Pricing and tariff setting: how to price your AI expert and what to offer clients

Getting your pricing and tariffs right is key to the profitability and sustainability of your AI product. Your pricing should reflect the high value your AI Expert brings to clients, while covering the real operating costs of AI reports and generating a profit. Remember, you are not creating just another chatbot, but a sophisticated "AI being" or "digital colleague" that has the potential to become an invaluable partner for your clients.

18.1 Understand the factors affecting your cost and pricing options

Before you set a price, it is important to understand what influences your operating costs:

A) The AI model (level of "intelligence") chosen for your instance: the BuddyPro platform allows you to choose from a variety of AI models that vary in their capabilities and cost per processing. The choice of model will directly affect the quality of your AI Expert's responses and your costs.

• 🧠 The "SMART" model (recommended for the highest quality answers): uses the most advanced language models available (e.g. models from companies like Anthropic). It provides the highest quality, deepest and most "human" conversations. It is the default model for most BuddyPro instances because it provides the best user experience.
  • Approximate cost per every message in conversation (user message and Buddy message both count separately): Approx. 0.08 USD.
• 💡 The "ECONOMY" model: Still a very capable model that offers an excellent compromise between quality of responses and operational costs. It may be suitable for tariffs where you want to offer a lower price, or if your AI Expert solves less complex tasks.
  • Approximate cost per message in conversation: Approx. 0.05 USD.
• 💲 "NANO" model (soon available): The most cost-effective option, ideal for situations where low price is the main criterion. Suitable if your AI Expert has to serve a very wide audience or solve simpler queries. It should be taken into account that the quality of the answers may be lower than with SMART or ECONOMY models.
  • Approximate cost per message in conversation: Approx. 0.03 USD.

You, as the instance owner, will be able to test which of the models is most suitable for your purposes and specific tariffs.

B) Set monthly message limit for the user:

For each tariff you define how many messages will be available to the user within their monthly subscription.

• Messages in both directions count: that is, both messages sent by the user and replies from your AI Expert. For example, if the plan includes 1000 messages, this could be 500 questions from the user and 500 replies from the AI.
• AI-initiated messages also count: if your AI Expert addresses the user himself (e.g. with a reminder or a proactive query, if you have this feature enabled), these messages also count towards the limit.
• Common limits can be e.g. 300, 500, 1000, or even more messages per month, depending on the expected usage intensity.
• Practical experience: most users usually do not use up their full monthly message limit (on average when using a limit of 1000 messages, they usually use only 30-50%). This is a good thing to take into account when calculating your average cost per user - it is usually not as high as the theoretical maximum would suggest.

C) Length and complexity of your System Prompt:

The System Prompt is the basic instruction for your AI Expert. If it is extremely long and complex (e.g., dozens of pages), it would be sent with every user query and may increase the cost of each individual message. Try to keep System Prompt as efficient as possible and keep key know-how in the sources folder.

D) Conversation history (long-term memory):

Your AI Expert remembers previous conversations with the user. The longer the user interacts with the AI, the more context the AI has and the richer the user profile. Processing more context can be slightly more expensive. Thus, a new user usually has cheaper first messages than a long-term user with a rich history.

note

TIP: How to find out your exact costs and statistics:

You can now use the /stats command as an owner or team member. This will show you a summary report containing:

• Total AI costs (including training costs).
• Average cost per report in your instance.
• The current status of AI credits.
• Statistics for the last week and month (number of messages and cost).
• Total revenue from linked payment systems (FAPI/Stripe).

You can also roughly calculate how many messages does an average subscriber use up per week/month by dividing the number of total messages in the (completed) week/month by the number of subscribers.

This way, you no longer have to rely on estimates, but can see real data about your instance's profitability.

18.2 How to determine selling price and margin - examples and recommendations

Your price should reflect the high value your AI Expert brings, while covering your costs and generating a profit.

• Positioning as a premium product: Don't be afraid to market your AI Expert as a premium service. Compare its value to the price of your in-person consultations, extensive online courses, or other expert products. For example, Pavel Riha successfully sells Online AI Strategist (built on the SMART model) for $140 per month.

• Consider offering longer plans (six months, a year) at a discounted price at the beginning. You can add monthly options later. This approach can help gain a core of loyal customers.

• Example sales price and margin calculation with 40-70% usage of the limit (for inspiration, adjust according to your situation and the AI model you choose):

  • SMART model, 1000 messages/month:

    

  • Variant STANDARD (ECONOMY model, 500 messages/month):

    

  • BASIC variant (NANO model, 300-350 messages/month):

    

• Discounts for longer subscriptions: Motivate clients to make a longer commitment by offering a discount when purchasing an annual or half-yearly plan.

18.3 Create different tariffs (if you want to offer more options)

You can choose to offer one main tariff, or create multiple subscription tiers that can vary in, for example:

• Monthly message limit (e.g. 300, 500, 1000 messages).
• Subscription length (monthly, quarterly, annual).
• Possible bonuses (e.g. access to an exclusive community, additional materials, priority support).

For each such plan, you will need to create a separate product and sales form in your payment system (FAPI or Stripe), and then link this individually to your BuddyPro instance (see point 19).

18.4 Extra message packages (a necessary addition to your offer!)

Your clients may occasionally exhaust their monthly message limit before renewing their subscription. For these situations, it is absolutely crucial to have the option of purchasing one-time extra message packages ready . Your AI Expert can automatically offer them this option.

• How extra message packages work:
  • These are one-time purchases (e.g., 100 extra message package, 300 extra message package).
  • Compared to messages from a monthly subscription, they have a major advantage: they do not expire at the end of the month, but carry over to future periods until the client has completely used them up.
  • They only start to be drawn after the client has exhausted his current monthly limit from the regular subscription.
• Package pricing: the price per message in the package being purchased may be slightly higher than the prorated price per message in the monthly plan. This encourages clients to stay on the regular subscription, which is more convenient for them.
• Implementation: you must also create a separate product and sales form for each extra message package in your payment system and link it to BuddyPro as a package type (see section 19.1.C).

18.5 Important command to inform your AI about the base price (for communication with users)

In order for your AI Expert to correctly inform the user about the subscription price (for example, when their trial is about to end or when they ask about the price directly), you need to "communicate" this information to them. The following command is used to do this, which is entered by you as the owner into a chat with your BuddyPro instance:

/setDefaultCost:{defaultCost}:{defaultCostCurrency}:{defaultCostPeriod}:{defaultCostPeriodCount}:{discountedCost}

• {defaultCost}: Your standard (non-discounted) price for the base subscription period.
• {defaultCostCurrency}: Currency code (e.g. CZK, EUR, USD).
• {defaultCostPeriod}: The unit of the period. Allowed values are: days, weeks, months, years.
• {defaultCostPeriodCount}: The number of units of the period (e.g. 1 for "monthly"; if you wanted to communicate e.g. "in 2 months", you would specify 2 and the period months).
• {discountedCost}: Optional parameter. If you specify it, your AI Expert will communicate that the default price is {defaultCost}, but that a temporary discount is currently available at {discountedCost}. If you do not offer a discount or do not want to communicate it in this way, simply omit this parameter (the command will end after {defaultCostPeriodCount}).

Examples of use:

• Example 1 (no discount, monthly payment):/setDefaultCost:2990:CZK:months:1Your AI Expert will inform the user that the subscription price starts at 2990 CZK per month.
• Example 2 (with discount, monthly payment):/setDefaultCost:3490:CZK:months:1:2990Your AI Expert will inform the user that the standard price is 3490 CZK per month, but is now temporarily available at the special price of 2990 CZK.

Important: This command does not affect the actual prices set in your payment system (FAPI/Stripe). It only serves to allow your AI instance to properly communicate basic pricing information to your users.

---

---

Frequently Asked Questions (FAQ) for this step:

• How many messages do users consume on average out of their monthly limit?
  • Experience shows that most users consume approximately 30-50% of their monthly limit of 1000 messages. The percentage can be higher at lower limits.
• Do unused messages from the monthly limit carry over to the next month?
  • No, messages from the regular monthly limit (within the subscription) are not carried over at the end of the billing period. Each month the limit is reset to the original value.
  • The exception to this is messages from purchased extra packages - these do not expire and carry over until they are completely used up.
• Can I change the AI model (SMART, ECONOMY, NANO) for my instance or for individual tariffs?
  • Yes, as an instance owner you will be able to change the default AI model for your entire instance using specific commands
  • If you would like to have different models for different tariffs, we are preparing this option. The default is that the entire instance runs on one selected model. For specific needs, we recommend consulting BuddyPro support to see if and how such a configuration can be implemented.
  • We recommend starting with the SMART model for best quality and possibly experimenting with other models if relevant to your strategy.
• What is the difference in cost per message for a new user versus a long-term user?
  • Costs are a bit lower for a brand new user because the AI doesn't need to process the user's extensive conversation history and context. For a long-term user with a rich history, the cost is slightly higher due to the need to search and use the accumulated context. In the longterm, this will average to the costs specified in 18.1
• What if a client uses up their monthly message limit?
  • Your AI Expert will automatically notify them when the limit is reached.
  • It will then offer him the option to purchase a one-time package of extra messages or upgrade to a higher plan (if you offer such options and have them linked). To do this, it redirects him to the URL you set with /setUpgradeSubscriptionUrl:{url}.
  • The user can ask your AI Expert at any time for the number of currently available/exhausted messages and it will tell him the current status
• How exactly are messages counted within the monthly limit?
  • The limit includes two-way communication: that is, both messages sent by the user and replies from your AI Expert are counted towards the limit.
  • Messages that your AI Expert initiates himself (e.g. automatic greetings, reminders, proactive queries) are also counted in the limit if you have this feature enabled(/shouldInitiateMessages:true). However, the user can tell their instance not to write to it by itself, which will limit these initiated messages.

---

---