AI Qualification

Overview

AI Qualification determines when a lead counts as Qualified for your organization and updates the lead accordingly—typically moving them from New Lead to Qualified, recording qualification timing, and generating an internal summary with inferred details where appropriate.

This feature works across the Kraya WhatsApp Extension, the Dashboard, the Kraya backend, and the AI services that power replies and summaries.

What you can do

Qualify leads automatically based on the live WhatsApp conversation and your stated criteria
Open Conversation Summary on the Dashboard or in the Extension—it updates after the AI sends a reply and gives you a short, easy-to-read overview of the chat
When a lead becomes Qualified, Kraya can create a qualification summary that is written into Notes so you keep a clear written record
See the lead move to Qualified when criteria are met (subject to pipeline and AI settings)
View qualification status, reason, and intent-related fields on the lead where your workspace shows them
Let the AI check your calendar, book calls and set call reminders for the lead mid-conversation (see AI actions below)
Let the AI send your approved WhatsApp templates mid-conversation — brochures, price sheets, carousels — when the moment your instructions describe comes up (WhatsApp API numbers only; see AI-Guided Template Sharing below)

How it works (user-facing)

When a lead sends a message and AI replies are enabled, Kraya generates an AI response using your organization knowledge base, qualification instructions, and conversation context.
Pause after a manual message (WhatsApp API numbers): When you or your team reply to a lead by hand on Chats (or from the WhatsApp Business App on a Coexistence number), AI replies and bump-ups for that lead pause for the time set in Pause AI after a manual message (15 minutes by default). When the pause ends, the AI answers the lead's last message if it is still unanswered. Chats shows AI paused until … with a Resume AI now link. See WhatsApp API Integration. Chrome extension, Hosted Account and Instagram replies are not paused this way.
Interactive options (WhatsApp API numbers): When the AI asks a question with a fixed set of answers (for example "Which course are you interested in?"), Kraya sends those choices as tappable WhatsApp reply buttons (up to 3 short options) or an interactive list (up to 10 options, opened from a button under the message) instead of a numbered text list. The lead taps a choice and the AI continues from that answer. If the options don't fit WhatsApp's limits — more than 10, duplicates, labels that are too long — or if WhatsApp rejects the interactive message, Kraya automatically falls back to the plain numbered text reply, so the lead never misses the question. This applies only to numbers connected via the WhatsApp API; Chrome Extension, hosted WhatsApp session, and Instagram replies stay as text.
The AI decides when the lead should be Qualified and can move the lead from New Lead to Qualified during that reply flow when your instructions are met. Moving to other stages (not only Qualified) depends on AI Stage Shifting; moving to Qualified from New Lead can still occur when AI Stage Shifting is off.
Qualification summary: After a lead is qualified, Kraya can generate a qualification summary (notes update, attributes, intent-related fields) using a separate summary step. Summaries consume credits when they run.

Prerequisites & basics

Credits are available in your organization account
AI replies are enabled globally and for the specific lead (where lead-level control exists)
The lead is in New Lead when you expect first-time qualification from that flow
Your pipeline’s phone number matches the WhatsApp Business phone number connected in Kraya
You are logged into WhatsApp Web with the Kraya Extension active (for Extension-driven conversations)
Your knowledge base and qualification requirements describe what “qualified” means and what to ask

Step-by-step guide

Set qualification expectations (knowledge base)

In the Extension Organization Info or your Dashboard knowledge base (depending on your workflow), complete What questions should the AI ask your leads before qualifying them? and related business context. The knowledge base field has no character limit — write as much context as your business needs (previously capped at 25,000 characters; the cap was removed so long qualification playbooks, FAQs, and product descriptions can live in one place).
Name stages clearly so the AI can align with New Lead and Qualified (and any other stages you use with AI Stage Shifting).

Enable AI replies

Turn on AI Auto-Reply in Global Settings (Extension) or the relevant user automation settings (Dashboard).
Enable AI replies for the lead in the sidebar if your workflow uses per-lead toggles.

Converse with the lead

The AI responds when messages arrive. Keep WhatsApp Web online if you use the Extension.

Qualification outcome

When criteria are met, the lead can move to Qualified and qualification timestamps or data can update as described in your product screens.
If your organization generates qualification summaries, notes may show a block starting with Updated Summary and a timestamp.

Review summary and attributes

Check Notes for appended summaries.
Check lead attributes for inferred Name, Email, or custom fields when the summary step fills them (only when empty and confidence is high enough).

AI actions: checking slots, booking calls and setting reminders

On top of replying, the AI can do things for the lead in the middle of a conversation:

Set a call reminder — when a lead asks to be contacted later ("call me tomorrow evening"), or puts off a decision your instructions say to follow up on, the AI creates a call reminder for your team.
Check availability — when a lead asks whether a time works, the AI checks your Calendar Booking slots and answers with what's free, offering up to five alternatives when the time asked for isn't. Asked about a day without a time, it gives that day's free ranges.
Book a call — when the lead asks to book, the AI books the slot on your calendar. If the lead only asked whether a time is free, it confirms availability and asks them to confirm first; it books only once they ask.
Look things up in your own systems — the AI can call your business's own APIs (an order-status or stock check, for example) and answer from the result. These lookups are set up for you by the Kraya team.

How a booking made by the AI behaves. It's a normal booking. The slot is re-checked against the same availability the public booking page uses, so the AI can't double-book or book outside your hours, and it then sets the lead's booked slot exactly as if the lead had used your booking page. Your reminder sequence, Google Calendar sync and any Call slot is booked Smart Trigger all run as usual. The AI never tells a lead a slot is "booked" by someone else — only that it isn't available.

Times and dates are read in your organization's timezone. The AI resolves phrases such as "tomorrow" or "next Monday" against the actual calendar and won't book on days your calendar is closed.

Reminders it sets carry an Auto chip (tooltip "Set by automation") in the Call Reminders table, Lead Details, the CRM leads list and the chat lead panel. The same chip marks any reminder created by automation — Smart Triggers, sequences or AI calls. The AI never overwrites a reminder a person on your team set.

If an action fails — the slot was taken a moment ago, or the time is in the past — the AI doesn't send a reply that claims it worked. It tries again with a fresh reply a few seconds later; if the booking still can't be made, that reply isn't sent at all. A failed lookup makes the AI say your team will confirm rather than guess.

Testing in the Chat Playground never books or creates reminders. It shows what would have happened instead, for example "AI would book 28 Oct, 11:00 AM".

Availability. These actions are available on paid plans and work on WhatsApp API numbers, hosted WhatsApp sessions, the Chrome Extension and Instagram. Booking needs Calendar Booking to be set up; reminders don't. They don't use extra credits — the reply costs the usual AI reply credit, charged only once the action succeeds. To turn them on for your account, or to connect your own systems for lookups, get in touch with the Kraya team.

Intent score criteria

Every conversation summary gives the lead an intent score from 0 to 10 — the badge you see next to the lead in the CRM, and the number Co-Pilot's Intent Score Threshold compares against. By default Kraya scores intent with a general rubric. You can replace it with your own rules.

Open Knowledge Base and find Intent Score Criteria.
Describe what buying intent looks like for your business, in plain language. For example: "Score 8–10 when the lead shares course, intake and budget and books a counselling call. Score 4–6 when they only name the course."
Save. The criteria apply from the next time a lead's conversation summary updates; existing scores are not recalculated.

How Kraya applies them:

The score is always a whole number from 0 to 10. If your criteria award points, they are added up and capped at 10. If they use another scale — hot / warm / cold, A / B / C, 1 to 5 — it is mapped proportionally onto 0 to 10, coldest to hottest.
The AI scores only on what the conversation actually shows. A lead who hasn't sent any message scores 0.
Your criteria change the score and nothing else — not the summary's wording, language, or the attributes it fills.
Up to 5,000 characters. Leave the field blank (it shows "Blank: Kraya's default intent scoring is used") to go back to the default rubric.

If your organization has a WhatsApp API number connected, admins see an AI-Guided Template Sharing section on the Knowledge Base page, just below AI-Guided File Sharing. Here you pick approved WhatsApp templates and tell the AI when to send each one. During a conversation, when the lead's message matches one of your instructions, the AI sends that template as its reply instead of typing out the same information.

Setting it up (admins)

Open Knowledge Base and scroll to AI-Guided Template Sharing.
Click Add Template and pick a template. Only templates with Approved status are offered.
In Add Template Instructions, write when the AI should send it — for example "Send this when the lead asks about fees or pricing." The instruction is required and can be up to 500 characters.
Click Save Instruction. The template joins the list.

To change the wording later, click Edit instruction on the row. To stop the AI using a template, click Remove template; this only removes it from the AI's list, not from your WhatsApp account.

How the AI uses it

The AI only considers a template when the conversation matches the instruction you wrote, so specific instructions work better than broad ones.
Variables are filled in automatically from the lead, the same way Auto Followup sequences and new-lead broadcasts fill them — the lead's name, for example. The AI never writes variable values itself.
Templates with an image, video or document header and carousel templates are supported and are sent with their media.
A template is sent only from the WhatsApp API number it belongs to. If you have several numbers, a template configured on one is never sent from another.
If a template can't be used at that moment — it is no longer approved, or sending fails — the AI falls back to a normal text reply, so the lead still gets an answer.
Sending a template counts as the AI's reply for that turn and uses the same AI reply credit; if nothing ends up being delivered, the turn isn't charged. WhatsApp's own template charges from Meta apply as usual.

When a template stops being available. If a configured template is later rejected, paused, deleted or resubmitted in WhatsApp, its row shows Unavailable (status) — remove this and the AI stops sending it. Remove it, or re-add it once it is approved again.

Settings and configuration

Qualification requirements (knowledge base): Drives what the AI asks and what “qualified” means.
Interactive options: On by default for organizations created after this launched. For older organizations it is switched on by Kraya support on request — there is no dashboard toggle yet. Turning it off returns every option set to the numbered text format.
AI Auto-Reply and AI Stage Shifting: See AI Stage Shifting. Stage shifting affects moves beyond New Lead → Qualified.
Pause AI after a manual message (WhatsApp Account Settings, per WhatsApp API number): How long AI replies and bump-ups wait after a manual message. 0 to 7 days; 0 turns it off.

Role-based notes

Admins: Keep pipeline phone numbers aligned with the connected WhatsApp Business number; maintain credits; ensure knowledge base content reflects current qualification rules.
Support agents: If qualification seems delayed, verify stage, credits, phone match, AI toggles, and knowledge base wording.
End users: Keep WhatsApp Web open with the Extension active for Extension-based AI and qualification flows.

Limits & behavior

Knowledge base size: The Organization Info / knowledge base field has no character limit. Use this space to describe your business, qualification criteria, stage definitions, and any rules the AI should follow — as much detail as you need.
AI Guide file sharing: You can upload files to share context with the AI. The maximum file upload size is 30 MB per file.
AI-Guided Template Sharing: WhatsApp API numbers only. Up to 2 templates on the Free plan, 50 on Basic and 100 on Pro; at the limit the Add Template button is disabled and the section shows the maximum for your pack. Only Approved templates can be added. Each instruction is up to 500 characters. The section is shown to admins only.
Interactive options: Reply buttons carry at most 3 options; lists carry at most 10. Anything beyond that, or an option set WhatsApp can't render, is sent as numbered text instead. Interactive replies cost the same credits as a normal AI reply.
Triggering replies: AI replies run only when global and lead-level AI settings allow, pipeline phone matches, and credits are available.
Credits (typical):
- Each AI reply consumes credits (for example, one credit per AI reply in standard setups).
- A separate qualification-check credit may apply for some cases.
- Qualification summary generation consumes credits when it runs.
- Exact totals per conversation depend on which steps run for your organization (reply only, reply + summary, or legacy reply + check + summary).
Qualification summary: Generated when your organization’s rules say a summary is needed for a qualified lead (for example, first qualification or when the conversation updates materially), and credits are available.

Common edge cases and how to handle them

Lead not in New Lead
- First-time qualification from the AI flow usually expects New Lead. Move the lead back if you need that path again.
Missing or mismatched phone numbers
- AI replies (and downstream qualification) may not run. Align pipeline and WhatsApp Business numbers.
No credits
- AI replies and summaries do not run without credits.
Attributes not updated
- Attributes fill only when fields are empty and the model infers values confidently; email and name have extra validation rules as before.
Qualification summary missing
- Summaries are not guaranteed on every message; they follow your organization’s summary rules and credits.

Troubleshooting

AI reply did not send

Check global and lead AI toggles, WhatsApp Web connectivity, pipeline phone match, and credits.
On a WhatsApp API number, check whether the chat shows AI paused until …. A team member replied by hand recently; the AI resumes at that time, or straight away if you click Resume AI now.

Lead did not become Qualified

Refine knowledge base qualification instructions; ensure a Qualified stage exists; confirm the lead was in New Lead if you expect first-time qualification via that path.

Qualification summary did not appear

Confirm credits; check whether the lead already had a recent summary; wait for the next eligible summary trigger.

Lead stage did not change

If you expect movement to a stage other than Qualified, enable AI Stage Shifting for the pipeline owner. New Lead → Qualified may still work with AI Stage Shifting off.

Attributes/Notes did not update

Summary appends to Notes with Updated Summary and a timestamp when generated.
Attributes update conservatively from the summary when fields are empty and values validate.

FAQs

How do I turn on AI booking and call reminders? Can the AI book a call slot for a lead during a WhatsApp chat? Yes. To have it switched on for your account, get in touch with the Kraya team — there is no setting you need to change yourself, and you don't need to add any instruction to your AI Playbook for it to work. What your account needs:
- A paid plan.
- AI replies on for the lead (AI Auto-Reply in your settings, and the AI toggle on for the lead's stage) — the actions happen as part of the AI's reply.
- For checking availability and booking: Calendar Booking set up, with Booking Page Status switched on. Without it, the AI can still set call reminders but won't offer or book slots. Once it's on, test it in the Chat Playground: ask something like "Can we talk tomorrow at 3 pm?" and you'll see what the AI would do (for example "AI would book 28 Oct, 11:00 AM") without anything actually being booked. See AI actions above for how bookings and reminders behave.
When does Kraya qualify a lead?
- For most accounts, when the AI determines criteria are met and moves the lead to Qualified (or equivalent behavior your admin confirms). Knowledge base content is central to that decision.
What gets written to Notes?
- When a qualification summary runs, a timestamped block is appended, starting with Updated Summary and the date/time, followed by summary text.
Will AI messages keep sending after qualification?
- Often yes, unless you change settings or disable AI for the lead.
What if I move a qualified lead back to New Lead?
- The lead may become eligible again for qualification-related flows on subsequent AI conversations, subject to your rules and credits.

Tips and best practices

Keep qualification questions and business facts up to date in the knowledge base.
Use clear stage names matching the Dashboard pipeline.
For multi-stage automation beyond Qualified, configure AI Stage Shifting and instructions together.

AI Stage Shifting — automatic moves between stages beyond New Lead → Qualified
Product Catalog — upload a product list so the AI recommends matching products in replies
WhatsApp API Integration — create and get templates approved before adding them to AI-Guided Template Sharing
Chrome Extension — Global Settings, sidebar, AI tools
Auto Followups — sequences by stage
Smart Triggers — automation when stages change
Dashboard — leads and pipelines

Screenshots

Prompt patterns for the Knowledge Base

These are common shapes for the "What questions should the AI ask your leads before qualifying them?" field. Combine them as needed.

Always send a welcome message first

The AI treats your prompt like a character script and follows lines in order. To guarantee a greeting on Q1:

Always start with the welcome message below,
Welcome Message: Hi there! Thanks for reaching out. I can help you quickly. Let's begin.

Q1: What are you looking for today?
- A) Product Demo
- B) Pricing
- C) Support

Q2: What city are you located in?

Q3: What is your preferred contact method?
- A) WhatsApp
- B) Email
- C) Call

Behavior:

Inbound chat — When a lead messages you first on WhatsApp with hi or hello, the AI sends the Welcome Message and Question 1 together.
New Lead stage — When you add or import a lead into New Leads, each new lead automatically receives the Welcome Message and Question 1.

Option-wise validation replies

To make each multiple-choice option trigger a specific reply, add a Validation block under the question:

Q1: What are you looking for today?
- A) Product Demo
- B) Pricing
- C) Support
Validation for Q1:
- If A: Great, I will arrange a short demo. Next, Q2.
- If B: I will share pricing details. Next, Q2.
- If C: I can connect you to support. Next, Q2.

You can personalise replies with placeholders like {{name}} or {{email}}.

Different question sets per audience

If your business serves more than one user group (e.g., teachers and nurses), use a single selector question and branch into separate flows:

Welcome:
Hi! I'll ask a few quick questions to guide you better.

Q1: What are you looking for?
Options: Teacher Abroad | Nurse Abroad
Validation: If answer includes "Teacher" go to TEACHER_FLOW
           If answer includes "Nurse" go to NURSE_FLOW

--- TEACHER_FLOW ---
Q2: Target country?
Q3: Highest qualification? (B.Ed | M.Ed | PhD | Other)
Q4: Teaching experience? (0 | 1-2 | 3-5 | 5+)
Q5: Do you have IELTS/TOEFL? (IELTS | TOEFL | None)

--- NURSE_FLOW ---
Q2: Registered Nurse status? (Yes | No | In progress)
Q3: Total experience? (Intern | 1-2 | 3-5 | 5+)
Q4: Preferred destination? (UK | Germany | Australia | Other)
Q5: IELTS/OET? (IELTS | OET | None)

Tips:

Keep Q1 as the first (selector) question.
Limit each flow to 3–5 questions.
Use clear 1–3 word option labels.
Test with a demo lead before activating.

On this page