Lorikeet FAQ

What makes Lorikeet different from other AI support tools?

Lorikeet is built for complexity, not just deflection. While many AI support tools only use simple RAG (retrieval-augmented generation) to answer FAQs, for more complex issues Lorikeet also executes multi-step workflows that access customer data, make decisions based on business rules, and take actions within your systems. Our AI is self-aware, meaning it knows its limits and escalates when uncertain rather than guessing an answer. We use instruction-based training allowing you to define complex workflows explicitly, rather than black-box self-training which can make unpredictable changes and is impossible to audit.

How does Lorikeet handle tickets it can't solve?

Lorikeet automatically escalates to human agents when it can't resolve an issue—whether it can't find a matching workflow, lacks relevant information, detects sensitive topics, or the customer requests a human. When escalated, agents receive complete conversation history, customer data, and context about what was already tried, allowing seamless handoff without customers repeating themselves. You're never charged for escalated tickets, only successful resolutions. Escalations can be configured to route automatically to your default team or to specialized teams based on issue type, with full audit trails showing exactly why each escalation occurred. 

Can Lorikeet handle regulated use cases for AI in CX?

Yes. Lorikeet is production-ready for high-stakes environments with complex compliance requirements. We support regulated industries including healthcare (HIPAA, HITECH), financial services (SOC 2), and insurance including claims processing and policy quotes. Customers with published stories include digital health companies Eucalyptus and easykind and financial services companies Taptap Send, Hnry and Flex. The platform includes built-in compliance workflows, comprehensive audit trails, and role-based access controls designed for regulatory requirements. You can read more about our case studies here or access our compliance and trust center here.

Does Lorikeet offer tools for internal support teams?

Yes. Lorikeet Coach is an internal AI agent that helps CX teams analyze and improve their support operation. Coach evaluates 100% of conversations (human, AI, or hybrid), diagnoses why metrics move, and for Lorikeet customers, implements fixes automatically. Key capabilities include automated QA with customizable quality scoring, conversational analytics you can query in natural language, root cause diagnosis when metrics change, and auto-generated test scenarios based on your actual ticket patterns. Coach conversational intelligence is included free for Concierge subscribers. Automated QA (Ticket Quality Score) is available as an optional add-on for 100% ticket coverage. For standalone Coach pricing, contact us.

What integrations does Lorikeet support?

Lorikeet offers built-in connectors for major ticketing systems including Zendesk, Intercom, Salesforce, HubSpot, Front, Help Scout and Kustomer. Helpdesks without a built-in connector, such as Freshdesk, connect through Flexible Ticketing as long as they send webhooks. For business systems, you can connect via webhooks (recommended for flexibility) or direct APIs. There's no migration needed as Lorikeet works alongside your existing tools. You can configure read-only or write access and toggle integrations on or off as needed. You can read more about integrations here

Is Lorikeet a helpdesk, or does it work with ours?

No. Lorikeet is not a helpdesk. It works inside the tools your team already runs, like Zendesk, Intercom, Salesforce and HubSpot, and resolves issues by taking action through your own APIs. No migration, no new helpdesk, no rip-and-replace.

Your team keeps its ticketing system and its workflows. Lorikeet resolves conversations end-to-end and hands over to your agents inside the same helpdesk when a human is needed.

Who owns our data, and what happens to it if we leave?

Your data stays yours. You own everything you put into Lorikeet and everything it generates from that data. We use it only to run your service, and we never use it to train AI models. Your ticket history, outcomes and quality scores can be exported, and if you leave, we delete your data within 60 days of your request.

These commitments are set out in our public Cloud Services Agreement.

What happens during onboarding?

Onboarding is connecting and testing, not migrating. You connect the helpdesk you already run, train Lorikeet on your existing help docs and policies, and test it on your own historical tickets before any customer sees it. Time to first live ticket is as little as hours, with an average under 29 days, and every plan includes forward-deployed support. You can also start with a 30-day free trial, where Coach, Lorikeet's setup agent, guides you through your first concierge.

The steps typically include: creating your account and configuring user roles, connecting your ticketing system, uploading and classifying reference materials, setting AI style and tone guidelines, configuring channel-specific settings, reviewing security settings including PII redaction, testing responses before going live, and establishing monitoring and improvement processes.

How long does implementation take?

Lorikeet's time to first live ticket is as little as hours, with an average under 29 days. The timeline depends on your internal systems and use case complexity. Companies with well-documented processes and straightforward integrations see measurable improvements within a week. We take an incremental training approach so you don't need perfect documentation before starting. We test thoroughly before production deployment.

Do you need perfect documentation before starting?

No. We take an iterative, coaching-based approach rather than requiring comprehensive documentation upfront. You can start with existing help center articles, FAQs, and policies, then continuously add content as you identify gaps. Our platform helps you discover where documentation is missing by showing you which inquiries the AI couldn't handle.

Does Lorikeet use their own models?

No, we use multiple leading third-party AI providers. This multi-provider architecture lets us use the best models for different use cases and continuously adopt the latest AI improvements. Your data isn't stored by providers or used for training.

How reliable is Lorikeet when AI providers have outages?

We use multiple leading third-party AI providers with automatic failover between them. If one provider experiences issues, we automatically switch to another, ensuring continuity.

How do we train Lorikeet's AI agent?

Training uses two approaches:

Reference Material Training: Upload help center articles, FAQs, documentation, policies, and guides manually or via our scraper. The AI uses RAG to search this content and answer straightforward questions.

Workflow Training: Create custom workflows with matching instructions, example questions, and specific steps including API calls, conditional logic, and variable passing. This handles complex, multi-step processes.

Subject matter experts review and approve workflows before deployment.

Can we control exactly how the AI responds?

Yes. You have extensive configurability including custom workflows with explicit steps, guardrails for what the AI can and cannot do, tone and style guidelines by channel, escalation thresholds based on confidence levels, and testing capabilities against historical tickets before going live. Every workflow is transparent and modifiable. You control which scenarios allow automatic responses versus requiring human review.

What safeguards does Lorikeet have to prevent incorrect information?

Lorikeet uses multiple safeguards: grounded responses based on your documentation and data, validation checks before providing answers, confidence scoring that escalates when uncertain, structured workflows for critical information using deterministic automation, limited LLM scope focused on specific tasks, and human review loops for high-stakes scenarios. Lorikeet doesn't speculate. If information isn't in your knowledge base, it says so rather than guessing. Task compartmentalization gives LLMs only specific focused tasks with clear inputs/outputs, reducing hallucination risk.

What happens when the AI makes a mistake?

Lorikeet provides observable decision-making with full audit trails showing exactly what the AI did and why. You can mark tickets as bad (no charge), and we use these errors to improve through continuous feedback loops. Our team works with you to analyze patterns and make improvements – AI doesn't auto-update without human oversight. All errors are logged, escalate immediately, and trigger alerts via Slack or on your dashboard.

How does Lorikeet's pricing work?

You pay an annual plan plus a rate for each ticket Lorikeet resolves, and escalations to a person and unresolved tickets are not charged. Start is $2,100 a month and Scale is $5,100 a month, both paid annually and both including the platform and a forward deployed support package. Resolutions cost $0.99 (Start) or $0.90 (Scale) per chat, email or SMS ticket, and $1.50 or $1.20 per voice call (based on a 3 minute average). Signature is custom-priced and includes a $75K platform fee. For example, a Scale customer resolving 5,000 chat tickets a month meters 5,000 x $0.90 = $4,500 in resolutions on top of the plan. There are no per-seat charges and no cost for failed attempts. You can mark any ticket as good or bad, and you're not charged for tickets you mark bad. Current rates are on the pricing page.

For Concierge subscribers, Coach conversational intelligence is included free. Automated QA (Ticket Quality Score) is available as an optional add-on at $0.30 (Start) or $0.27 (Scale) a ticket for 100% ticket coverage. Every plan can start with a 30-day free trial: you add a card at checkout and nothing is charged until the trial ends. For standalone Coach pricing, contact us.

Why is per-conversation pricing bad for buyers?

Per-conversation pricing creates misaligned incentives. Vendors get paid whether they solve your problem or not, which means they're motivated to maximize conversation volume rather than resolution quality. You end up paying for failures, poor responses, and tickets that still need human follow-up. With per-resolution pricing, the vendor is incentivized to actually solve problems, not just generate activity.

What ROI should we expect?

Your return is the gap between what a ticket costs you today and the per-resolution rate: $0.99 (Start) or $0.90 (Scale) per resolved chat, email or SMS ticket, plus the annual plan, with escalations not charged. Lorikeet reduces handle time for tickets it resolves, provides 24/7 coverage without additional headcount, and handles higher complexity tickets than competitors (not just simple FAQs). Typical customers see significant reduction in average handle time, resolution of complex multi-step processes that would otherwise require senior agents, and freed capacity for human agents to focus on high-value interactions. ROI varies by use case, but we track resolution rates, CSAT, response times, and cost per resolution to demonstrate impact. You can access a calculator on RoI here: https://www.cxroi.ai/

What is Team of Agents?

Team of Agents is Lorikeet's multi-agent architecture. Instead of a single AI attempting everything, we deploy specialized agents to handle specific tasks and collaborate on complex inquiries. Agents with specific expertise (e.g., data lookup, policy checking, action execution, response generation) work together under an orchestrator agent that manages coordination and ensures proper sequencing.

This architecture provides better accuracy through task-specialized agents, reduced hallucination by working within narrow domains, improved compliance with domain-specific rule enforcement, easier maintenance by updating individual agents, better scalability by adding new specialists, and clear audit trails tracking which agents handled which parts of inquiries.

Can Lorikeet make phone calls or send texts?

Yes. Lorikeet provides voice support with sub-1 second latency and natural conversation flow. The Team of Agents architecture enables multi-channel orchestration, where specialized AI agents can coordinate across voice, SMS, email, and chat. Voice capabilities include natural interruption handling, emotional intelligence that detects frustration and adjusts tone, warm transfers to human agents with context, and extensive personality customization.

Does Lorikeet work across email, chat, and voice?

Yes. You train workflows once, and they deploy consistently across all channels: web chat widgets, email support, live chat platforms, CRM ticketing systems, voice (phone support), and SMS. Each channel can have channel-specific rules e.g. auto-respond in web chat but create internal notes in email, while maintaining consistent underlying logic.

What types of actions can Lorikeet take?

Lorikeet can execute a wide range of actions through API integrations and MCP servers. For example:

• Process refunds and payment adjustments

• Update customer accounts and subscriptions

• Freeze accounts or cards for security

• Trigger workflows in third-party systems

• Look up customer data across multiple systems

• Send notifications via email or SMS

• Create, update, or close tickets

• Route to specialized teams based on inquiry type

All actions maintain complete audit trails and can require human authorization for high-value scenarios.

How does Lorikeet handle complex workflows?

Lorikeet executes workflows precisely, from simple 3-step forms to complex 40-step processes:

• Follows standard operating procedures (SOPs)

• Accesses customer data and context

• Makes decisions based on business rules

• Takes actions within integrated systems

• Maintains compliance and security throughout

Workflows can include conditional logic, loops, variable passing, API calls, and human approval gates.

How does Lorikeet adapt to different workflow complexities?

Lorikeet offers flexible approaches:

• Agentic RAG search: For searching across documents and unstructured information

• Natural language workflows: For conversational, adaptive processes

• Structured decision trees: For deterministic, compliance-critical processes

This allows appropriate complexity for each use case – from simple FAQs to complex multi-step processes.

What can't Lorikeet do?

We're honest about limitations. Lorikeet doesn't make clinical judgments (healthcare), doesn't replace human judgment for complex edge cases or highly sensitive situations, doesn't handle situations requiring empathy that only humans can provide, and doesn't self-train without human oversight. We're focused on automating processes with clear rules and procedures, not trying to be all things to all people.

Does Lorikeet have a free trial?

Yes. Lorikeet offers a 30-day free trial. You can sign up at app.lorikeetcx.ai/sign-up with your work email and start setting up your first AI concierge.

How do I start a free trial of Lorikeet?

Go to lorikeetcx.ai/free-trial, which takes you to the Lorikeet sign-up page, and create an account. A work email is required to sign up.

Can I try Lorikeet without talking to sales?

Yes. Start a 30-day free trial at app.lorikeetcx.ai/sign-up with your work email. You add a card at checkout to start the trial, and nothing is charged until the trial ends. You can also let your AI assistant, such as ChatGPT or Claude, start the sign-up for you through api.lorikeetcx.ai/v1/agent-signup, and you finish in your own browser.

What happens during the free trial?

After you sign up, Coach, Lorikeet's setup agent, guides you through setting up your first concierge: the AI agent that answers and resolves your customers' questions. You have 30 days to try Lorikeet for free.

Can I book a demo instead of starting a trial?

Yes. If you would rather talk through your use case with the Lorikeet team, book a demo. You can also do both: start a free trial to explore Lorikeet on your own and book a demo to plan your rollout with us.

How do we evaluate if Lorikeet is right for us?

We recommend a pilot approach where you test against historical tickets, run side-by-side comparisons with your current process, and measure quality metrics over a defined period. Our testing framework includes historical ticket replay (see how AI would have performed), synthetic test scenarios for edge cases, and batch testing of hundreds or thousands of inquiries. This reveals projected resolution rates, accuracy, and knowledge gaps before customers experience any issues.

Do you have references from companies like ours?

Yes. Customers with published stories include digital health companies (Eucalyptus, easykind), an insurance company (SageSure), and fintech companies (Taptap Send, Hnry, Carmoola, Flex). We can provide references for specific use cases including HIPAA-compliant patient communication, insurance claims processing (FNOL), complex financial transactions, and regulated industry compliance. Contact us to connect with customers in your industry.

What metrics should we track during a pilot?

Focus on quality over deflection:

• Resolution accuracy: How often does AI provide correct information?

• CSAT: Customer satisfaction with AI interactions

• Resolution time: How quickly are tickets resolved?

• Escalation patterns: When and why does AI escalate to humans?

• Coverage: What percentage of inquiry types can AI handle?

• Cost per resolution: Operational efficiency gains

Don't just measure deflection rate – a high deflection rate with poor quality creates more work downstream.

What QA capabilities does Lorikeet provide before going live?

Lorikeet provides numerous options for pre-implementation QA:

• Historical ticket replay: test against your ticket database to see how AI would have performed

• Synthetic test scenarios: create test cases for edge cases, common inquiries, compliance scenarios

• Batch testing: run hundreds/thousands of test inquiries through workflows

• Accuracy scoring: scores AI responses against expected answers

• Coverage analysis: identifies which types of inquiry AI can handle and where there are gaps

• Compliance validation: verify the AI followed required processes

This testing reveals projected resolution rates, accuracy assessment, escalation predictions, knowledge gaps, workflow issues, and compliance gaps – allowing you to fix issues before customers experience them.

What QA capabilities does Lorikeet provide after implementation?

Lorikeet evaluates every conversation against your quality standards - no random samples or generic criteria. You define your rubric in plain language, and Lorikeet reviews conversations as they happen.

Lorikeet scores each conversation against your standards and surfaces patterns across your team. You get visibility into what's working and what's drifting, with specific feedback on individual conversations rather than abstract scores weeks after the fact.

Lorikeet supports your custom quality rubrics alongside industry-standard metrics like containment and CSAT. You can track accuracy, compliance, tone, process adherence – whatever matters most to how you want to show up for customers.

The same evaluation applies whether conversations are handled by AI or your team. This gives you a consistent view of quality across your entire support operation.

What compliance capabilities does Lorikeet provide for regulated industries?

Lorikeet includes built-in compliance workflows pre-configured for common regulatory requirements like HIPAA and SOC 2. We provide comprehensive audit trails that log all interactions, decisions, and actions, plus consent management that tracks and enforces customer consent preferences. Data retention policies are automatically enforced, and role-based access controls ensure only authorized personnel can access sensitive data. All data is encrypted in transit and at rest.

For regulatory reporting, we support audit trail exports, compliance dashboards, exception reporting, customer consent reports, data access logs, incident reports, and custom reports for your specific regulatory requirements.

What compliance frameworks does Lorikeet support for fintech and healthcare?

Financial Services:

For payment card data, Lorikeet doesn't store or process raw payment card information. We work with payment processors using secure tokens, initiate refunds and payment updates via secure APIs, and maintain complete logging of all payment-related actions. Payment workflows require strict authorization, and all payment data transmission is encrypted.

Healthcare:

We support HIPAA (Health Insurance Portability and Accountability Act) and HITECH Act requirements, sign Business Associate Agreements, and support state-specific health privacy laws. Lorikeet is a customer support platform, not a medical device: it does not diagnose or give clinical advice, and guardrails escalate clinical questions to your team. For healthcare customers, all Protected Health Information (PHI) is encrypted and access-controlled. The AI only accesses PHI required for specific tasks (following the minimum necessary standard), and we maintain complete logging of all PHI access and modifications.

Lorikeet signs Business Associate Agreements (BAAs) and implements administrative, physical, and technical safeguards per HIPAA requirements. We support breach notification processes, patient consent workflows, and offer de-identification options for non-clinical workflows.

General Compliance:

GDPR, CCPA, SOC 2 Type II, ISO 27001, regional data centers for data sovereignty compliance, data transfer controls, and adherence to local compliance requirements. Enterprise customers can integrate with SSO/identity systems like Okta and Azure AD.

How does Lorikeet address AI-specific regulations?

As AI regulations evolve globally, Lorikeet is built to meet emerging requirements. We provide transparency with clear disclosure that customers are interacting with AI, and explainability so you can understand why the AI made specific decisions. Human oversight is built in for high-stakes decisions, and we conduct regular bias monitoring to assess for discriminatory patterns.

Our audit capabilities demonstrate AI behavior and decision-making processes for regulatory review. Customers have the right to request human review of AI decisions, and we maintain comprehensive documentation of AI systems for regulatory compliance.

Ongoing compliance includes regular audits, automatic updates when new regulatory requirements emerge, compliance alerts, training updates when regulations change, documentation maintenance, and access to Lorikeet compliance specialists who stay current on evolving AI regulations. Refer to https://trust.lorikeetcx.ai/ for more information.

How does Lorikeet compare to competitors?

Unlike simple chatbot platforms, Lorikeet handles multi-step workflows and complex processes. Unlike customer service platforms that added AI as an afterthought, we're built AI-first for automation. Our per-resolution pricing aligns incentives differently than per-seat or per-conversation models. We focus on complex support scenarios, not just FAQ deflection—if you only need a chatbot, we're probably overkill.

When are human agents still needed?

Humans excel at complex judgment calls, highly empathetic situations, creative problem-solving for unprecedented issues, and building relationships with high-value customers. Lorikeet handles repeatable processes with clear rules, allowing your human agents to focus on work that truly requires human expertise. The best outcomes come from thoughtful collaboration between AI and humans, not trying to replace humans entirely.

How does Lorikeet improve over time?

Through continuous optimization: we identify knowledge gaps where AI couldn't find content, add new articles or FAQs, refine workflow matching instructions based on performance, monitor resolution rates and satisfaction, review escalation patterns, and learn from human agent overrides. Your support team makes ongoing improvements based on data, not automatic self-training that could introduce unpredictable changes.