Grow on AgentDukaan · Intermediate · 6 min read
A practical, India-aware guide to pricing your AI agent: value-based pricing, tiers, launch offers, and a fill-in worksheet.
Normally ₹1,499 — free while we launch.
How to Price Your AI Agent
Pricing is the difference between an agent that sells and one that sits in your dashboard collecting zero subscribers. This guide walks you through choosing a model, setting anchors and tiers, and running launch offers without training buyers to expect everything for free. Bring a real agent you've built — by the end you'll have a number you can defend.
Start from value, not from your costs
The most common mistake is cost-based pricing: add up your API spend, hosting, and time, slap on a margin, and call it a price. It feels safe and it's almost always too low. Your buyer doesn't care that GPT tokens cost you ₹0.40 per conversation — they care that your WhatsApp support agent saved them from hiring a ₹18,000/month junior, or that your invoice-chaser recovered ₹40,000 in late payments last month.
Value-based pricing starts from the buyer's outcome and prices a fraction of it. The rule of thumb: capture 10–25% of the measurable value you create. If your agent saves a clinic 15 hours of front-desk time a month (worth ~₹6,000 at local wages), pricing it at ₹999–₹1,499/month is an easy yes.
Use cost as a floor, never as the price. Know your unit economics so you don't go bankrupt on a power user — then forget them when you talk to the customer.
| Cost-based | Value-based | |
|---|---|---|
| Anchor | Your expenses + margin | Buyer's saved time/money |
| Typical outcome | Underpriced, thin margins | Healthy margins, easier "yes" |
| When it's okay | Pure commodity utility | Almost everything else |
| Conversation | "It costs me X" | "It saves you Y" |
To find your value number, ask three prospective buyers a single question: "What does this problem cost you today — in hours, money, or missed sales?" Their answers are your pricing research.
Subscription vs one-time vs done-for-you
On AgentDukaan you can sell the same agent three ways, and they have very different economics. You don't have to pick one — many sellers offer all three.
- Subscription (hosted): Buyer pays monthly/annually; you host and maintain. Best recurring revenue, but you carry ongoing API and support cost. Price so that one month covers several months of your serving cost — aim for the subscription to clear your monthly cost-to-serve within the first week.
- One-time (source code): Buyer pays once, gets the code, runs it themselves. Higher sticker price (5–15× a monthly fee is normal), zero recurring revenue, near-zero ongoing cost for you. Good for technical buyers and agencies who want to self-host and white-label.
- Done-for-you (setup): You configure, connect channels (WhatsApp/Telegram/email), and hand over a working system. This is consulting — price your time honestly. A ₹5,000–₹25,000 setup fee on top of a subscription is common, and it dramatically lowers buyer hesitation.
A simple pattern that works: low monthly subscription + optional DFY setup fee. The subscription gets people in the door; the setup fee captures buyers who'd rather pay than fiddle. Add annual billing at ~2 months free to improve cash flow and cut churn.
GST note: list whether your prices are inclusive or exclusive of 18% GST, and show it clearly at checkout. Indian business buyers will ask for a GST invoice — sort this before you launch, not after.
Anchors and tiers
People judge price by comparison, so give them the comparison you want. Three tiers work better than one because they create a frame: most buyers avoid the cheapest and the most expensive and land in the middle — so design your middle tier to be the one you actually want to sell.
A reliable structure:
| Tier | Who it's for | Lever that scales | Example |
|---|---|---|---|
| Starter | Solo / trying it out | Low volume cap | ₹499/mo, 200 conversations |
| Growth ⭐ | The target buyer | Higher cap + 1 channel extra | ₹1,499/mo, 1,500 conversations, WhatsApp + Telegram |
| Pro / Business | Agencies, high volume | Generous cap + priority support | ₹4,999/mo, 10,000 conversations, all channels, priority support |
Pick a metric that grows with the buyer's value — conversations handled, leads captured, invoices processed — not something arbitrary. Anchor against the alternative they'd otherwise pay for ("cheaper than one part-time hire"), and use charm pricing (₹999, ₹1,499) since it reliably reads as a deal in Indian markets. Round annual numbers (₹9,999/yr) feel cleaner for the commitment tier.
Trials, launch offers, and "first N free"
Free is a powerful acquisition tool and a dangerous habit. Use it to remove risk, not to set your real price.
- Free trial (time-boxed): 7–14 days, ideally requiring channel setup so they reach the "aha" moment. Card-on-file trials convert better than no-card trials but reduce signups — for a new agent, favour no-card to build reviews.
- "First N free": Give the first 1,000 conversations (or first 10 buyers) free. It's concrete, generous, and naturally caps your exposure — much safer than an open-ended free tier.
- Launch offer: A genuine, time-boxed discount ("₹999 → ₹699/mo for the first 50 subscribers, locked for life") rewards early adopters and creates urgency without devaluing the product. Always show the crossed-out original so the anchor survives.
Avoid a permanent free tier unless free users actively help you (referrals, content, reviews). Otherwise you pay API costs forever for people who'll never convert.
Don't race to the bottom
When buyers compare listings, it's tempting to undercut. Resist it — a price war you win still loses, because the floor is set by whoever values their time least, and you can't out-cheap a hobbyist.
Compete on differentiation instead: a sharper niche (agents for dental clinics, not "support bots"), proven outcomes in your listing, faster setup, better channel coverage, real support. A ₹1,499 agent with reviews and a 2-minute WhatsApp setup beats a ₹399 agent nobody trusts. If you're the cheapest option, that's usually a positioning problem, not a price problem.
A simple pricing worksheet
Fill this in for your agent before you publish. Most of it takes 20 minutes.
- Outcome: In one line, what does the buyer get? (e.g. "Never miss a WhatsApp lead again.")
- Value: What does this problem cost them now per month? (hours × wage, or ₹ lost)
- Cost-to-serve: Your API + hosting + support per active buyer per month.
- Floor: Cost-to-serve × 2. Never price below this.
- Target price: 10–25% of monthly value, rounded to charm pricing.
- Three tiers: Define Starter / Growth / Pro and the metric that scales them.
- One-time price: 5–15× your monthly fee (if you sell source).
- DFY setup fee: Your honest hours × rate.
- Launch offer: Discount, cap, and end date — written down.
- GST: Inclusive or exclusive, stated clearly.
Use this prompt to pressure-test your numbers with an AI before you commit:
You are a SaaS pricing strategist for the Indian market. Here is my AI agent:
- What it does: <one line>
- Buyer: <who, what size business>
- Monthly value it creates for them: ₹<number> (and how I estimated it)
- My monthly cost-to-serve per buyer: ₹<number>
- Proposed tiers: <Starter / Growth / Pro with prices and limits>
Critique my pricing. Flag where I'm underpricing, whether my tier metric
scales with buyer value, whether the middle tier is the obvious choice,
and suggest a launch offer with a cap and end date. Use rupee pricing and
note GST handling. Be specific and push back on weak assumptions.
Next steps
- Run the worksheet above on your agent and write down a target price today.
- Test that price with three real prospects using the value question.
- Pressure-test with the prompt, then publish a tiered listing — see /guides/listing-that-converts to make it convert.
When you're ready, list your agent on AgentDukaan and try a small "first 50 subscribers" launch offer — you can adjust tiers anytime as the data comes in. Questions on billing or GST setup? The help center has you covered.