AI Startup Building

AI startup বানানো আজকের সময়ের সবচেয়ে বড় opportunity — কিন্তু ৯০% fail করে market না বোঝার কারণে, technology এর জন্য না। Build last, sell first।

• নিজের কাজে repeated pain — best source।
• Boring industry + AI = blue ocean (insurance, logistics, accounting)।
• ‘AI-native’ rebuild of existing tool (Notion vs Word)।
• ChatGPT এর শখ না — recurring + paid problem।
• Y Combinator Request for Startups list দেখো।

1. Customer interview × 15-20  (no leading questions)
2. Landing page + waitlist  (Carrd / Framer)
3. Manual prototype  (you = the AI)
4. Concierge MVP  (5 paying customers)
5. Build software only when manual breaks

‘Mom Test’ পড়ো — কীভাবে honest feedback বের করতে হয় শেখায়।

• Plain GPT wrapper = no moat, ৩ মাসে replicated।
• Real moat: proprietary data, workflow integration, distribution, brand, network effect।
• ‘Last mile’ — domain workflow embedding ই বাঁচায়।
• Fine-tuned/own model on niche data — strong moat।
• Community + content engine — slow but durable।

• Solo possible, কিন্তু 2 co-founder = higher survival rate।
• Complement: hacker + hustler (build + sell)।
• Equity vesting 4 yr, 1 yr cliff — সবসময়।
• Co-founder agreement আগে — bad breakup costly।
• Early hire: GTM > more engineer (usually)।

Bootstrap     — revenue থেকে grow (ownership 100%)
Friends/Family — $25-100k
Pre-seed      — $250k-1M, angel / micro-VC
Seed          — $1-4M, traction চাই
Series A      — $5-15M, repeatable GTM
Series B+     — scale

• Accelerator: Y Combinator, Antler, Techstars, On Deck।
• AI specific: a16z Speedrun, Neo, AI Grant।
• VC pitch = traction story + market size + team।
• Revenue থাকলে negotiation power বাড়ে।

• MRR / ARR growth (15-20%/month early)।
• Net revenue retention > 100%।
• CAC payback < 12 month।
• Gross margin > 70% (AI inference cost সমস্যা)।
• Logo concentration < 20% per customer।
• Activation rate, weekly active usage।

• Foundation model এর next version তোমার feature কে obsolete করতে পারে।
• API price drop / TOS change — vendor lock risk।
• Inference cost > price — unit economics ভেঙে যায়।
• Hallucination liability — legal/medical এ বড় সমস্যা।
• Data privacy — EU AI Act, GDPR comply।

• Content + SEO (programmatic landing pages — AI advantage)।
• Community-led (Discord, Slack, subreddit)।
• Founder-led sales (first 100 customer)।
• Product Hunt + Hacker News launch।
• Integration marketplace (Zapier, Notion, Slack apps)।
• Partnership / white-label।

• প্রতি সপ্তাহে ৩ customer call।
• Weekly metric review (north star + 3 input metric)।
• Build in public — distribution + accountability।
• Burn rate জানো — runway = cash / monthly burn।
• Mental health — therapist, exercise, sleep।