Cernio — Production Simulation

Version: 1.0 Date: 2026-03-25 Source: Hetzner VPS Plan, Strategy Docs (02, 04), Handbook (Ch. 189-210) Purpose: Simulate infrastructure, AI costs, and operational load at 10, 50, 200, and 1,000 users

1. Infrastructure Baseline

Current Architecture (Post-Migration to Hetzner)

Component	Server	Specs	Monthly Cost
Galata (Production)	Hetzner CX43	8 vCPU, 16GB RAM, 160GB NVMe	€12.80 (~$13.90)
Kadikoy (Worker/AI)	Hetzner CX33	4 vCPU, 8GB RAM, 80GB NVMe	€7.40 (~$8.03)
Snapshots	Both servers	Weekly automated	~€1.72 ($1.87)
Cloudflare	CDN + R2	Free tier	$0
Total infrastructure			~$23.80/mo

Prices reflect Hetzner post-April 2026 increase (+30-37%).

What Runs Where

Galata (Production):

Supabase (PostgreSQL + GoTrue Auth + PostgREST + Kong Gateway)
dose.com.tr (corporate site)
admin.dose.com.tr (admin panel)
Coolify orchestrator + Traefik
WireGuard VPN, Uptime Kuma, Vaultwarden

Kadikoy (Worker/AI):

Cernio Next.js application (SSR)
Python scraper
16 API routes (including AI-calling routes)
Batch scripts (finder, cleaner, bulk_fixer)

Scaling Decision: CPU vs Network Bound

Workload	Bound By	Impact
AI discovery API calls	Network (waiting for LLM response)	Low CPU, high latency
AI batch classification	Network (25 companies per call)	Low CPU, moderate latency
Web scraping (Perplexity/Google)	Network (search API response)	Low CPU, high latency
Python PDF/Excel parsing	CPU (text extraction)	High CPU, brief duration (max 60s)
Next.js SSR	CPU (React rendering)	Moderate CPU, scales with concurrent users
PostgreSQL queries	I/O (disk reads)	Low CPU, depends on query complexity

Key insight: Most workloads are network-bound (waiting for AI APIs). CPU is rarely the bottleneck. This means CX (shared CPU) servers are appropriate even at moderate scale.

2. Simulation: 10 Users (Beta — Month 0-2)

User Behavior Profile

Metric	Value
Total registered users	10
Daily active users	4-6
Concurrent users (peak)	2-3
Discovery searches/day	5-8
Contact reveals/day	2-4
Page views/day	100-200

Resource Consumption

Resource	Usage	Capacity	Utilization
Kadikoy CPU	~5% avg, 15% peak	4 vCPU	Low
Kadikoy RAM	~2.5GB (Next.js + Node)	8GB	31%
Galata CPU	~8% avg (Supabase + other sites)	8 vCPU	Low
Galata RAM	~10GB (Supabase + all services)	16GB	62%
Database size	~500MB	160GB	<1%
API calls/day	5-10 AI calls	Rate limit: 10/min	Minimal
Bandwidth	~2GB/day	20TB/month	<1%

Monthly Costs

Item	Cost	Notes
Infrastructure	$23.80	Fixed
AI API (240 searches/mo)	$14.40	240 × $0.06
AI API (90 reveals/mo)	$2.70	90 × $0.03
Total	$40.90
Revenue	$0	Beta — free access
Net	-$40.90	Acceptable beta cost

Scaling Triggers: None

Everything comfortably within capacity. No action needed.

3. Simulation: 50 Users (Early Paid — Month 4-6)

User Behavior Profile

Metric	Value
Total registered users	50 (35 free, 12 Pro, 3 Team)
Daily active users	15-25
Concurrent users (peak)	5-8
Discovery searches/day	30-50
Contact reveals/day	20-40
Batch operations/week	2-3 (Team users)
Page views/day	500-1,000

Resource Consumption

Resource	Usage	Capacity	Utilization
Kadikoy CPU	~12% avg, 35% peak	4 vCPU	Comfortable
Kadikoy RAM	~3.5GB	8GB	44%
Galata CPU	~12% avg	8 vCPU	Low
Galata RAM	~11GB	16GB	69%
Database size	~2GB	160GB	1.3%
API calls/day	30-60 AI calls	Rate limit covers	Moderate
Bandwidth	~8GB/day	20TB/month	1.2%

Monthly Costs

Item	Cost	Notes
Infrastructure	$23.80	Fixed
AI API (1,200 searches/mo)	$72.00	1,200 × $0.06
AI API (900 reveals/mo)	$27.00	900 × $0.03
Stripe fees	$22.00	~15 transactions
Total costs	$144.80
Revenue	$1,155	12 Pro × $59 + 3 Team ×$ 149
Net profit	+$1,010	87.5% operating margin

Scaling Triggers: None Yet

All resources below 50% utilization on average. Database well within capacity. No upgrade needed.

4. Simulation: 200 Users (Growth — Month 9-12)

User Behavior Profile

Metric	Value
Total registered users	200 (120 free, 65 Pro, 14 Team, 1 Enterprise)
Daily active users	60-90
Concurrent users (peak)	15-25
Discovery searches/day	120-200
Contact reveals/day	80-150
Batch operations/week	8-12
Page views/day	3,000-5,000

Resource Consumption

Resource	Usage	Capacity	Utilization	Status
Kadikoy CPU	~25% avg, 60% peak	4 vCPU	Moderate	Watch
Kadikoy RAM	~5GB	8GB	62%	Watch
Galata CPU	~18% avg	8 vCPU	Low	OK
Galata RAM	~12GB	16GB	75%	Watch
Database size	~8GB	160GB	5%	OK
Database connections	~30 active	Default pool: 100	30%	OK
API calls/day	120-250 AI calls	Rate limit: 600/hr	Moderate	OK
Bandwidth	~25GB/day	20TB/month	3.8%	OK

Monthly Costs

Item	Cost	Notes
Infrastructure	$23.80	Fixed — still on original servers
AI API (4,800 searches/mo)	$288.00	Growing
AI API (3,450 reveals/mo)	$103.50	Growing
AI API (batch, 500 ops/mo)	$15.00	Team/Enterprise users
Stripe fees	$190.00	~80 transactions
Total costs	$620.30
Revenue	$5,785	65 Pro + 14 Team + 1 Enterprise
Net profit	+$5,165	89.3% operating margin

Scaling Triggers

Trigger	Threshold	Action	Cost Impact
Kadikoy CPU avg >50%	Approaching	Rescale CX33 → CX43 (8 vCPU, 16GB)	+€5.40/mo (+$5.86)
Galata RAM >80%	Approaching	Monitor — Supabase is the heavy consumer	—
DB connections >60	Not yet	Enable PgBouncer connection pooling	$0 (config change)

Recommended Action at 200 Users

Rescale Kadikoy from CX33 to CX43 (~

30/mo total infra instead of

24/mo). This is a Hetzner “Rescale” operation — one-click, zero-downtime (virtual hardware upgrade).

5. Simulation: 1,000 Users (Scale — Month 24-36)

User Behavior Profile

Metric	Value
Total registered users	1,000 (520 free, 350 Pro, 110 Team, 20 Enterprise)
Daily active users	300-450
Concurrent users (peak)	50-80
Discovery searches/day	500-800
Contact reveals/day	300-500
Batch operations/day	50-100
Page views/day	15,000-25,000

Resource Consumption

Resource	Usage	Capacity (after upgrade)	Utilization	Status
Kadikoy CPU	~45% avg, 85% peak	8 vCPU (CX43)	High	Upgrade needed
Kadikoy RAM	~10GB	16GB (CX43)	62%	OK
Galata CPU	~30% avg	8 vCPU (CX43)	Moderate	OK
Galata RAM	~14GB	16GB	87%	Upgrade needed
Database size	~40GB	160GB	25%	OK
Database connections	~80-120 active	Pool: 100 (needs PgBouncer)	High	Configure
API calls/day	600-1,200 AI calls	Rate limit: 1,200/hr	Moderate	OK
Bandwidth	~80GB/day	20TB/month	12%	OK

Monthly Costs

Item	Cost	Notes
Infrastructure (upgraded)	$60.00	Galata CX53 + Kadikoy CX53 (see below)
AI API (20,000 searches/mo)	$1,200.00	Largest cost center
AI API (12,000 reveals/mo)	$360.00
AI API (batch, 2,000 ops/mo)	$60.00	Enterprise + Team
AI API (deep analysis, 500/mo)	$50.00	Enterprise users
Stripe fees	$1,100.00	~480 transactions
Monitoring tools (PostHog, Sentry)	$50.00	May exceed free tiers
Total costs	$2,880.00
Revenue	$35,200	350 Pro + 110 Team + 20 Enterprise
Net profit	+$32,320	91.8% operating margin

Required Infrastructure Upgrades

Server	Current	Upgrade To	New Specs	New Cost
Galata	CX43	CX53	16 vCPU, 32GB RAM, 320GB	~€22/mo
Kadikoy	CX43	CX53	16 vCPU, 32GB RAM, 320GB	~€22/mo
			Total infra	~€50/mo ($54)

Additional Infrastructure Needs at 1,000 Users

Need	Solution	Cost	Priority
Connection pooling	PgBouncer (already in Supabase stack)	$0	High
CDN for static assets	Cloudflare (already configured)	$0	Already done
Background job queue	BullMQ + Redis on Kadikoy	$0 (on existing server)	High
Search result caching	Redis on Kadikoy	$0 (on existing server)	Medium
Monitoring upgrade	PostHog Cloud ( $0-50/mo) + Sentry ($ 0-26/mo)	$0-76/mo	Medium
Backup verification	Automated restore tests (monthly)	$0 (script)	High

6. Scaling Roadmap

Users    Infra Action                                    Monthly Infra Cost
──────   ────────────────────────────────────────────   ──────────────────
0-50     Original config (CX43 + CX33)                  $24
50-200   Rescale Kadikoy CX33 → CX43                    $30
200-500  Rescale Galata CX43 → CX53                     $44
500-1K   Rescale both to CX53, add PgBouncer + Redis     $54
1K-5K    Add VPS3 (dedicated worker), separate DB        $80-120
5K+      Consider managed DB or Supabase Cloud           $200+

         $200 ┤                                              ████
              │                                         ████ ████
         $120 ┤                                    ████ ████ ████
              │                               ████ ████ ████ ████
          $54 ┤                          ████ ████ ████ ████ ████
          $44 ┤                     ████ ████ ████ ████ ████ ████
          $30 ┤                ████ ████ ████ ████ ████ ████ ████
          $24 ┤  ████ ████ ████ ████ ████ ████ ████ ████ ████ ████
            0 ┤
              └──10───50──100──200──500──1K───2K───5K───10K──20K──
                                Users

7. AI Cost Simulation

Cost by Provider Strategy

Strategy	Searches/mo	Cost/Search	Monthly AI Cost	Notes
All Gemini Flash	20,000	$0.03	$600	Cheapest but no web search
Gemini + Perplexity	20,000	$0.06	$1,200	Current strategy (balanced)
All Perplexity Sonar	20,000	$0.10	$2,000	Best search quality, most expensive
Self-hosted OSS	20,000	$0.01	$200	Future: GPU VPS (~$50/mo)

Cost Reduction Strategies

Strategy	Savings	Implementation
Search cache hit rate (30-day TTL)	20-40%	Already built (`export_ai_search_history`)
Smart provider routing	30%	Use Gemini for classification, Perplexity only for web search
Batch classification (25/call)	80% vs individual	Already built (AI batch pipeline)
Result deduplication	10-15%	Skip companies already enriched recently
Off-peak processing	0% (no time pricing)	N/A for API-based
Self-hosted models (future)	60-80%	Requires GPU VPS ($50-100/mo) at 5K+ users

Projected AI Cost at Scale

Users	Searches/mo	Cache Hit	Net Searches	AI Cost/mo	As % of Revenue
10	300	5%	285	$17	N/A (beta)
50	1,500	15%	1,275	$77	6.7%
200	6,000	25%	4,500	$270	4.7%
1,000	30,000	30%	21,000	$1,260	3.6%
5,000	150,000	35%	97,500	$5,850	2.8%

AI cost as % of revenue decreases at scale because subscription revenue scales linearly while AI cost benefits from caching and batch efficiencies.

8. Database Growth Simulation

Storage Projection

Users	Companies in DB	Contacts	Searches	Est. DB Size	% of 160GB
10	2,500	500	3,000	500MB	<1%
50	12,000	3,000	15,000	2GB	1.3%
200	50,000	15,000	60,000	8GB	5%
1,000	250,000	80,000	300,000	40GB	25%
5,000	1,000,000	350,000	1,500,000	150GB	94%

Scaling Actions

Threshold	Action
DB > 100GB	Upgrade to CX53 (320GB disk) or add volume
DB > 250GB	Consider dedicated DB server or Supabase Cloud
1M+ companies	Implement data archiving for stale records
10M+ search_history	Aggregate and purge raw records older than 6 months

9. Concurrent User Simulation

Next.js SSR Performance

Concurrent Users	Avg Response Time	Server Load (Kadikoy CX43)	Status
1-10	<100ms	<10% CPU	Excellent
10-30	100-200ms	15-30% CPU	Good
30-60	200-400ms	30-50% CPU	Acceptable
60-100	400-800ms	50-70% CPU	Upgrade soon
100+	>800ms	>70% CPU	Upgrade required

API Route Performance (AI-Calling)

Route	Avg Latency	Bottleneck	Concurrency Limit
`POST /api/discover`	8-15s	LLM response time	10 req/60s per user
`POST /api/headhunt`	5-10s	Web search API	10 req/60s per user
`POST /api/ai/classify`	3-8s	Batch LLM call	10 req/60s per user
`POST /api/ai/complete`	2-5s	LLM response	20 req/60s per user
`GET /api/companies`	<200ms	PostgreSQL query	No limit
`GET /api/contacts`	<200ms	PostgreSQL query	No limit

AI routes are latency-bound by external APIs (8-15s), not server resources. Even at 1,000 users, only 50-80 are concurrent, and only a fraction are running AI queries simultaneously. The rate limiter prevents abuse.

10. Breakpoint Analysis

When Does Each Resource Hit Its Limit?

Resource	Limit	Breakpoint (Users)	Action	Cost to Fix
Kadikoy CPU	4 vCPU sustained >60%	~150-200	Rescale to CX43	+$6/mo
Kadikoy RAM	8GB >80%	~300-400	Rescale to CX43 (16GB)	Included above
Galata RAM	16GB >85%	~500-800	Rescale to CX53 (32GB)	+$13/mo
DB storage	160GB >80%	~4,000-5,000	Upgrade disk or CX53	+$13/mo
DB connections	100 concurrent	~500-700	PgBouncer pooling	$0
Bandwidth (20TB)	>60% utilization	~10,000+	Unlikely to hit	$0
Stripe free tier	N/A (pay per txn)	Never	Always pay-per-use	% of revenue
PostHog free tier	1M events/month	~2,000-3,000 users	Upgrade to $0-50 tier	$0-50/mo

Infrastructure Cost vs Revenue at Each Breakpoint

Revenue/Costs
     │
$35K ┤                                              ●  Revenue
     │                                         ●
$25K ┤                                    ●
     │                               ●
$15K ┤                          ●
     │                     ●
$10K ┤                ●
     │           ●
 $5K ┤      ●
     │  ●                                              ● Costs (incl. AI)
 $1K ┤● ─ ─ ● ─ ─ ● ─ ─ ● ─ ─ ● ─ ─ ● ─ ─ ● ─ ─ ● ─ ─ ●
   0 ┤
     └──10───50──100──200──500──1K───2K───3K───5K──
                        Users

Revenue grows much faster than costs at every scale. The gap widens because fixed costs are minimal and AI costs benefit from caching.

11. Disaster Scenarios

What If Traffic Spikes 10x Overnight?

Scenario	Impact	Recovery
Viral LinkedIn post → 500 signups in a day	SSR slows to 2-3s, some 503s	Hetzner Rescale (15 min) — bump Kadikoy to CX53
Trade fair demo → 100 discovery searches in 1 hour	Rate limiter kicks in (10/min/user), queue forms	Temporarily increase rate limit, AI APIs handle it
Bot attack → 10,000 requests/min	Cloudflare WAF blocks, rate limiter rejects	Already protected at network + app level
Database corruption	WAL-PITR restores to any point in time	Max 5 min data loss (WAL shipping interval)
Hetzner datacenter outage	Full downtime	No HA — accept ~4h/year downtime for cost savings

Acceptable Downtime Target

Tier	SLA	Monthly Downtime Budget	Appropriate At
Current	99.5%	3.6 hours	0-1,000 users
Growth	99.9%	43 minutes	1,000-10,000 users (add HA)
Scale	99.95%	22 minutes	10,000+ users (multi-region)

12. Cost Summary — All Scales

Scale	Users	Infra/mo	AI/mo	Stripe/mo	Total Cost/mo	Revenue/mo	Margin
Beta	10	$24	$17	$0	$41	$0	-100%
Early	50	$24	$99	$22	$145	$1,155	87%
Growth	200	$30	$407	$190	$627	$5,785	89%
Scale	1,000	$54	$1,670	$1,100	$2,824	$35,200	92%
Big	5,000	$120	$6,850	$5,000	$11,970	$175,000	93%

Operating margin improves with scale. This is the hallmark of a well-architected SaaS: fixed costs become negligible, AI costs benefit from caching, and revenue scales linearly with users.

Summary

Question	Answer
Can current infra handle 200 users?	Yes — one Kadikoy upgrade (~$6/mo more)
When is the first real upgrade needed?	~200 users (Kadikoy rescale)
What’s the biggest cost at scale?	AI API calls (~48% of costs at 1K users)
Is the cost model sustainable?	Yes — margin improves from 87% to 93% at scale
When might we need a 3rd server?	~2,000-5,000 users (dedicated worker VPS)
Can we handle a viral spike?	Yes — Hetzner Rescale (15 min) + rate limiting

Next: INDEX.md — Document index and PM Knowledge Base guide

Overview

Strategy

Economics

Operations

​Cernio — Production Simulation

​1. Infrastructure Baseline

​Current Architecture (Post-Migration to Hetzner)

​What Runs Where

​Scaling Decision: CPU vs Network Bound

​2. Simulation: 10 Users (Beta — Month 0-2)

​User Behavior Profile

​Resource Consumption

​Monthly Costs

​Scaling Triggers: None

​3. Simulation: 50 Users (Early Paid — Month 4-6)

​User Behavior Profile

​Resource Consumption

​Monthly Costs

​Scaling Triggers: None Yet

​4. Simulation: 200 Users (Growth — Month 9-12)

​User Behavior Profile

​Resource Consumption

​Monthly Costs

​Scaling Triggers

​Recommended Action at 200 Users

​5. Simulation: 1,000 Users (Scale — Month 24-36)

​User Behavior Profile

​Resource Consumption

​Monthly Costs

​Required Infrastructure Upgrades

​Additional Infrastructure Needs at 1,000 Users

​6. Scaling Roadmap

​7. AI Cost Simulation

​Cost by Provider Strategy

​Cost Reduction Strategies

​Projected AI Cost at Scale

​8. Database Growth Simulation

​Storage Projection

​Scaling Actions

​9. Concurrent User Simulation

​Next.js SSR Performance

​API Route Performance (AI-Calling)

​10. Breakpoint Analysis

​When Does Each Resource Hit Its Limit?

​Infrastructure Cost vs Revenue at Each Breakpoint

​11. Disaster Scenarios

​What If Traffic Spikes 10x Overnight?

​Acceptable Downtime Target

​12. Cost Summary — All Scales

​Summary

Cernio — Production Simulation

1. Infrastructure Baseline

Current Architecture (Post-Migration to Hetzner)

What Runs Where

Scaling Decision: CPU vs Network Bound

2. Simulation: 10 Users (Beta — Month 0-2)

User Behavior Profile

Resource Consumption

Monthly Costs

Scaling Triggers: None

3. Simulation: 50 Users (Early Paid — Month 4-6)

User Behavior Profile

Resource Consumption

Monthly Costs

Scaling Triggers: None Yet

4. Simulation: 200 Users (Growth — Month 9-12)

User Behavior Profile

Resource Consumption

Monthly Costs

Scaling Triggers

Recommended Action at 200 Users

5. Simulation: 1,000 Users (Scale — Month 24-36)

User Behavior Profile

Resource Consumption

Monthly Costs

Required Infrastructure Upgrades

Additional Infrastructure Needs at 1,000 Users

6. Scaling Roadmap

7. AI Cost Simulation

Cost by Provider Strategy

Cost Reduction Strategies

Projected AI Cost at Scale

8. Database Growth Simulation

Storage Projection

Scaling Actions

9. Concurrent User Simulation

Next.js SSR Performance

API Route Performance (AI-Calling)

10. Breakpoint Analysis

When Does Each Resource Hit Its Limit?

Infrastructure Cost vs Revenue at Each Breakpoint

11. Disaster Scenarios

What If Traffic Spikes 10x Overnight?

Acceptable Downtime Target

12. Cost Summary — All Scales

Summary