AI Credits: Pay for What You Use, Skip What You Don't

I hate SaaS billing that punishes you for using the product.

When I was building OT1-Pro, I looked at how AI tools charge. Some bill per message sent. Some bill per AI reply. Some charge per seat, per month, whether your 11 team members used the AI 500 times or zero times. I watched a small Egyptian retailer in my beta group get a $280 bill from a competitor because their AI auto-responder fired on a week when they ran a campaign and volume spiked 4×. They hadn't done anything wrong — they just got a lot of customers.

That model is broken. So I built something different.

How OT1-Pro AI Billing Actually Works

There are two distinct things the AI does, and they cost different amounts of compute to run:

1. AI auto-replies — responding to incoming messages in real-time. These are fast, run on a lightweight model, and happen constantly. If you have 300 customers messaging per day, that's 300 small inference calls. I include these in every subscription plan, unlimited, because throttling them would break the whole product.
2. Deep Analysis — the AI reads all your stored conversations and surfaces patterns, top objections, missed opportunities, customer segments. This is a heavy operation: it processes months of data, runs multiple LLM passes, and generates a structured report. It costs real compute. It runs once per 24 hours per team, cached to avoid repeat charges for the same data.

So the split is: subscription covers auto-replies, AI credits cover Deep Analysis.

One Deep Analysis run = one AI credit.

The Four-Tier Credit Ladder

Every plan includes credits in the bundle. You can also buy extra credits à la carte. Here's the current structure:

At one credit per Deep Analysis run, the Starter plan's 30 monthly credits = one deep analysis per day, which is the max cadence anyway (the cache window is 24 hours). Most teams use it 2-3 times per week, so 30 credits lasts well over a month in practice.

Why I Chose Credits Over Per-Seat or Per-Message Pricing

Per-seat pricing has a dirty secret: you pay for seats whether they produce value or not. A 10-person team where 3 people actively use AI still pays for 10. That's not how I want to run a tool.

Per-message billing is worse for anyone running campaigns or seasonal businesses. I've seen operators get destroyed in December. A shoe retailer running a Eid al-Adha campaign shouldn't pay 10× more for AI that month just because 10,000 customers replied "متوفر إيه من المقاسات؟"

Credits solve this cleanly: the expensive thing (Deep Analysis) uses credits; the cheap thing (auto-replies) is flat-rate. You know your fixed monthly cost, and the only variable is how often you trigger a compute-heavy analytical operation.

The Billing Meter Chip: What You Actually See

In the OT1-Pro dashboard, every team sees a small credit meter in the top navigation — a chip that shows your remaining credits and refreshes in real time after each operation. It looks like:

• Green chip: 20+ credits remaining
• Yellow chip: 5-19 credits remaining (soft warning)
• Red chip: 1-4 credits remaining (top-up prompt appears)

When you click "Run Deep Analysis," the system shows a confirmation dialog: "This will use 1 AI credit. You have 14 remaining." No surprise deductions. No silent billing at month end. You see the cost before you approve it.

What One AI Credit Buys You in Real Terms

I ran a Deep Analysis last week on a test account with 2,400 stored conversations from a fashion retailer in Jeddah. What came back in one credit:

• Top 7 customer objections ranked by frequency (price was #1 at 34%, not product quality)
• 3 conversation patterns flagged as missed sales: customers who asked for size availability and got no reply within 2 hours
• A "loyal segment" of 41 customers who had bought 3+ times and hadn't been messaged in 90+ days
• Peak inquiry hours: 9pm–11pm Saudi time, Thursdays and Fridays (the brand was sending campaigns at 10am, missing peak completely)
• Language split: 68% Arabic, 22% English, 10% mixed — but the AI was replying in English-first to everyone

That analysis — if done manually — would take a junior analyst 2-3 days. It ran in about 90 seconds and cost $0.30 in credits.

Competitor Comparison: What $29/Month Gets You Elsewhere

Intercom's AI pricing model charges per "AI resolution" — if the AI closes a conversation, you pay. I've seen teams with high-volume support discover that their monthly AI bill at Intercom was more than their base subscription. That is exactly the model I refused to copy.

How Credits Roll Over (and When They Don't)

Monthly credits reset on your billing date. They don't roll over month to month — this is a deliberate choice to keep the accounting simple and the pricing predictable. What you don't use this month doesn't accumulate silently and then confuse you about how many you actually have.

There's one exception: extra credits you purchase (top-up credits) do not expire. If you buy 50 credits à la carte, they stay in your account until you use them, even across plan changes.

The Edge Case: What Happens When You Run Out Mid-Month

If your monthly credits hit zero, the Deep Analysis button still appears — it just prompts for a top-up before proceeding. AI auto-replies are completely unaffected. Your customers keep getting answered. The only thing that pauses is the analytical heavy-lifting.

I designed this deliberately: I never want a billing edge case to make a customer's inbox go silent. Auto-replies are mission-critical uptime — they don't touch the credit system.

Free Plan: 10 Credits on Signup, No Expiry

Every new OT1-Pro account gets 10 AI credits on signup, no credit card required. That's 10 Deep Analysis runs — enough to genuinely evaluate whether the insights are valuable for your business before you pay anything.

I added this because I hate tools that lock analytics behind a paywall before you can determine if they're worth it. Run the analysis. See what your AI found. Then decide if the subscription makes sense.

Budgeting AI Credits as a Business Expense

Here's how I'd think about this if I were the customer: a Starter account at $29/mo with 30 included credits = $0 per Deep Analysis run (amortized into the subscription). Extra top-up credits at $3 per 10 = $0.30 each. If one Deep Analysis run surfaces one missed opportunity — say, a loyal-customer segment of 40 people you haven't messaged in 90 days — and you send a re-engagement campaign that converts 20% at an average order value of $50, that's $400 in revenue from a $0.30 analysis.

The math isn't close. The only question is whether you remember to run it.

Upcoming: Automatic Scheduled Analysis

The team is working on a scheduler that will auto-run a Deep Analysis every Monday morning and email you a digest — no manual trigger needed. This would still consume 1 credit per run. The goal is to make sure the insights get in front of you even when you're too busy to log in and manually start it.

If you're on the Pro plan, with 100 credits per month, the weekly auto-schedule would use 4 credits per month — leaving 96 for manual runs.

The Single Rule That Governs All of This

You should never be surprised by an AI bill. If you are, the billing model is wrong. Credits exist so you know, before you click, exactly what you're authorizing — and so the expensive operations are always your choice, never automatic.

See how Deep Analysis works in practice before you decide if credits are worth it. Or compare the full picture at OT1-Pro vs WATI. Ready to try it? Start with 10 free credits.