Business Model Analysis

Context

The CRM compliance demo has validated that Condelo works as a platform for building custom intelligence apps. The model going forward: we build custom apps on top of Condelo for clients, charging a build fee, a platform license, and a 30% cut of revenue to cover AI/LLM costs.

What You're Proposing

Strengths of This Model

• Low barrier to entry: Clients pay for the build, you own the platform IP
• Compounding revenue: Each new client adds recurring + rev-share income
• Proven concept: CRM compliance demo validates the build pattern
• AI cost pass-through: 30% cut covers variable LLM costs and creates margin
• Platform leverage: Each app build strengthens the platform (shared components, patterns, bug fixes)

The Revenue Model: You Control the Payment Rail

How it actually works (clarified):

1. You build the app for a 3rd party (the "app partner")
2. End users of the app pay Condelo (you handle all billing/subscriptions)
3. You take 30% to cover platform costs, LLM costs, and profit
4. You pay out 70% to the app partner's bank account

This is a marketplace/payment-rail model — like the App Store, Shopify, or Stripe Connect. The key insight: you control the money flow, which means:

• No audit problems — you see every transaction because you process them
• No trust issues — the partner sees their revenue dashboard, you see yours
• Built-in enforcement — the 30% cut happens automatically at payment time, not as a separate invoice
• Transparent — both sides see the same numbers

Pricing Model Options (Evaluated)

The question is: how do end users pay, and how does the 30/70 split work?

Option A: 30% of Gross Revenue (Flat Commission)

How it works: Whatever end users pay for the app, you take 30%, partner gets 70%. Simple.

End user pays £100/mo subscription
  → £30 to Condelo (platform + AI costs + profit)
  → £70 to App Partner

Risks:

• Your costs aren't proportional to revenue. A £50/mo app and a £500/mo app might use the same amount of LLM tokens, but you take 10x more from the expensive one. This is fine for YOU but partners building premium apps may resent it.
• Comparison to Shopify (0.5-2%), Apple (15-30%), Stripe (2.9%). 30% is at the top end. Justifiable because you provide the entire AI platform (not just payment processing), but you'll need to defend it.
• Model works best when: AI costs are a significant % of the subscription price. If LLM costs are £20/mo and the app charges £100/mo, your 30% (£30) covers costs + profit. If the app charges £1,000/mo, your 30% (£300) is mostly pure profit — partner will push back.

When to use: Best for apps where AI is the core value and costs scale with usage.

Option B: 30% of AI/Platform Costs + Fixed Fee (Cost-Plus)

How it works: You charge a transparent markup on actual platform resource consumption, plus a small fixed platform fee. Separate from what the partner charges end users.

Monthly platform costs for the app: £200 (LLM + storage + compute)
  → Partner pays: £200 × 1.3 = £260 (30% markup on costs)
  → Plus: £300/mo platform fee
  → Total to Condelo: £560/mo
  → Partner keeps: 100% of end-user revenue minus the £560

Risk: If you're NOT on the payment rail for end users, you're back to invoicing the partner directly. Loses the "built-in enforcement" advantage.

Option C: Tiered Commission (Graduated Split) — RECOMMENDED

How it works: 30% split on end-user revenue, but the rate decreases at higher volumes. You still control the payment rail.

Example: App generates £30K/mo
  → First £5K:  30% = £1,500 to Condelo
  → Next £15K:  25% = £3,750 to Condelo
  → Next £10K:  20% = £2,000 to Condelo
  → Total: £7,250 to Condelo (effective rate: 24.2%)
  → Partner gets: £22,750

Why this is the best fit:

• Addresses the fairness concern of flat 30% at high revenue
• Incentivizes partners to grow (they keep more as they scale)
• Your take ALWAYS exceeds your costs (even at 15%, £7,500/mo on £50K revenue covers any realistic LLM spend)
• Industry-standard pattern — easy to explain: "same as the App Store but better"
• Still simple: one rate card, automated split, no invoicing

Option D: Build Fee + Commission Hybrid

How it works: Charge a build fee upfront, then take a lower commission (15-20%) because the build fee already recovers your initial investment.

Build fee: £25,000 (upfront)
Commission: 20% of end-user revenue (ongoing)

Risk: The lower commission rate means you need more volume to cover ongoing LLM costs. If an app has low end-user revenue, the 20% may not cover your costs.

Recommendation: Option C (Graduated Commission) + Build Fee

Best of all worlds:

Why this works:

• Build fee covers your investment in building the custom app
• You handle all billing via Stripe Connect — end users see the app's brand, payments flow through your Stripe account, 70%+ auto-transfers to partner
• Both you and the partner see a real-time revenue dashboard
• No invoicing, no chasing payments, no audit disputes
• Partners are incentivized to grow (better rates at higher volumes)
• Your per-client revenue is always well above your per-client costs

Build Fee Guidance

The build fee is calculated as: weeks × weekly rate × complexity multiplier.

With a weekly rate of £2,500 over 5 weeks at 1.2× complexity, the build fee comes to £15,000.

• Fixed-price for defined MVP scope. T&M for iterations after launch.
• Include 2 weeks of post-launch support in the build fee.
• Any scope additions during build → change order at T&M rate.

Technical Implementation: Stripe Connect

Stripe Connect is purpose-built for this model. It handles:

• End-user subscription billing (on behalf of the app)
• Automatic revenue splitting (your 30% vs partner's 70%)
• Automated payouts to partner bank accounts
• Tax reporting (1099s, VAT, etc.)
• Dispute handling
• Dashboard for both you and the partner

Implementation effort: 2-3 weeks (Stripe Connect integration, subscription management, payout dashboard).

Cost: Stripe charges ~2.9% + 30p per transaction + 0.25% for Connect. This comes off the top before the 30/70 split, or you absorb it in your 30%.

LLM Cost Coverage Analysis

The key question: does 30% of end-user revenue actually cover your LLM/platform costs?

*At graduated rates, this would be ~£12,500 (effective 25%)

Conclusion: 30% is very healthy. Even at 15% (highest tier), your margins are strong because LLM costs don't scale linearly with end-user revenue. The riskiest scenario is a very cheap app (£100/mo) with heavy AI usage — but even then, your 30% (£30) likely covers the £5-15 in LLM costs.

Risks & Considerations

Legal & Contracts

• Service agreement template: Define SLA (uptime), data ownership, liability
• Data processing agreement (DPA): Required if handling client data, especially in UK/EU
• IP ownership: Clarify that the platform is yours, the client owns their data and custom app config
• Termination clause: What happens to client data when they leave? Export format, retention period

LLM Cost Volatility

• LLM pricing drops ~50% per year. Your margins will improve over time, but clients may expect price reductions
• Consider locking in pricing for 12 months at a time
• Use the cheapest model that works (GPT-4o-mini for most tasks, GPT-4o for complex reasoning)
• Cache aggressively — many queries are similar across clients in the same domain

Security

• Penetration testing: Before going live with paying clients, get at least a basic pentest
• SOC 2 / ISO 27001: Not needed immediately, but enterprise clients will ask. Start tracking your security practices now
• Encryption at rest: Postgres supports it, ensure it's enabled. Qdrant and MinIO too

Data Residency

• UK/EU clients may require data stays in UK/EU. Deploy infrastructure in EU region from day one
• This is much easier to do from the start than to migrate later

Operational Overhead

• With 2 people, every client adds support burden. Budget for ~2-4 hours/week per client for support/maintenance
• Automate everything you can: monitoring, alerting, backups, deployments
• Build a simple knowledge base/FAQ for common client questions

Client App Maintenance

• Each custom app creates a maintenance obligation. Client systems change, webhooks break, APIs update
• Factor ongoing maintenance into the platform license fee
• Consider: apps should be as thin as possible — business logic in the platform, not the app

Competitive Positioning

• You're not selling "RAG" — you're selling "AI-powered intelligence for your domain"
• Example: the CRM compliance app's value prop is "automated compliance monitoring that catches violations in real-time" — that's delivered through the client's own stack (BFF + frontend), powered by Condelo's AI platform underneath
• Each client app has a different value proposition. The platform is invisible to end users

Revenue Examples (Build Fee + Commission Model)

Example 1: CRM Compliance Monitor

The app partner charges their end users £200/mo per seat for compliance monitoring.

Example 2: Legal Document Review

Partner charges law firms £500/mo for document analysis.

Example 3: Research Intelligence Dashboard

Partner charges £50/mo per researcher.

Portfolio Revenue Projections (Conservative)

With 5 app partners at an average build fee of £25,000 and average commission of £1,400/mo:

The compounding effect is powerful: Build fees front-load revenue, commissions compound. Adjust the side panel inputs to model different scenarios.

Break-even: Infra costs ~£150-200/mo initially. You're profitable from day one on recurring revenue. Build fees cover your time investment from the first client.