What determines the cost
Two solutions with the same label can have very different costs. An assistant answering approved FAQs is not the same as an agent that checks reservations, updates records and communicates through several channels.
The budget depends on the number of processes, properties, languages, users and integrations, along with data quality, security requirements, availability and support.
Pilot, production and scale
A pilot tests one hypothesis with limited scope, data and users. Production adds authentication, monitoring, error handling, support, documentation, security and integration into daily work.
Hotel groups also need permissions, catalogues and rules for each property. Comparing prototype prices alone ignores what makes the service reliable.
Recurring costs
Beyond development, costs may include model usage, hosting, databases, communication tools, integration licences and maintenance. Estimate usage from real volumes, not from a demonstration with a handful of conversations.
Allow for content review, process changes and updates to supplier APIs.
Preparing for a comparable proposal
Describe the current process, monthly volume, people involved, systems, frequent exceptions and intended result. Ask suppliers to distinguish discovery, pilot, production, licences and maintenance.
BigLearn’s initial assessment determines whether AI is appropriate and defines a first use case before confirming schedule and investment.
Frequently asked questions
Can a price be given without knowing the hotel?
Only a very broad range. A useful estimate requires information about the process, integrations, volume and risk.
Is a chatbot always the cheapest option?
Not necessarily. It may be inexpensive to demonstrate but costly to maintain when information is disorganised or many integrations and exceptions are involved.
How should proposals be compared?
Compare scope, data, integrations, security, support, recurring costs, acceptance criteria and ownership of delivered components.