AI Use & Limitations

Artificial intelligence and automation can help businesses respond faster, follow up more consistently and manage customer opportunities at greater scale.

However, AI is not infallible.

The purpose of this Policy is to explain:

• How ReviveX may use AI
• What ReviveX AI systems are designed to do
• What they are not designed to do
• Known limitations of AI technology
• When human involvement may be required
• Client responsibilities
• Data and privacy considerations
• Appropriate use of automated systems
• High-risk and prohibited uses
• Automated decision-making considerations

This Policy should be read together with the ReviveX:

• Service Terms
• Privacy Policy
• Data Processing Agreement
• Acceptable Use Policy
• Subprocessors & Data Handling

ReviveX uses artificial intelligence as a business tool.

The objective is not to use AI simply because the technology exists.

ReviveX systems are intended to use AI and automation where they can help businesses:

• Respond to enquiries
• Capture opportunities
• Understand customer intent
• Collect relevant information
• Qualify enquiries
• Route conversations
• Trigger follow-up
• Book appointments
• Reactivate dormant opportunities
• Update CRM records
• Support customer communication
• Reduce repetitive manual activity
• Maintain consistent customer journeys

Our principle is:

AI is a tool. The business outcome is the point.

AI should operate within an appropriate business process rather than as an uncontrolled substitute for judgement.

Depending on the Client's selected services and configuration, AI may be used within areas such as:

• Website conversations
• Inbound telephone conversations
• Lead qualification
• Speed-to-lead communication
• Database reactivation
• Reputation and review journeys
• CRM activity
• Conversation summaries
• Lead categorisation
• Enquiry routing
• Appointment booking
• Follow-up
• Internal workflow automation
• Information extraction
• Custom AI assistants
• Custom applications
• Business-process automation

Not every ReviveX service uses AI in the same way.

The exact functionality depends on the Client's implementation.

ReviveX solutions may use both:

A ReviveX workflow may use:

• AI only
• Automation only
• A combination of AI and automation

Depending on the implementation, a ReviveX system may be configured to:

• Answer common questions
• Ask qualifying questions
• Capture contact information
• Identify the nature of an enquiry
• Determine likely intent
• Collect information needed by the business
• Route an enquiry
• Create or update CRM records
• Schedule appointments
• Trigger follow-up
• Summarise conversations
• Categorise opportunities
• Escalate specified matters
• Support database reactivation
• Support reputation workflows
• Provide information from an approved business knowledge base

The exact permitted actions should be determined by the agreed Client workflow.

Unless expressly designed, tested and agreed for a particular lawful use, a ReviveX AI system should not be assumed to:

• Replace professional judgement
• Provide legal advice
• Provide medical diagnosis
• Provide clinical advice
• Provide financial advice
• Provide tax advice
• Make emergency-response decisions
• Guarantee factual accuracy
• Guarantee a sale
• Guarantee qualification accuracy
• Guarantee customer intent
• Guarantee business outcomes
• Understand every possible context
• Have perfect knowledge of the Client's business
• Have access to information that has not been provided or connected
• Know information that changed after its knowledge source was last updated

Artificial intelligence may occasionally:

• Misunderstand a question
• Misinterpret intent
• Produce an inaccurate response
• Produce incomplete information
• Misunderstand context
• Infer something incorrectly
• Repeat information
• Fail to recognise unusual phrasing
• Provide an outdated answer
• Generate information that sounds plausible but is incorrect
• Handle an unexpected situation poorly

These limitations are inherent in current AI technology.

The use of AI therefore requires appropriate:

• Configuration
• Testing
• Knowledge management
• Monitoring
• Human escalation
• Optimisation

where relevant.

ReviveX does not warrant that every AI-generated response will be:

• Correct
• Complete
• Appropriate
• Current
• Error-free

AI output should be understood within the context and risk level of the relevant workflow.

A low-risk question about business opening hours is different from a matter involving:

• Health
• Law
• Financial consequences
• Safety
• Contractual rights
• Material customer disputes

Higher-risk situations require stronger controls and appropriate human involvement.

AI performance depends heavily on the information available to the system.

The Client is responsible for providing accurate and current information where relevant. This may include:

• Business hours
• Services
• Products
• Pricing
• Policies
• Contact information
• Locations
• Qualification rules
• Escalation instructions
• Appointment rules
• Promotions
• FAQs
• Service areas

If Client information is incorrect, incomplete or outdated, the AI may rely on that incorrect, incomplete or outdated information.

Where appropriate, ReviveX may configure AI to rely primarily on Client-approved information or connected knowledge sources.

This can reduce unnecessary speculation but cannot guarantee that an AI system will never produce an incorrect response.

Clients should notify ReviveX when important business information changes.

ReviveX workflows may include human handover or escalation.

A system may be configured to escalate where:

• The customer asks to speak to a person
• The system does not have sufficient information
• A complaint requires human involvement
• An unusual situation occurs
• A high-value opportunity requires personal attention
• A sensitive matter arises
• The workflow reaches a defined escalation point
• The AI has low confidence
• A Client-defined rule requires escalation

The availability and method of human handover depends on the Client's implementation.

A handover feature does not necessarily mean that a Client employee is available immediately.

Human availability depends on:

• Client operating hours
• Staff availability
• Escalation procedures
• The communication channel
• Client response times

ReviveX should not represent a Client's staff as immediately available unless that availability has been confirmed.

ReviveX encourages appropriate human oversight where AI is used for activities involving material business, personal or professional consequences.

Depending on the workflow, human oversight may include:

• Reviewing escalated conversations
• Reviewing unusual outputs
• Reviewing important decisions
• Monitoring live conversations
• Reviewing lead qualification
• Updating knowledge
• Correcting workflow behaviour
• Reviewing complaints
• Reviewing regulated information

The appropriate level of oversight depends on the risk of the use case.

Automated systems can support business decisions.

However, particular care must be taken where automated processing could produce a decision that:

• Has legal consequences for a person, or
• Affects that person to a substantial degree

Where personal information is involved, such processing must comply with applicable data-protection law, including the requirements relating to automated decision-making under POPIA.

Standard ReviveX lead-conversion workflows should not be configured to make final, solely automated decisions with significant legal or similarly substantial effects on individuals unless:

• The processing is lawful
• The Client has identified the use case
• Appropriate safeguards have been considered
• Applicable rights of the Data Subject are respected
• Human involvement or review is provided where required
• The workflow complies with applicable law

Where applicable law requires safeguards relating to an automated decision, appropriate measures may include allowing the affected person to:

• Raise concerns
• Make representations
• Request appropriate review

and providing sufficient information about the relevant automated processing where legally required.

Clients intending to use ReviveX technology for material automated decisions must disclose that intended use during implementation.

ReviveX AI may assist with lead qualification.

For example, it may:

• Ask qualifying questions
• Identify stated needs
• Categorise an enquiry
• Identify urgency
• Identify location
• Determine whether criteria appear to be met
• Route a lead accordingly

AI qualification is intended to support business workflow.

It does not guarantee that:

• A lead will buy
• Information supplied by a lead is true
• The lead is financially qualified
• The lead meets undisclosed criteria
• The Client should accept or reject the lead

Material final decisions should be made according to the Client's appropriate business process.

Automations may create or update:

• Contacts
• Opportunities
• Pipeline stages
• Tasks
• Tags
• Notes
• Appointments
• Follow-up actions

Automated CRM activity depends on:

• Workflow configuration
• Information received
• Connected systems
• Integration availability

Clients should not assume that every automated CRM action is immune from technical or configuration error.

AI may assist with appointment scheduling where a calendar or booking system is connected.

Appointment availability depends on the connected calendar and configuration.

Clients remain responsible for:

• Maintaining accurate availability
• Blocking unavailable times
• Managing staff schedules
• Updating appointment rules

A technical or synchronisation issue may occasionally cause incorrect availability.

Where ReviveX provides AI-assisted telephone functionality, AI may:

• Answer calls
• Converse with callers
• Capture information
• Identify intent
• Route enquiries
• Book appointments
• Trigger follow-up
• Escalate calls

Voice AI may occasionally:

• Mishear a caller
• Misunderstand accents
• Misinterpret names
• Misinterpret numbers
• Be affected by background noise
• Produce transcription errors

Important information should be confirmed where appropriate.

Website AI may interact with visitors using conversational technology.

It may:

• Answer questions
• Capture details
• Qualify enquiries
• Schedule appointments
• Route leads
• Trigger workflows

Website users should not assume that every response is written by a human.

Where appropriate to the implementation, the interface should make the automated nature of the interaction reasonably clear.

Do not falsely represent an AI system as a specific real human employee.

AI and automation may be used to assist with re-engaging contacts in a Client database.

ReviveX does not guarantee that:

• A contact will respond
• A contact remains interested
• A contact's information remains current
• A campaign will produce sales
• A particular response rate will be achieved

The Client remains responsible for ensuring that the database may lawfully be used for the intended purpose.

Please refer to the:

• Service Terms
• Data Processing Agreement
• Privacy Policy

ReviveX technology may support review-request journeys.

ReviveX does not guarantee:

• Positive reviews
• Five-star ratings
• Removal of negative reviews
• Increased ratings
• A particular number of reviews

ReviveX should not configure workflows intended to:

• Fabricate reviews
• Impersonate customers
• Create false testimonials
• Manipulate reviews unlawfully
• Prevent legitimate customers from expressing genuine opinions

AI does not make an unlawful marketing campaign lawful.

Where ReviveX technology is used for:

• Email outreach
• SMS
• WhatsApp
• Automated calling
• Database reactivation
• Other electronic direct marketing

the Client remains responsible for determining whether the communication is lawful.

The relevant requirements of POPIA and other applicable law must be respected.

Clients are responsible for:

• Describing their intended use
• Providing accurate business information
• Identifying regulated activities
• Identifying high-risk workflows
• Identifying legally significant decisions
• Providing escalation rules
• Ensuring their own staff use the system appropriately
• Monitoring business-specific outcomes
• Informing ReviveX of material inaccuracies
• Ensuring that Client instructions are lawful

ReviveX cannot reasonably configure safeguards for a risk that the Client has not disclosed.

Additional care may be required where AI is used within:

• Healthcare
• Legal services
• Financial services
• Insurance
• Credit
• Employment
• Recruitment
• Education
• Property
• Other regulated sectors

Clients remain responsible for identifying applicable industry requirements.

ReviveX technology does not replace an appropriately qualified professional.

Unless expressly designed and lawfully agreed for an appropriate professional use, ReviveX AI must not be relied upon as a substitute for:

• Medical diagnosis
• Clinical judgement
• Legal advice
• Financial advice
• Investment advice
• Tax advice
• Accounting advice
• Regulatory advice

Where a conversation enters such an area, the workflow should be configured to provide an appropriate limitation or human referral where relevant.

Standard ReviveX systems are not emergency-response services.

They should not be relied upon as the sole channel for:

• Medical emergencies
• Threats to life
• Criminal emergencies
• Fire emergencies
• Immediate safety threats
• Crisis intervention

Clients whose businesses may receive emergency or urgent requests should provide appropriate emergency instructions and escalation procedures.

AI systems may process personal information when necessary to perform an authorised workflow.

Examples may include:

• Name
• Contact details
• Enquiry information
• Conversation content
• Appointment information
• CRM information

Processing of personal information is governed by the ReviveX:

Privacy Policy

and, for Client-controlled data:

Data Processing Agreement

Clients must inform ReviveX where a proposed AI workflow may involve:

• Health information
• Biometric information
• Religious or philosophical information
• Political information
• Trade union information
• Sexual information
• Criminal-behaviour information
• Other special personal information regulated by applicable law

Additional safeguards or restrictions may be required.

ReviveX may decline a workflow where the risk cannot reasonably be managed.

Clients must identify where a proposed AI workflow is expected to process personal information relating to children.

The workflow should not proceed until:

• The legal basis has been considered
• Appropriate safeguards are identified
• The intended use is appropriate

ReviveX may rely on external AI and technology providers.

The exact providers may vary according to:

• Service
• Functionality
• Client requirements
• Technology availability
• Integration architecture

Third-party providers may:

• Update models
• Change functionality
• Change pricing
• Change technical behaviour
• Experience outages
• Change service terms

ReviveX cannot guarantee that independent third-party AI technology will remain unchanged.

Do not list specific AI model providers on this page unless they have been verified as part of ReviveX's current service architecture.

Current verified providers, where appropriate, should be maintained on the ReviveX:

Subprocessors & Data Handling page.

AI models may change over time.

An updated model may behave differently from an earlier version.

ReviveX may therefore:

• Test updated models
• Adjust prompts
• Modify workflows
• Change model configurations
• Change providers
• Add additional safeguards

where reasonably necessary to maintain or improve the service.

Material changes affecting a Client's agreed workflow should be communicated where appropriate.

ReviveX may use:

• Prompts
• System instructions
• Business rules
• Knowledge sources
• Guardrails
• Workflow logic

to guide AI behaviour.

These measures can reduce unwanted behaviour but do not guarantee perfect compliance by an AI model.

Users may attempt to manipulate AI systems through malicious or misleading instructions.

This is sometimes referred to as:

• Prompt injection
• Jailbreaking
• Adversarial prompting

ReviveX may implement safeguards intended to reduce these risks.

However, no safeguard can guarantee that every possible manipulation attempt will fail.

Systems should therefore avoid exposing:

• Credentials
• Confidential system instructions
• Security information
• Protected Client information

merely because a user requests it.

Clients should consider carefully which confidential information is made available to an AI workflow.

AI should receive only information reasonably required to perform the intended task.

Clients should not unnecessarily place:

• Passwords
• Financial credentials
• Authentication secrets
• Private keys
• Highly sensitive confidential information

into conversational AI knowledge or prompts.

AI workflows should seek to collect only information reasonably necessary for the identified business purpose.

Do not ask users for personal information merely because the technology is capable of collecting it.

Do not make a blanket claim on this page that no third-party AI provider ever uses, retains or processes data for model improvement unless that has been verified for every applicable provider and service configuration.

ReviveX's handling of Client-controlled personal information is governed by the Data Processing Agreement.

Provider-specific data handling should be described accurately on the Subprocessors & Data Handling page where verified.

ReviveX will take reasonable technical and organisational measures appropriate to its role and the relevant AI implementation.

No AI or internet-connected system can be guaranteed to be:

• 100% secure
• Hack-proof
• Error-proof
• Immune from misuse

For additional information, refer to:

Subprocessors & Data Handling

and:

Data Processing Agreement

AI workflows may require optimisation after launch.

ReviveX may review relevant system behaviour for purposes such as:

• Identifying errors
• Improving workflows
• Updating knowledge
• Improving routing
• Improving qualification
• Addressing recurring misunderstandings

Any processing of Client-controlled personal information for these purposes remains subject to applicable contractual and privacy requirements.

Clients should report material AI errors to ReviveX when identified.

Where possible, include:

• The conversation
• Date and time
• Channel
• Expected behaviour
• Actual behaviour

This information can assist with troubleshooting and optimisation.

AI functionality does not guarantee:

• Sales
• Revenue
• Profit
• Lead quality
• Conversion
• Appointment attendance
• Review scores
• Customer satisfaction
• Business growth

ReviveX provides technology and workflows intended to improve business processes.

Actual commercial outcomes depend on many factors outside ReviveX's control.

ReviveX may restrict, pause or disable an AI function where reasonably necessary because of:

• Security concerns
• Unlawful use
• Material misuse
• Provider restrictions
• Serious technical problems
• Compliance concerns
• Risk to users
• Risk to Client data
• Risk to third parties

Where reasonably appropriate, ReviveX should inform the affected Client.

Clients must not intentionally use ReviveX AI to:

• Commit fraud
• Impersonate another person unlawfully
• Generate malicious software
• Conduct unlawful surveillance
• Harass individuals
• Discriminate unlawfully
• Facilitate criminal activity
• Violate privacy rights
• Circumvent lawful opt-outs
• Send unlawful spam
• Deceive people about material matters
• Fabricate customer reviews
• Obtain credentials through deception
• Expose another Client's confidential information
• Violate applicable law

More detailed restrictions appear in the ReviveX Acceptable Use Policy.

ReviveX expects AI systems to be deployed in a manner that is:

• Lawful
• Purpose-driven
• Proportionate
• Appropriately transparent
• Privacy-conscious
• Secure
• Subject to reasonable human oversight

The level of control required should increase as the potential consequence of the AI's action increases.

ReviveX supports reasonable transparency around automated interaction.

Where appropriate to the channel and context, users should not be intentionally misled into believing that an automated AI system is a real human employee.

This does not require every automated action to include unnecessary technical detail.

The level of disclosure should be appropriate to the interaction and applicable requirements.

ReviveX AI assistants may use a conversational name or persona.

A conversational identity is used to make interaction easier and more natural.

It must not be used to impersonate a specific real individual without appropriate authority.

Where asked directly whether the system is automated or AI-powered, it should respond truthfully according to its configuration.

Liability relating to use of ReviveX AI is governed by:

• Applicable law
• The ReviveX Service Terms
• The applicable Client Service Agreement

Nothing in this Policy excludes liability that cannot lawfully be excluded.

This Policy does not create a separate arbitrary financial liability cap.

Artificial intelligence develops rapidly.

ReviveX may update this Policy to reflect:

• New capabilities
• New limitations
• New providers
• New legal requirements
• Changes to our services
• Changes to responsible AI practices

The latest version will be published on this page.

Questions about AI use within a ReviveX system can be directed to:

ReviveX International
Christopher Parry Photography (Pty) Ltd

Registration Number: 2015/179935/07

Physical Address:
626 32nd Avenue
Villieria
Pretoria
Gauteng
0186
South Africa

Email: chris@revivex.co.za

Telephone: +27 84 511 7736

Website: revivex.co.za

This AI Use & Limitations Policy should be read alongside the following documents:

• Privacy Policy
• Website Terms of Use
• PAIA Manual
• Cookie Policy
• Service Terms
• Data Processing Agreement
• Acceptable Use Policy
• Subprocessors & Data Handling