👥 CONDITIONAL

Telegram Expert Platform

Expert marketplace with an AI assistant and Telegram-based monetization

Category: marketplace
Date: 2026-03-20
ID: 274882d7…
AI Score ?Idea potential score (0-100). The final verdict (GO / CONDITIONAL / NO-GO) is the qualitative consensus of 5 AI models based on all risks and conditions - it can differ from the numeric score alone.
62/100
⚠ Conditional GO — risks exist, hypotheses need validation
Financial Dashboard — Key Numbers
Investment Required ?Total capital needed to reach break-even: servers, marketing, development.
$5,000
to get started
Break-even ?The month when monthly profit will cover all startup costs.
Month 15
from launch
MRR Target ?Monthly recurring revenue at which the project is considered successful and ready to scale.
$3,000
per month
Margin ?Share of each dollar remaining after infrastructure, APIs, and direct costs. 70%+ is healthy for SaaS.
96%
of revenue retained
Monthly Revenue Growth Forecast
Unit Economics — Numbers per Customer
Revenue per Customer ?How much one customer pays per month (ARPU). The higher, the fewer customers needed for strong profit.
$9.99/mo
per customer avg.
Profitability ?Share of each dollar remaining after servers, APIs, and other direct costs.
~96%
of revenue retained
Break-even ?When startup costs will be fully recovered and the business begins generating net profit.
Month 14–16
to profitability
Customer Lifetime Value ?Total revenue from one customer over the entire relationship (LTV). Ideally 3× above acquisition cost.
$199.8
lifetime total
Max Acquisition Cost ?Maximum ad spend per customer while keeping the business model profitable (CAC target).
$30
per new customer
LTV / CAC ?Ratio of customer lifetime value to acquisition cost. 3× and above is healthy.
6.7×
✓ Above benchmark
Development Scenarios
Scenario Revenue by Month 6 Revenue by Month 12 Key Assumption
Pessimistic $41 $192 CAC higher than forecast, conversion below 5%
Realistic $119 $550 On plan: CAC ≤ target, churn ≤ 5%/month
Optimistic $297 $1,375 Virality kicked in, CAC 2x below target
Why This Verdict

✓ Arguments FOR

  • Telegram as a distribution channel: 900M+ users
  • AI scales expertise without linear cost growth
  • Native monetization through Telegram Stars

✗ Why not higher

  • Regulatory risks in professional niches (medicine, law)
  • AI quality versus a real expert: trust is hard to build
  • The marketplace chicken-and-egg problem
🛑 When to Stop the Project
K1
Zero conversion per channel
experiment.starts == 0 AND experiment.channel_views >= 1000
→ pause experiment
K2
CAC exceeds LTV threshold
avg_cpl_rub > 2000 AND total_starts >= 50
→ stop all paid traffic
K3
Budget exhausted
budget.remaining_telega <= 300
→ stop new campaigns and escalate
K4
No paying users after 200 starts
total_starts >= 200 AND paying_users == 0
→ pause traffic and fix paywall
K5
No data for 7 days
days_since_last_experiment_update >= 7
→ escalate check server
What to Validate Before Scaling
1
Telegram users prefer getting expert consultations through a familiar interface without installing new apps
2
An AI assistant trained on a specific expert's knowledge scales their income without losing quality
3
A vertical focus (legal, financial, or medical niche) reduces the chicken-and-egg problem
What the AI Models Said
~
Claude Opus (Critic)
A marketplace of AI experts in Telegram runs into the chicken-and-egg problem. It's hard to attract experts without users, and vice versa. Telegram doesn't provide marketplace infrastructure.
~
GPT-4.1 (Market Strategist)
Telegram as a platform for monetizing expertise through AI assistants is a real trend. Telegram Stars creates native monetization. The key is the right vertical focus.
~
Grok-3 (Technical Analyst)
Expert AI in Telegram is the next step after ChatGPT. A 'personal AI financial advisor' or 'AI lawyer' through a bot taps mass demand with high willingness to pay.
Milestones & Stages
M1
First CPL data, Wave 1+2
In Progress
M2
CPL below the kill threshold
Pending
M3
First paying user
Pending
M4
Outreach start, warm messages
Done
M5
6+ paying users ($60 MRR)
Pending
Investment & Exit Scenarios

Total Investment Needed

$5,000
to reach profitability
Marketing $2,500
Development $1,250
Infrastructure $750
Operations $500
🚦 Strengths & Risks at a Glance

✓ Green Flags

  • Personal memory creates a moat and switching difficulty after 4–6 months of active use
    4–6 months until meaningful switching cost
  • The freemium plus $9.99/month model with ~96% margin shows healthy unit economics
    96% margin, cash LTV/CAC = 120x (P50)
  • Candy.ai proves the model: $25M ARR, ~250K paying users, no venture funding
    $25M ARR, 250K paying users
  • Telegram has 1 billion MAU and built-in payments, zero distribution cost
    1 billion MAU
  • Infrastructure cost improves with scale: $0.17/month per user at 10K users
    $0.17 per user at 10K scale
  • An AI team through Claude Code replaces 5–10 people at $0 extra cost
    80–90% of roles replaced by AI
  • TAM of $120M for AI companions in 2025, an open position as a niche expert with memory
    $120M TAM

✗ Red Flags

  • All 8 critical hypotheses (H1, H2, H3, H4, H8, H9, H12) are UNVERIFIED before MVP development
    high
  • ChatGPT GPTs offer specialization for free, a direct high-priority threat
    high
  • Stars friction: no precedent of $1K+ MRR through Stars alone, real conversion unknown
    high
  • Legal risk in health and law niches; requires disclaimers and positioning as a 'navigator,' not a consultant
    high
  • Freemium-to-Pro conversion cut to a realistic 1.5% in the Anti-Optimism Audit; the original 3% was overstated
    medium
  • Churn of 12% (M1–3) / 8% (M4+), higher than the optimistic scenario (4%); a slower path to payback
    medium
  • Telegram may build native AI chat within 12 months, potentially displacing bots
    medium
  • The product can be cloned in 5–10 days; the only moat is accumulated user data
    medium
⚠️ Risk Matrix
Risk Probability Impact Mitigation
Free ChatGPT GPTs / direct competition
🛑 Kill trigger
 65% High Hyper-specialization with personal memory; GPTs have no switching cost
Stars friction / payment conversion failure
🛑 Kill trigger
 60% High Diversification: Stars + YooKassa (RU) + Stripe (international)
Free-to-paid conversion < 1% within 3 months
🛑 Kill trigger
 45% High Launch only after H1 verification (≥3 of 20 say yes, would pay)
Cloning by competitors within 5–10 days
 80% Medium Being first doesn't win; whoever has more data and a better product wins
Legal risk (health + law)
 30% High Mandatory disclaimers; positioning as informational, not diagnostic
Telegram builds native AI chat
 50% High Build up the user base within 12 months; personal history creates switching cost
Churn > 15%/month at month 4+
 25% Medium Minimize through personal memory; check retention early
AI commoditization / price pressure
 70% Medium Focus on specialization and memory instead of cheap LLM pricing
💸 Monthly Cash Flow (Realistic Scenario)
Period Revenue Expenses Net Cumulative
M0 / Preparation $0 $500 -$500 -$500
M1 / MVP launch $30 $560 -$530 -$1,030
M3 / +2 experts $140 $700 -$560 -$2,150
M6 / MAX DRAWDOWN $380 $900 -$520 -$3,710
M9 / Channel growth $750 $1,100 -$350 -$4,235
M12 / Break-even excluding salary $980 $3,330 -$2,350 -$9,485
M18 / PAID BACK $1,500 $1,400 $100 $115
🏁 Competitive Landscape
📡 Market catalyst: Users' willingness to pay for personal AI experts instead of generic chatbots; Telegram as a platform with 1 billion MAU and built-in payments
Competitor Size Take Rate Weakness
ChatGPT Plus
 N/A N/A
Generic, no personal memory, doesn't adapt, $20/month above our price
Character.AI
 N/A N/A
Focused on roleplay, not specialized experts with memory
ChatGPT GPTs
 N/A N/A
Free, but no personal memory or built-in context
Replika
 N/A N/A
Positioned as an AI friend, not an expert; retention is low (7+ months)
Candy.ai
 250K paying users N/A
$25M ARR proves the model, but positioned as 18+, not niche expertise
Claude.ai Pro
 N/A N/A
$20/month, generic AI, no memory or niche focus
🛠 MVP — Week-by-Week Plan
Week 1
  • Telegram webhook + aiogram 3.x
  • LiteLLM integration
  • PostgreSQL connection (Supabase)
  • First response from Claude Haiku
Bot responds to messagesHistory is saved to the database
Week 2
  • Three-tier memory system: short-term plus long-term
  • Semantic search over history (vectors)
  • Redis (Upstash) for sessions
Bot remembers context for ≥20 messagesKey facts are extracted correctly
Week 3
  • Freemium limits: 10 messages/day counter
  • Telegram Stars integration
  • Upgrade to Pro on payment
Limit triggers correctlyPayment through Stars is recorded
Week 4
  • First expert integration: Biohacker
  • Onboarding: asking for goals and parameters
  • Personalized recommendations
The Biohacker gives recommendations in the context of historyUser perceives the personalization
Week 5
  • Add +2 experts (Mom Advisor, Legal Navigator)
  • Switching between experts
  • Feedback form for testers
3 experts available in the botUser can choose an expert
Week 6
  • Beta test with 20 testers
  • Collecting feedback
  • Iterating on bug fixes and UX
≥3 of 20 say they'd pay $9.99NPS > 40Average messages/day < 20
🏰 Competitive Moat

✗ Easy to Copy

  • Core Telegram bot (aiogram, FastAPI)
  • LLM integration (Claude/GPT)
  • Freemium model with a message limit
  • Basic history storage system

✓ Hard to Copy

  • Personal user memory (long-term database plus context)
  • Accumulated conversation history creates switching cost after 4–6 months
  • Specialized prompts for each niche (require testing and iteration)
  • User data and preferences are the core asset
⏱ Moat forms by: M4–M6
📊 Acquisition Cost by Channel
Channel CAC Notes Profitable?
Organic Telegram (channel + referrals) $1.00 Freemium to 1.5% conversion to Pro; average churn 8–12%/month at M4+; ARPU $7.99; LTV $120 ✓ Yes
Referral chats (parenting, biohacking, legal) $0.50 Low friction, high WTP target audience; CAC ~50% of the main channel ✓ Yes
Paid Telegram ads (future, M6+) $3.00 Potential for scaling; requires ROI validation with real data ✗ No
🔬 Anti-Optimism Audit
1
Original free-to-paid conversion of 3% and churn of 7% (M1–3)
→ A realistic 1.5% conversion and 12% churn (M1–3) / 8% (M4+) based on comparables and user behavior
-25 points (10 paying users by M2 → M4–M5; $7K MRR by M18 → ~$1.5K by M18)
2
Fully loaded CAC was calculated as cash-only ($1.00); founder time was ignored
→ Fully loaded CAC ~$27 (15 hrs/week × $80/hr ÷ 200 new users); fully loaded LTV/CAC = 4.5x (acceptable, but not exceptional)
-10 points (the economics are less outstanding than they looked)
3
Max drawdown $2.2K (original forecast)
→ Max drawdown $3.71K (M6) due to slower growth in paying users
-8 points (more capital needed at the start)
4
The deferred status of critical hypotheses (H1, H2, H3, H4, H8, H9, H12) wasn't explicitly flagged as a blocker
→ All 8 critical hypotheses are UNVERIFIED; MVP launch only after H1 verification (≥3 of 20 = yes, would pay)
-7 points (higher risk of building a product that doesn't validate)

Have your own idea?

5 AI models attack it from all angles and deliver an honest GO / NO-GO verdict with real numbers in 24 hours.

Validate for $39 →