AI Use Policy

1. Scope of AI use

PulseForge configures AI for narrow, well-defined business tasks:

• Voice intake — answering inbound calls during overflow and after-hours, greeting callers, collecting job details, qualifying urgency, booking appointments where possible, sending summaries
• Missed-call follow-up — sending SMS text-backs to callers who couldn't be answered live
• Web enquiry response — AI-written initial reply to website enquiries
• Quote follow-up — drafting and sending scheduled follow-up messages using approved scripts
• Review collection — sending post-job review request SMS/email
• Old lead reactivation — sending re-engagement messages to dormant leads
• Admin support — summarising calls, tagging leads, routing escalations

AI is not deployed for: legal/medical/financial decision-making, fully autonomous transactions, persuasive sales of regulated products, or any task requiring a human licence or professional judgement.

2. Human fallback rules

Every AI voice and messaging system is configured with explicit human fallback rules, agreed during onboarding. Typical triggers for human escalation:

• Caller asks for a manager or human
• Emergency keywords (fire, gas leak, injury, urgent, ASAP, etc. — adjusted by industry)
• Complaint or refund request
• Request for regulated advice
• Caller appears distressed or agitated
• AI confidence in understanding drops below threshold
• Topic outside defined scope

Escalation typically sends an immediate SMS alert to the business owner (or nominated contact) with caller details and call recording. Critical industries (e.g. locksmiths handling lockouts, emergency plumbers) can have additional escalation paths.

3. Prohibited advice areas

AI configured by PulseForge does not, and is not permitted to, provide:

• Legal advice — including contractual, family, immigration, employment or commercial legal matters
• Medical advice — diagnosis, treatment, medication, symptom interpretation
• Financial advice — investment, tax, superannuation, retirement, insurance product recommendations
• Credit advice — loan suitability, credit product recommendations
• Migration advice — visa eligibility, application guidance
• Specialist trade advice requiring physical inspection or licensed assessment

Where callers ask about these areas, the AI is configured to acknowledge the question, decline to give advice, and offer to escalate to a qualified human or take a message.

4. Regulated industries

For clients in healthcare, finance, legal, migration, credit, NDIS or other regulated sectors, we apply additional controls:

• Stricter human-only handling of sensitive matters
• Tighter approved scripts with industry-specific compliance language
• Mandatory disclosure that the AI cannot provide regulated advice
• Default escalation for anything outside intake/booking/FAQ
• Higher confidence thresholds before AI proceeds
• Optional supervisor review of AI-generated communications before sending

We discuss compliance requirements during onboarding and may decline engagements where AI cannot be configured safely for the use case.

5. Call disclosure

Where required by Australian best practice or industry regulator guidance, AI voice agents disclose their AI nature to callers — typically at the start of the call. Example wording:

"Hi, you've reached [Business Name]. You're speaking with an AI assistant who can take your details and book your job. How can I help?"

Call recording disclosures are included where required by state/territory recording laws. We work with clients to ensure disclosure aligns with their jurisdiction and industry.

6. Script approval

Every AI script — voice prompts, SMS templates, email sequences — is approved by the client before going live. Scripts cover:

• Greeting and disclosure
• Qualification questions and order
• Booking and quoting language
• Pricing references (or refusal to give pricing)
• Fallback statements
• Escalation triggers
• Closing and summary

Script changes after launch also require client sign-off, except for minor corrections (typos, wording fixes) which we may make and disclose in the next monthly report.

7. Data handling by AI tools

AI providers process call audio, transcripts, message content and configuration to deliver and improve their services. Key points:

• Where supported by the AI provider, we configure settings to minimise data retention and disable training on your data
• Call recordings and transcripts are accessible to authorised PulseForge staff and the client
• We document the AI providers used for your engagement and their applicable terms
• You can request a sub-processor list at any time

For more detail, see our Privacy Policy.

8. Monitoring and quality control

We monitor AI performance on an ongoing basis:

• Sample call reviews to confirm scripts are followed and tone is appropriate
• Escalation rate monitoring — too few or too many escalations both indicate tuning is needed
• Confidence scoring on AI responses
• Misunderstanding/error pattern detection
• Periodic client review meetings (monthly for Full Stack clients)

Issues identified are reported transparently to clients and fixed. We do not hide AI mistakes — we surface them and improve the configuration.

9. Known limitations of AI

We're upfront about what AI cannot do well:

• AI voice can struggle with strong accents, fast speech, or poor call quality
• AI can occasionally misunderstand or miss nuance — fallback rules exist for this reason
• AI does not have human judgement or empathy; it follows configured scripts
• AI transcription accuracy varies — typically 85–95%, less for technical jargon
• AI providers occasionally have outages or service degradation outside our control
• AI cannot perform tasks requiring a physical presence, licence or in-person assessment

Setting expectations correctly is part of every PulseForge engagement. AI augments your business — it doesn't replace skilled humans.

10. Client responsibility

Clients are responsible for:

• Reviewing and approving all AI scripts and sequences before launch
• Reviewing critical AI outputs where required (e.g. legal/financial sectors)
• Ensuring customer consent for SMS/email outreach where required by Australian law
• Notifying us promptly of script changes, service changes, or compliance concerns
• Using captured leads ethically and lawfully
• Reviewing monthly reports and raising any quality concerns

11. Updates to this policy

We review this policy quarterly. Material updates are communicated to active clients by email. This page always reflects the current policy.

Questions about this policy: [email protected]. This is a general policy and doesn't constitute legal advice. Specific industry compliance requirements should be reviewed with a qualified legal professional.