Product Case Study · Artificial Intelligence and Self-Service
Adding Context-Aware AI Support to the Booking Flow
Anolla’s AI support does not operate as a generic chatbot detached from the booking software. The assistant connects the user’s question with the active page, user role, account status, and software rules to suggest a next step suited to the specific workflow instead of providing generic guidance.
This technical overview describes how Anolla uses context to transform AI support from a general information layer into a product-specific guidance tool.
The AI considers the active page, user role, account status, and product rules.
Number of Context LayersSupported languages within a single support logic.
Public Platform CapabilityA support layer integrated into the platform for service providers and customers.
Product Design PrincipleWhy Is the User’s Question Alone Not Enough for AI Support?
A software support response may be factually correct but useless to the user if it does not account for their role, open view, or currently active configuration. Anolla treats context as part of generating the response rather than as a detail added to the conversation later.
Challenge
The same question can have completely different underlying causes. “Why can’t I book?” may indicate a fully booked schedule, an advance notice restriction, an incorrect user role, an account status issue, or a rule configured by the service provider.
Design Decision
The assistant starts with the user’s language and query and, when necessary, considers the active page, role, account status, and relevant configuration.
Platform Outcome
The goal is not merely to answer the question, but to explain the cause, point to the appropriate setting, or direct the user to the correct self-service or feedback flow.
What context makes an AI response truly relevant?
A typical support bot searches for the help text closest to the question. For booking software, this may not be enough because the user's actual situation depends on several interconnected layers. A provider may change the schedule in the admin interface, a customer may try to make a booking on the public page, and an account owner and an employee may see different actions.
Even a technically identical question may require a different answer. A lack of available time slots may be caused by a resource schedule, service duration, buffer time, the maximum advance booking window, or another overlapping booking.
Active Page
Shows which feature or booking step the user is currently viewing.
User Role
Determines which actions the user can perform and which settings they can access.
Account status
May affect feature availability and the appropriate next step.
Provider rules
Determine availability, rescheduling, cancellation, and other booking logic.
The right answer depends not only on the question, but also on where it was asked, by which user role, and with what configuration.
What principles make AI support context-aware?
Context is not treated as one large data package. The assistant uses the layers needed to resolve the query incrementally and should limit itself to the minimum information actually required to provide accurate guidance.
01 · The active page provides the starting point
The assistant considers whether the user is on the booking page, in the calendar, in the service settings, in the account view, or at another stage of the workflow.
02 · Role and account status limit available actions
The guidance must reflect whether the question is being asked by a customer, employee, or account owner, as well as which features are available to them.
03 · Product rules come before explanations
The response must be based on booking logic, the current configuration, and documented features rather than on a plausible-sounding assumption.
04 · The response has boundaries and a next step
If the necessary context is missing, the assistant should ask for clarification, acknowledge uncertainty, or direct the user to the appropriate workflow.
How does the user’s situation lead to a context-aware response?
The assistant uses only the context required to resolve the specific question. The necessary information is combined with the product rules to provide an explanation, a next step, or, if necessary, a clarifying question.
Required context
Context engine
Identify · connect · verify
The question is connected to the active workflow, the user's available options, and documented product rules.
Appropriate response method
The conceptual diagram describes the logic behind generating a response, not the complete technical architecture of Anolla’s internal systems.
How does AI support adapt to different usage scenarios?
The assistant does not need to use all available data for every question. The required context depends on the decision the user is trying to make and the constraint that may affect their actions.
| Scenario | User question | Required context | Appropriate response |
|---|---|---|---|
| Availability issue | “Why can’t the customer see any available times for this day?” | Active service, resources, schedule, duration, buffer times, and booking restrictions. | Explain the most relevant restriction and indicate which setting to check. |
| Role and access | “Why can’t I see the company calendar or services?” | Selected account, active role, the user’s association with the service provider, and permissions. | Guide the user to select the correct account or role without assuming they lack access. |
| Modifying a booking | “Why can’t I reschedule or cancel my appointment?” | Booking status, service provider rules, time restrictions, and the user’s association with the booking. | Explain the applicable restriction and direct the user to an allowed action. |
| Unclear case | “Everything seems correct, but the action still doesn’t work.” | Clarification of the question, the visible error message, and the minimum required product context. | Do not assume a hidden cause; ask for clarification or direct the user to the feedback flow. |
How does a question lead to a verifiable next step?
A context-aware response is formed through a sequential decision flow, not through a single large query about the user and account.
Identify the user's language and active workflow
The assistant starts by identifying the language in which the question was asked and the page or feature the user is currently using.
Determine the intent of the query
The question is classified by purpose, such as configuration, availability, access, or modifying a booking.
Gather the minimum necessary context
The role, account status, or configuration is used only when the answer cannot be reliably narrowed down without it.
Map the situation to product rules
The situation is compared against documented features, booking logic, and currently applicable restrictions.
Provide an action-oriented response
The response explains the reason, identifies the setting to check, and guides the user to an appropriate next step.
Acknowledge limitations and escalate when necessary
If the cause cannot be verified, the AI should not invent a solution but should ask for clarification or direct the user to the appropriate workflow.
What characteristics result from a context-aware support architecture?
Context awareness does not mean that AI makes every decision on the user's behalf. The same product logic that the application uses to manage bookings and permissions also helps shape the explanation provided to the user.
Fewer generic instructions
The user does not need to work through a long list of possible causes when the active context allows the answer to be narrowed down reliably.
A workflow-specific next step
The response can direct the user to the correct account, setting, booking details, or feedback workflow.
Shared logic across languages
Multilingual support can rely on the same product rules instead of a separate process created for each language.
An extensible knowledge layer
When new features are added, the central product context can be expanded, benefiting both service providers and customers.
Usability of AI-generated response drafts in human support
52.4% of response drafts required no substantive changes
This metric describes AI-generated response drafts prepared for human support: technical support did not need to make substantive changes to 52.4% of the responses. It represents a human support workflow, not the share of automatically resolved conversations or an accuracy rate.
Source: Anolla's anonymized AI and support usage data. Last updated July 1, 2026. Method: AI-generated responses are compared with responses edited or manually replaced by technical support.
What exactly do the claims in this case study mean?
This is a product case study, not a customer story or a promise that AI will automatically resolve every support request. The page describes the design principles behind Anolla's context-aware support and the platform capabilities that have been publicly documented.
Basis of capability
Anolla's current AI support description, product workflows, and the principle of using the context of the active page and account.
Functional evidence
Typical test scenarios related to booking, setup, access, and account status.
Data usage
The case study does not use conversations from identifiable clients or individual performance metrics from any company.
Interpretation
The page demonstrates the structure and scope of integrated support, not flawless responses or a guaranteed automation rate.
Limitations. AI may misinterpret a question, rely on incomplete context, or require clarification. The assistant should not claim that it changed settings, accessed data, or identified the cause unless it has a verifiable basis for doing so. Account and user data must be used minimally and only to the extent necessary to resolve the request.