Support agents,a fixed amount per month
Your support agent already answers customers in the evening and on weekends, from their history and your help articles, and what it costs depends on the number of conversations and their length. With Agent Libre behind it, the model that writes its answers is served in Canada for an amount written in the contract. Your flows, your handoff rules and your console stay as they are.
Talk about your customer serviceWhat breaks when the support agent is billed by usage
Three situations the support team recognizes after a few months in production.
Monday morning after an outage
A service interrupted on Saturday night fills the conversation queue. The agent answers all of them at once, and answers well. Billed by usage, that single weekend costs more than the previous month, and the support budget finds out on the next invoice. The holiday season and recall campaigns produce the same effect, on known dates.
The customer pastes whatever is at hand
To move faster, a customer pastes a full invoice, an account number or the sixteen digits of a card into the chat box. All of that text goes to the model. If the model sits with a foreign provider, that information has left the country before anyone could remove it.
One total on the invoice, no detail
The model provider’s invoice gives one amount for the month. It does not say what chat, email and ticket triage each consume, or which channel doubled. The support lead cannot decide what to automate first, because they do not know what each channel costs.
You, your tool, and the model that answers it
Your tool stays the same. What changes is the model that answers it.
Your organization
Your teams, your data and your rules. You decide what the tool does, and with which data.
Your support agent
The agent that answers your customers by chat or email, evenings and weekends included. Your support platform or your team puts it in place and decides when it hands off to a person.
The model that answers the tool
Every time the tool needs an answer, it calls a language model. That model can come from different providers. This is where Agent Libre comes in.
What Agent Libre changes behind your tool
Your support platform calls a language model to write every answer. That model can come from Anthropic, OpenAI or Agent Libre. Agent Libre serves that model. The platform itself stays yours.
The same flow, a different model behind it
Your flows, your handoff rules and your help articles stay in your platform. Your vendor or your team replaces the address of the model it calls with Agent Libre’s. Human agents see no difference in their console.
A queue instead of a surcharge
Agent Libre reserves a processing speed for your organization. When every conversation arrives at once, answers take a few seconds longer instead of costing more. You choose that speed from the measurement taken during the 24-hour to one-week audit.
One account per channel
Agent Libre’s gateway logs every call with the application that made it. You see what chat, email and ticket triage each consume, separately, and you can set a ceiling on each. The detail missing from the provider’s invoice is here.
Transcripts in Canada
The model is hosted and operated in Canada. The text of each conversation is processed, then forgotten: nothing is kept beyond processing and nothing is used to train a model. The archives are kept by your platform, under your own retention periods.
Three rules of the plan
Twelve months
The plan is written in the contract for twelve months. It is revised at renewal from measured consumption, up or down.
Re-reading is included
The customer history and the articles the agent re-reads before every answer are part of the plan, with no volume limit under fair use.
24 hours of measurement
The free audit places a gateway in front of your current provider and measures what your agent really consumes, channel by channel, before any commitment.
What each team decides before connecting the tool
Changing the model behind the agent takes a few minutes. The decisions below take longer and are better made first.
Decide which types of requests the agent can close on its own (order tracking, address change) and which go to a person (complaint, refund, cancellation), and write it into your platform’s flow before changing the model behind it. The model does not make that decision for you.
Check that your platform lets you choose the model it calls; some impose one. Then gather a hundred real, already resolved conversations and have both models answer the same requests for a week, to compare tone and accuracy on your own cases before switching.
List what the agent needs to read to answer (order number, product, purchase date) and what it does not need (card number, date of birth, full address). The gateway masks the second list before sending to the model. That list is what documents the processing under Quebec’s Law 25.
Take the busiest month of the last twelve to choose the reserved speed. A plan that is too tight slows answers at peak; a plan that is too large is corrected at renewal, downward.
What we get asked on this topic
Does Agent Libre provide the support agent?
No. The agent is put in place by your platform vendor or by your team, with its flows and handoff rules. It calls a language model to write its answers. Agent Libre serves that model in Canada for a fixed amount per month.
Does the model answer as well as the one we use today?
On everyday support requests, answer quality depends mostly on the instructions and articles your platform gives the model. The only test that counts runs on your conversations: during the audit, have both models answer the same requests and compare before switching.
What happens if volume doubles, at peak or for good?
At peak, answers wait their turn for a few seconds when the reserved speed is reached, with no cut-off and no charge. If volume doubles for good, the plan does not move before renewal; at that point it is revised from measured consumption and you know the new amount before signing.
What does the model see of a conversation?
It sees the text your platform sends it, after passing through the gateway that masks the fields on your list. It has no access to your customer database or your archives. Nothing is kept beyond processing and nothing is used to train a model.
What does our support platform need to allow?
One thing: letting you choose the model it calls. The connection then comes down to three lines of configuration, the address, the key and the model name. For Zendesk, Intercom, Freshdesk or a home-built platform, the connection point is checked during the 24-hour to one-week audit.
See also
Autonomous AI agents
What an agent that re-reads its history on every turn costs, and what changes when the model behind it is served in Canada for an amount known in advance.
Financial services
Banks, insurers, credit unions and fintechs: what compliance, privacy and finance decide before putting Agent Libre’s model behind their AI tools.
The flexibility premium: what the right to consume at any time costs
Why paying by usage means paying for the peak, and what a plan written in the contract changes.
Your company has two supply chains
The first one you already know. Materials come in, they are transported, stored, transformed, delivered. Every step has a supplier, a lead time, a unit cost. When the margin tightens, you walk down that chain step by step, and you generally know where to look.
The second one moves information instead of materials. A price request comes in and becomes a quote. An order comes in and becomes a production order, an invoice, a confirmed delivery. A site report becomes a change order, or it does not. This chain runs through every one of your departments and it employs a large share of your people.
The 24-hour to one-week audit
We measure your actual consumption at your current provider: who calls what, what it costs, and what is leaving your walls. Free, no commitment, and the report is yours even if you go no further.