The True Einstein AgentForce Pricing

If you are thinking of purchasing Einstein AgentForce or Co-Pilot that uses the Chat GPT Large Language Models (LLM); the next 2 minutes reading will give back the highest money saving you will earn this year.

To get Einstein AgentForce you have to purchase Einstein Service for an additional ￡60 or $75 per user per month

( https://www.salesforce.com/uk/service/ai/pricing )

However, most customers will have an additional charge. This is because many Einstein services use API calls made through the LLM gateway and this consumes what are called Einstein Requests. It seams Salesforce is being purposefully opaque about this part of the pricing, partly because calculating the cost is quite complicated as it is dependent on consumption levels, and because, the author believes Salesforce does not want customers to know the likely cost of AgentForce.

Using many of the?Einstein features will include both an input API call to the LLM and an API call for its response. Therefore, each Einstein feature request will consume 2 API calls. This now needs to be converted into an Einstein Request value.

Einstein Request usage is calculated for each API call made through the LLM gateway to a Large Language Model (LLM). Consumption calculations involve:

?

1. The usage type multiplier associated with the type of LLM used, as described in the Einstein Generative AI Rate Card.
2. The API call size factor, which is the combined size of the prompt and the LLM-generated response, in words, divided by 1,500, and then rounded up to the next integer. There are no fractional API call size factors.

?

For example, if an API call of 3,000 words uses an LLM with a usage type multiplier of 10, it consumes 20 Einstein Requests. I told you the calculation is complicated; it is quite understandable that Salesforce is not publicising this cost because it would be difficult to communicate in simple form. However, the author believes this cost will be by far the biggest part of the cost of AgentForce for most companies.

Each license of Einstein comes with a set amount of free Einstein Requests. Don't get confused with the free Einstein Requests provided by Einstein Co-Pilot, which are two separate things. I spoke to many people at Salesforce to find out what the free amount will be, but received no definitive answer.

Once you know how many Einstein Requests you have used; if this is more than your allowed monthly usage, there will be an additional charge above your monthly cost.

It is easy to miss, but the 1 line of text in the image below is the most important information published by Salesforce, and unless you know exactly where to go to view this information, you might not realise the true cost of Einstein.

?

Salesforce states "Pre-loaded AI Requests and Data Cloud Credits. Additional AI Requests and Data Cloud Credits purchased based on business need."

The above text from Salesforce is quite non-descript and opaque, since it doesn't state how much each??Request and Data Cloud Credit cost and how they are calculated. Only by researching many different areas of Salesforce can you find this information, and still, you will not be able to calculate your expected cost. We have done as much of the research we can for you.

If you made the Einstein call in our example and have consumed 20 Einstein Requests, this on its own wouldn't cost you anything additional in your monthly cost. However, you are not going to use Einstein to make 1 call per month.

Let's say 1 user makes on average 2 calls per hour, and on average 12 calls per day. 1 user would consume 12 x 20 days x 20 = Each user consumes on average 480?Einstein Requests per month at a conservative level of only 2 calls per hour. Unfortunately, this is where our research hit dead ends, and if anyone does find the answer to the true cost, please enter a comment to this article. We could not find details stating how to convert the number of Einstein Requests into monetary cost.

What I have heard from prominent MVPs is that the cost of each AgentForce request, consumed each time the customer presses the "Enter" button to ask Einstein AgentForce a question or provide an answer, will cost $2 to $3. If that is true, to fulfil a customer inquiry, say for an average chat, both asking questions and answering, the customer presses the "Enter" button on a chat 10 times; the cost of closing each customer support inquiry could be between $20 to $30.

You may conclude that it is expensive, but remember AgentForce should be concluding customer support inquiries far quicker, and therefore, there could be an additional benefit of improved customer relations and reputation. Also, there will be a saving of office space and everything else that goes with that, plus AgentForce will be able to support customers 24 hours a day.

Let me now introduce SES and how this fits in with AgentForce. SES is an AI app, built natively on Salesforce. SES learns code of a customer's system and from a community of connected systems, and leverages code of these systems. Most importantly SES generates new code, referencing existing code. Our unique AI does not increase code duplication, defects, or code churn, that is experienced by Github Co-pilot and other AI's, substantiated in a study by GitClear.

Customers will truly realise the benefits of AgentForce by utilising the full arsenal of their existing Apex functions and Flows. This is achieved by creating Agent Actions that are serviced by Apex or Flows, see below.

?

SES AI compliments AgentForce by building new AI generated Apex functions whilst utilising and referencing your existing Apex functions. AgentForce is the ideal AI tool to communicate with the user, whilst SES is ideal for creating the logic to serve AgentForce.

If you are interested to learn more or discuss anything connected to AI code generation, contact me at [email protected], or visit our website at www.selfevolvingsoftware.com

?