Data Footprint Assessment
Data is the new oil. Answer these:
- Where do you collect unique data?
- What processes generate valuable insights?
- What domain secrets need protection?
- Could this data train AI that competes with you?
If data is strategic, own it. If commodity, rent is fine.
Step 4: Industry Vertical Check
Different industries have different sovereignty requirements:
| Vertical | Data Sensitivity | Recommendation |
|---|---|---|
| Healthcare | Very High (PII, PHI) | Build or crypto-enabled |
| Real Estate | High (transactions, valuations) | Crypto-enabled ideal |
| Finance | Very High (regulated) | Build with compliance |
| Gaming | Medium (player data) | Crypto-native (NFTs) |
| Supply Chain | High (provenance) | DePIN infrastructure |
See Vertical RaaS for industry-specific patterns.
The Hybrid Approach
Most businesses shouldn't choose pure buy OR pure build. The hybrid path:
PHASE 1: Buy (Speed)
↓
Use SaaS for fast market entry
↓
PHASE 2: Bridge (Data)
↓
Export/sync data to owned infrastructure
↓
PHASE 3: Build (Control)
↓
Replace commodity SaaS with owned systems
↓
PHASE 4: Crypto-Enable (Sovereignty)
↓
On-chain objects, verifiable data, portable identity
Key insight: Start with speed, end with sovereignty. Plan the migration from day one.
In-House Development with AI
If building, evaluate your AI capabilities:
AI Coding Maturity
| Level | Capability | Build Readiness |
|---|---|---|
| 1 | Code completion | Low—still need full team |
| 2 | Function generation | Medium—AI assists humans |
| 3 | Feature development | High—AI builds, humans review |
| 4 | Autonomous agents | Very High—AI maintains itself |
Costs and Timelines
| Factor | Traditional | AI-Assisted | AI-Native |
|---|---|---|---|
| Initial build | Months | Weeks | Days |
| Maintenance | Team | Individual + AI | AI + oversight |
| Iteration speed | Slow | Medium | Fast |
| Technical debt | Accumulates | Managed | AI-refactored |
Risks
- AI may not deliver full functionality—have fallback
- External dependencies (APIs, data) create fragility
- Compliance of AI-generated code needs auditing
- Model access and pricing can change
Decision Checklist
Before deciding:
- Is this function commodity or core to our differentiation?
- Does this data train competitive AI models?
- Can we export ALL data in standard formats?
- What are switching costs if vendor fails/pivots?
- Does our industry have special sovereignty requirements?
- Can we bridge to crypto-enabled infrastructure later?
- Do we have AI coding capability to build/maintain?
Applied Verdicts
How the framework maps to the commissioning dashboard. Each verdict links back to the Capability Landscape.
| Category | Verdict | Rationale |
|---|---|---|
| Identity + Access | Hybrid | Clerk today, own auth layer later — data sovereignty demands it |
| AI + Intelligence | Build (core) | Multi-agent orchestration IS the platform — edge > 7 |
| CRM | Build (core) | Data model IS the business — can't rent the thing you sell |
| Workflows + Automation | Buy + Bridge | n8n/Zapier for glue, own core workflows via Agent Platform |
| Analytics + Data | Hybrid | PostHog for product analytics, own BI for business metrics |
| Content + Marketing | Buy | Commodity — Midjourney, Arcads, Reel Farm rotate by quality |
| Billing + Commerce | Buy + Bridge | Stripe today, bridge to crypto-enabled settlement later |
| Notifications | Buy | Twilio/Resend until volume justifies owning the channel |
| Blockchain | Build (core) | On-chain settlement, smart contracts — can't outsource trust |
| UI Components | Build (core) | Shadcn + Radix — the interaction layer must match the brand |
The test: If the data it generates trains your competitive advantage, build it. If it's a commodity pipe, buy it. If it's both, bridge it.
Context
Decision Framework:
- Decisions — The P-C framework for choices
- Products — JTBD methodology
- RFP Process — End-to-end procurement lifecycle from need to sign-off
- Tech Platform — Business advisor's view of which tech decisions matter
Data & Sovereignty:
- Data Flow — Clean, Fast, Open + ABCD stack
- Sui Trust Stack — Crypto-enabled sovereignty
- DePIN — Physical infrastructure layer
Industry Patterns:
- Industries — Vertical-specific requirements
- Vertical RaaS — Industry software patterns
- AI Coding — Build capability
Links
- Klarna: End of SaaS — AI makes build viable
- a16z: The End of SaaS — Agents replace applications
Questions
If your data trains a model that could compete with you, what does that tell you about your buy verdict?
- Which "Buy" category in your stack would become "Build" the moment your business outgrows a single vendor's pricing tier?
- What is the actual switching cost when you move off a SaaS tool — and have you measured it against the savings from renting?
- At what point does the hybrid path (Buy → Bridge → Build) become more expensive than starting with Build?