Skip to main content

Going Live

Ready to launch your AI assistant? This guide covers everything you need for a successful launch.

Pre-Launch Checklist​

Configuration​

  • ✅ Assistant fully configured - Name, instructions, voice, greeting set
  • ✅ Phone numbers assigned - At least one active number
  • ✅ Testing completed - 5-10 test calls, all scenarios tested
  • ✅ Integrations working (if applicable) - Green checkmark, test bookings created
  • ✅ Knowledge base uploaded (if applicable) - Current, accurate information
  • ✅ Transfer tools configured - Transfer number works, humans ready
  • ✅ Team informed - Staff knows it's launching, can handle escalations
Critical

Don't skip testing or phone number assignment - these are the most common reasons for failed launches.

Launch Strategies​

Launch to a small group before going public.

Phases:

  1. Internal (1-3 days) - Team testing only
  2. Trusted customers (3-7 days) - 10-20 friendly customers
  3. Limited public (1-2 weeks) - One location or service line
  4. Full launch - All customers

Timeline: 1-4 weeks

Best for: First-time users, complex integrations, high call volume

Limited Hours Launch​

Start with specific hours only, expand gradually.

Schedule example:

  • Week 1: After-hours only (6pm-8am, weekends)
  • Week 2: Add lunch hours (12pm-1pm)
  • Week 3: Add morning rush (8am-10am)
  • Week 4: All hours

Best for: Testing after-hours coverage, replacing overflow/busy coverage

Parallel Running​

Run AI alongside humans with separate numbers.

  1. Keep existing number for human staff
  2. New number for AI assistant
  3. Track performance of both
  4. Decide next steps after 2-4 weeks

Best for: Testing before replacing humans, comparing performance

Full Launch (Advanced)​

Launch to everyone immediately.

Best for: Experienced users, simple use cases, low call volume, thorough testing completed

⚠️ Higher risk - only if very confident!

Announcing Your AI​

To Your Team​

Sample email:

Subject: New AI Assistant Launching [Date]

Team,

We're launching an AI assistant on [date] for [phone number].

It handles: [list main functions]
Transfers to: [number/person]

What to do:
- Test it by calling [number]
- Be ready for transferred calls
- Report feedback to [person]

Questions? Contact [support person].

To Your Customers​

Option 1: No announcement (seamless transition)

  • Just launch it naturally
  • Best for: Replacing IVR, adding after-hours, natural-sounding assistant

Option 2: Soft mention

Our new 24/7 phone line is powered by AI technology.
Call us anytime at [number].

Option 3: Full transparency

We've launched an AI assistant!

✅ Answers questions 24/7
✅ Books appointments in our calendar
✅ Transfers to our team when needed

Call [number]. Need a human? Just ask!

Update These Materials​

Digital: Website, Google Business, social media, email signatures, directories

Print: Business cards, brochures, signage, invoices (update as reprinted)

First Week Monitoring​

Monitor closely during days 1-7.

Daily Tasks​

Morning:

  • Check call volume
  • Review all transcripts
  • Note errors
  • Verify integrations working

Afternoon:

  • Quick check ongoing calls
  • Respond to feedback
  • Make urgent fixes

Evening:

  • Review day's performance
  • Note patterns
  • Plan improvements

Key Metrics​

Call metrics:

  • Volume, duration, completion rate
  • Transfer rate, drop rate

Quality metrics:

  • Transcript accuracy
  • Task completion
  • Caller sentiment (if you add it as a custom insight)

Integration metrics:

  • Successful bookings
  • Data accuracy
  • Connection stability

Common First-Week Issues​

High Transfer Rate (>20%)​

Fix:

  1. Review transfer transcripts - what's triggering them?
  2. Add missing info to instructions/KB
  3. Adjust transfer criteria
  4. Train on new scenarios

Misunderstandings​

Fix:

  1. Review patterns in transcripts
  2. Add common phrasings to instructions
  3. Try different transcription model
  4. Add examples of correct interpretations

Wrong Information​

Fix immediately:

  1. Update instructions with correct info
  2. Review all knowledge base documents
  3. Make instructions more explicit
  4. Test thoroughly

Integration Failures​

Fix:

  1. Check integration status (should be green)
  2. Reconnect if needed
  3. Verify service names match exactly
  4. Review integration logs

Making Adjustments​

For minor changes:

  1. Edit instructions
  2. Save (changes apply immediately)
  3. Test quickly

For major changes:

  1. Create duplicate assistant
  2. Make changes to duplicate
  3. Test thoroughly
  4. Swap phone numbers when ready

Best practices:

  • ✅ One change at a time
  • ✅ Test after every change
  • ✅ Track what changed and why
  • ❌ Don't make multiple changes simultaneously

When to Change​

Immediately: Wrong information, integration failures, major misunderstandings

Within 24-48 hours: Frequent minor errors, awkward phrasing

After more data: Voice adjustments, optimization, nice-to-haves

Success Metrics​

Operational​

  • 📞 Calls handled - Volume per day/week
  • ⏱️ Average duration - Usually 2-5 minutes
  • ✅ Completion rate - Target: >80%
  • 🔄 Transfer rate - Target: <20%
  • 📅 Appointments booked - Per day/week
  • ⏰ Time saved - Staff hours freed up

Quality​

  • 😊 Customer satisfaction - From your own surveys or reviews
  • ✍️ Data accuracy - Correct information captured
  • ✅ Task completion - Successfully completed actions
  • ❌ Error rate - Misunderstandings or failures
Set Baselines

Track week 1 metrics as your baseline. Aim for continuous improvement!

Handling Customer Feedback​

Negative Feedback Process​

  1. Acknowledge - Thank them, apologize, explain you're improving
  2. Investigate - Find call in analytics, review transcript
  3. Fix - Update instructions, add to KB, fix incorrect info
  4. Follow up - Test fix, monitor for similar issues

Common complaints:

ComplaintSolution
"Didn't understand me"Improve instructions, add common phrases
"Gave wrong info"Fix immediately, update instructions/KB
"Couldn't reach human"Make transfer more accessible
"Too robotic"Change voice or provider
"Too slow"Optimize conversation flow

Rollback Plan​

When to Rollback​

Rollback if:

  • 50% of calls have major issues

  • Giving consistently wrong information
  • Integration creating bad data
  • Multiple customer complaints per day

Don't rollback for:

  • Minor issues affecting <10% of calls
  • Issues fixable within an hour
  • Single confused customer
  • Optimization opportunities

Emergency Steps​

  1. Immediate: Turn off call forwarding with your phone carrier so calls ring your business line again
  2. Short-term: Investigate, determine if fixable
  3. Decision: Fix and re-launch, or keep using backup
  4. Recovery: Fix issues, test, re-launch softly

Long-Term Maintenance​

Weekly​

  • Review analytics
  • Read sample transcripts
  • Check for new patterns
  • Track KPIs
  • Make minor adjustments

Monthly​

  • Comprehensive performance review
  • Update knowledge base
  • Identify optimization opportunities
  • Gather team feedback

Quarterly​

  • Deep-dive analysis
  • Customer satisfaction survey
  • Explore new features
  • Report to stakeholders

Scaling Up​

Week 2-4: Optimization​

  • Reduce call duration
  • Improve completion rates
  • Decrease transfer rates
  • Enhance satisfaction

Month 2+: Expansion​

  • Add more phone numbers
  • Create specialized assistants
  • Add more integrations
  • Implement advanced features

Getting Help​

Contact Support When:​

  • Platform-wide issues
  • Integration not working after reconnecting
  • Phone number routing problems
  • Technical errors in logs

Have Ready:​

  • Assistant name/ID
  • Phone number(s) affected
  • Call examples (timestamps)
  • Screenshots of errors
  • What you've already tried

What's Next?​

  1. Keep improving - Use Best Practices
  2. Monitor closely - Check Analytics
  3. Scale up - Add more capabilities
Continuous Improvement

The best AI assistants continuously improve based on real customer interactions!

Need Help?​