Bring your own API key chatbot
A bring your own api key chatbot lets a team supply its own provider credentials for model requests. If your support team wants control over model choice and provider billing, the key setup needs to be clear before rollout. momo uses a model chosen through OpenRouter when hosted; bring-your-own-key is available on the self-hosted version.
Meaning and cost
What bringing your own AI model key means
Bringing your own key means that a team supplies provider credentials for model requests instead of relying only on a service's managed model arrangement. Teams may consider it for provider choice, account-level data policies, or visibility over provider billing. Those are reasons to evaluate the arrangement, not a promise that it will lower total costs.
For this product, the distinction is specific: the hosted version uses a model chosen through OpenRouter, while the self-hosted version can use a key the company brings. Provider support, key format, billing setup, and provider data handling depend on the integration and should be settled before deployment. No provider or model support should be assumed from the general BYOK term. Teams considering a bring your own key AI setup should ask the development team which providers are supported.
A provider may bill separately for model requests when a team uses its own account. That is distinct from the flat plan fee and included AI conversation allowance on hosted plans. Review the chatbot pricing model guide when comparing plan fees with provider charges. The exact provider billing arrangement for the self-hosted setup needs confirmation for the chosen setup.
How it works
From model choice to a supported answer
Treat key configuration as one part of the rollout, not the whole chatbot setup. Confirm the self-hosted integration's supported setup with the development team, then add business knowledge and test how a customer question reaches an answer or a person.
Plan
Choose the deployment route
Decide whether the hosted version, where the model is chosen through OpenRouter, or self-hosting fits the team's requirements. BYOK applies to the self-hosted version. For details on plan charges, see the chatbot pricing model guide.
The team knows which deployment and model arrangement it is evaluating.
Establish
Settle provider and key details
Ask the development team which providers and model APIs the self-hosted setup supports, what credentials it accepts, and how billing and key handling work. Do not infer a settings path or storage method.
Provider-specific requirements are understood before a key is entered.
Prepare
Add the business's support knowledge
Supply relevant content using a website crawl, PDFs, Word files, plain text, or Q&A pairs. This knowledge is the basis for customer support answers, separate from the provider account decision.
The chatbot has business material to retrieve for support questions.
Test
Follow a question through the answer flow
Use a realistic support question and see whether the reply is based on retrieved passages, checked against those sources, and cited when the system is confident. Also exercise the uncertain path to make sure the team can take over.
The team can inspect both a grounded answer and the route to human help.
The support workflow stays explicit
A model key changes the provider-account arrangement for self-hosting. The customer-support flow remains based on business knowledge, a confidence check, and a human inbox for questions the system cannot answer confidently.
Answers cite sources
Answers are grounded in business knowledge and include citations when the system is confident.
Uncertainty opens a ticket
Below its confidence line, the chatbot says it is not sure and opens a ticket for the support team.
Inbox on every plan
The helpdesk inbox gives teammates a place to handle conversations that need human support.
Flat hosted plan fee
Hosted plans have a flat fee and an included number of AI conversations, with no per-resolution fee or credit packs.
Questions to settle before using a key
The product behavior and the provider-specific setup are separate questions. These answers state what the support workflow does and identify the integration details the team still needs to establish.
What does bringing your own AI model key mean?
It means using a company-supplied model key with the self-hosted version. The hosted version uses a model chosen through OpenRouter. Which providers, model APIs, key formats, and account permissions are supported needs confirmation before setup.
Can I use a chatbot with openai api key?
Support for that specific provider and key setup has not been verified, so do not assume it is available or follow generic setup instructions. Ask the development team which providers and APIs work with the self-hosted version, and what configuration steps apply.
Does it still cite sources and hand conversations to a human?
Yes. It retrieves passages from the business's knowledge, drafts an answer, and checks that draft against those sources. When confident, it answers with citations; when it is not confident, it says so and opens a ticket for the team.
Who pays for model usage when a team uses its own key?
Provider billing for a self-hosted key depends on the provider account and confirmed integration, so establish the billing arrangement before rollout. This is separate from the hosted plan pricing, which is a flat fee with an included number of AI conversations and no per-resolution fee.
What should we check about key storage and provider data policies?
Ask the development team how keys are stored, transmitted, accessed, rotated, and deleted, and what account permissions are needed. Check the provider's retention, training, processing-location, and account-level controls directly; no particular policy or certification should be assumed. For related data-location considerations, review EU hosting and GDPR.
Try the support workflow with momo
Start free to explore the hosted workflow; the hosted model is chosen through OpenRouter, while bringing your own key applies to self-hosting.