AI credits — pay for what you use
যেখানে যতটুকু AI দরকার, ততটুকুই কিনুন। কোনো hidden seat-fee নেই — credit fixed-rate, কেনার পর ৩০ দিন valid, BDT-তে EPS-এ pay।
Different actions cost different amounts — honestly
A 30-page tender analysis is genuinely 5× the work of a clarifying chat message. We model that directly in credits instead of hiding it behind a flat “X queries / month”.
Ask the copilot anything about a tender — eligibility, BoQ rates, PPR clauses, deadlines. About 5,000 input + 1,000 output tokens.
Upload a ZIP, get a complete brief — fit score, BoQ structured, eligibility verdict, PPR 2008 compliance flags. ~30K tokens of work.
Follow-up questions about an already-analyzed tender. Same cost as a chat message — context is reused.
Searching across your tenders or PPR rules — embedding cost is negligible and bundled into the chat answer that uses it.
Four packs · BDT-priced · 30-day validity
- Valid for 30 days
- Stacks with subscription
- Valid for 30 days
- Stacks with subscription
- Valid for 30 days
- Stacks with subscription
- Valid for 30 days
- Stacks with subscription
Why credits, not seats?
A clarifying chat is genuinely 5× cheaper than a full analysis. Modeling this in credits keeps the price-fairness story clear; flat “X queries / month” quotas hide it.
Power users on the smallest plan can buy more credits without forcing a full plan upgrade. Avoids the “upgrade-to-keep-using-AI” rage near a tender deadline.
Every consumption is recorded with a ledger row. If the AI gives you garbage on a specific query, support can refund in credits — no “your monthly bucket is gone” drama.
Cloud providers may re-price AI inference without notice. Credits give us a stable abstraction layer — we absorb the change quietly, you keep the same per-credit cost on what you've already bought.