If you feel any tags have been applied incorrectly, review the description for each tag to ensure it is clear. ### **Ho**w much does it cost to automatically tag and get a sentiment score for each ticket or conversation? It costs \$0.05 for each conversation property per ticket, this will be an add-on to your existing subscription and charged at the end of each billing period. For example, if you were to have 3 conversation properties applied to a ticket, this would cost $0.15 for that ticket, if just one conversation property needs populating it will be just $0.05 for that ticket. ### What tagging or categories can I use for my Freshchat conversations and tickets? You can use your own custom tagging, conversation properties, reason codes or groupings when tagging your Freshchat conversations or tickets. For example, if you were to have 3 conversation properties applied to a ticket, this would cost \$0.15 for that ticket, if just one conversation property needs populating it will be just \$0.05 for that ticket. ### What do I write for my AI tagging description? Your AI tagging description is used to classify your ticket (in conjunction with the tag name), so you should describe when you want the tag to be applied. If you are unsure, or to begin with, click the ✨ AI generation button and it will generate a description for you automatically, you can then tweak this if you want to make it more accurate. ### How can I ensure the AI agent only responds to certain types of Freshchat conversations? If you are using Freshchat you can use our tagging product to tell the AI agent which tickets it should respond to. For example, you might have "reason for contact" conversation property in Freshchat that your agents update today, and there might be 10 category options within that. If "unsubscribe" was the tag added by our AI agent, and that tag was on your block list, then the AI wouldn't reply to it and it will force a human escalation. To add a tag to your block list, go to **AI Agent Setup > AI Tagging**, select the conversation property that has the tag you want to block, then find the tag and click **Block.** If a conversation isn't replied to because it was blocked (and you have [created an AI Chat Status property, following the instructions here](/features/channels/freshchat#how-do-i-add-conversation-properties-so-i-know-what-the-ai-status-of-a-conversation-is)) the AI Chat Status will be updated to `ai-agent-tag-blocked`. Alternatively, you could set up a triage flow for your AI agent, using rules or automations within Freshchat for where certain tickets should be routed depending on the "reason for contact".
If you feel any tags have been applied incorrectly, review the description for each tag to ensure it is clear. ### **Ho**w much does it cost to automatically tag and get a sentiment score for each ticket or email? It costs \$0.05 for each tag or custom field per ticket, this will be an add-on to your existing subscription and charged at the end of each billing period. For example, if you were to have 3 tags/custom fields applied to a ticket, this would cost \$0.15 for that ticket, if just one tag/custom field needs populating it will be just \$0.05 for that ticket. ### What tagging or categories can I use for my Freshdesk emails and tickets? You can use your own custom tagging, categories, reason codes or groupings when tagging your Freshdesk emails or tickets. The tagging will be captured in up to 3 Freshdesk custom fields or tags you define, for each email or ticket, or within the ‘Tags’ field if you don’t use custom fields. ### What do I write for my AI tagging description? Your AI tagging description is used to classify your ticket (in conjunction with the tag name), so you should describe when you want the tag to be applied. If you are unsure, or to begin with, click the ✨ AI generation button and it will generate a description for you automatically, you can then tweak this if you want to make it more accurate. ### How can I ensure the AI agent only responds to certain types of Freshdesk tickets? If you are using Freshdesk Tickets you can use our Ticket tagging product to tell the AI agent which tickets it should respond to. For example, you might have "reason for contact" custom field in Freshdesk that your agents update today, and there might be 10 category options within that. If "unsubscribe" was the tag added by our AI agent, and that tag was on your block list, then the AI wouldn't reply to it and it will force a human handover. To add a tag to your block list, go to **AI Agent Setup > AI Tagging**, select the Custom Field or Tag that has the tag you want to block, then find the tag and click **Block.** If a ticket isn't replied to because it was blocked the tag `ai-agent-tag-blocked` will also be added to the ticket. Alternatively, you could set up a triage flow for your AI agent, using rules or automations within Freshdesk for where certain tickets should be routed depending on the "reason for contact".
You can create a separate account for this if you’d like to customise the name or image of the AI agent.
In general, set the sender as a team account or designated AI agent seat within Intercom.
`"Hi there, I'll pass you back to our AI assistant for any further questions."`
This approach is the best long term one as you can keep feeding in more information about that user's account. 2. [You can add a note](/features/channels/intercom/intercom-playbooks#provide-your-intercom-ai-agent-with-conversation-attributes-from-intercom-to-answer-user-data-specif), as part of the current workflow, just before the user is asked to type their question.
You can add in static and dynamic content here (from Intercom), so if you're able to add the current user's page as a user attribute, you can feed it in here.
The AI will look at this note as well as the user's message for context. You can even do both. Option #1 will help give the AI additional knowledge about the user, and Option #2 will enrich that user's question to be more contextual to where they are. ### How to collect email on handover/escalation in Intercom When a Intercom conversation has been handed over or escalated to an agent, you will likely want to collect the email address of the user or customer if they were a visitor. To do this, [follow our guide on escalation data collection here](/features/escalation-data-collection). # Intercom (Tagging) Source: https://support.myaskai.com/features/channels/intercom/intercom-tagging Automatically tag your tickets and conversations in Intercom with custom tags and sentiment scores.
If you feel any tags have been applied incorrectly, review the description for each tag to ensure it is clear. ### **Ho**w much does it cost to automatically tag and get a sentiment score for each ticket or conversation? It costs \$0.05 for each attribute per ticket, this will be an add-on to your existing subscription and charged at the end of each billing period. For example, if you were to have 3 attributes (tags) applied to a ticket, this would cost \$0.15 for that ticket, if just one attribute needs populating it will be just \$0.05 for that ticket. ### What tagging or categories can I use for my Intercom emails and conversations? You can use your own custom tagging, categories, reason codes or groupings when tagging your Intercom emails or conversations. The tagging will be captured in up to 3 Intercom attributes you define, for each conversation or ticket, or within the ‘Tags’ field if you don’t use custom fields. ### What do I write for my AI tagging description? Your AI tagging description is used to classify your ticket (in conjunction with the tag name), so you should describe when you want the tag to be applied. If you are unsure, or to begin with, click the ✨ AI generation button and it will generate a description for you automatically, you can then tweak this if you want to make it more accurate. ### What reporting do I have for my tags and sentiment? You can use Intercom's reporting to interrogate and report on your tags and sentiment scores to see which and how they are changing over time. ### How many options can I apply for each attribute to my Intercom emails or conversations? We would recommend using fewer than 200 options for your attributes. ### What if I update my attributes in Intercom? Your Auto-Tagging isn't automatically kept in sync in Intercom, so if you add or remove a field or tag in Intercom, you will have to do the same in My AskAI and vice versa (make sure the value name matches exactly). Alternatively, you can just remove and re-add the attribute in My AskAI for it to re-sync, although be aware that descriptions will be re-written (in case you have changed many of them). ### How can I ensure the AI agent only responds to certain types of Intercom conversations or tickets? If you are using Intercom you can use our Ticket tagging product to tell the AI agent which tickets it should respond to. For example, you might have "reason for contact" attribute in Intercom that your agents update today, and there might be 10 category options within that. If "unsubscribe" was the tag added by our AI agent, and that tag was on your block list, then the AI wouldn't reply to it and it will force a human handover. To add a tag to your block list, go to **AI Agent Setup > AI Tagging**, select the attribute that has the tag you want to block, then find the tag and click **Block.** If a conversation isn't replied to because it was blocked the tag `ai-agent-tag-blocked` will also be added to the conversation. Alternatively, you could set up a triage workflow for your AI agent, using rules or triggers within Intercom for where certain conversations should be routed depending on the "reason for contact".
Add My AskAI to 'a team' and then add to any 'channel'.
2. **Add My AskAI** to your Slack to your Workspace\ \ Click below :point\_down: [](https://slack.com/oauth/v2/authorize?client_id=3639755760257.5036828617799\&scope=channels:history,channels:join,chat:write,commands,groups:history,im:history\&user_scope=)
3. **Get your My AskAI's ID and API Key**\ \ Go to ***Channels > Slack & Teams*** and click the Slack icon then copy your My AskAI's **ID** and **API Key**.
4. **Add your My AskAI's ID and API Key to Slack** 1. Select the My AskAI app within Slack 2. Select 'Connect My AskAI' 3. Input the ID and API Key, from step 3, into the popup within Slack 4. Select 'Submit' to save these settings
All Zendesk Messaging bot integrations require you to have Sunshine Conversations access (on the Zendesk Suite Professional plan or above) and usage purchased or included in your Zendesk plan. If you would like to learn more about purchasing Sunshine Conversations, please speak directly with Zendesk who will be happy to help.
*https\://**teammyaskai**.zendesk.com*
5. Enter your subdomain and click "Sign in"
Create the following trigger:
**Trigger name**: Auto-Close Solved Ticket
**Description**: Close a ticket that has been marked as solved
**Category**: Ticket Status (you may have to + Add Category if this is not already available)
**Conditions**: Meet ALL of the following conditions: *Ticket > Status category* | *IS* | *Solved*
**Actions:** *Ticket Status > category* | *Closed*
You can, however, pause replies at any time once set up by going to **Channels > Zendesk > Zendesk Messaging**, scrolling to the bottom of the page and toggling "Pause AI agent" on.
If you chose a Custom Field and don't have one setup in Zendesk already or you chose Tag then you will need to add a list of values/tags and descriptions for those values/tags so the AI knows under which circumstances each value/tag should be applied.
4. Once you have reviewed the descriptions for each tag and checked they are correct then you can turn on the auto-tagging. 5. You will then see your tags auto-applied in Zendesk, additionally the tag `ai-agent-tags-applied` will be applied so you know when it has been used.
If you feel any tags have been applied incorrectly, review the description for each tag to ensure it is clear. ### **Ho**w much does it cost to automatically tag and get a sentiment score for each ticket or conversation? It costs \$0.05 for each tag or custom field per ticket, this will be an add-on to your existing subscription and charged at the end of each billing period. For example, if you were to have 3 tags applied to a ticket, this would cost \$0.15 for tags on that ticket, if just one custom fields needs populating it will be just \$0.05 for that ticket. ### What tagging or categories can I use for my Zendesk tickets and conversations? You can use your own custom tagging, categories, reason codes or groupings when tagging your Zendesk tickets or conversations. The tagging will be captured in up to 3 Zendesk custom fields you define, for each conversation or ticket, or within the ‘Tags’ field if you don’t use custom fields. ### What do I write for my AI tagging description? Your AI tagging description is used to classify your ticket (in conjunction with the tag name), so you should describe when you want the tag to be applied. If you are unsure, or to begin with, click the ✨ AI generation button and it will generate a description for you automatically, you can then tweak this if you want to make it more accurate. ### What reporting do I have for my tags and sentiment? You can use Zendesk Explore to interrogate and report on your tags and sentiment scores to see which and how they are changing over time. ### How many tags can I apply to my Zendesk tickets or conversations? We would recommend using fewer than 200 tags. ### What if I update my tags in Zendesk? Your Auto-Tagging isn't automatically kept in sync in Zendesk, so if you add or remove a field or tag in Zendesk, you will have to do the same in My AskAI and vice versa. ### How can I ensure the AI agent only responds to certain types of Zendesk tickets? If you are using Zendesk Tickets you can use our Ticket tagging product to tell the AI agent which tickets it should respond to. For example, you might have "reason for contact" field in Zendesk that your agents update today, and there might be 10 category options within that. If "unsubscribe" was the tag added by our AI agent, and that tag was on your block list, then the AI wouldn't reply to it and it will force a human handover. To add a tag to your block list, go to **AI Agent Setup > Auto-Tagging**, select the Custom Field or Tag that has the tag you want to block, then find the tag and click **Block AI Replies.** If a ticket isn't replied to because it was blocked the tag `ai-agent-tag-blocked` will also be added to the ticket. Alternatively, you could set up a triage flow for your AI agent, using rules or triggers within Zendesk for where certain tickets should be routed depending on the "reason for contact".
Trigger name: Assign AI Agent Ticket
Tag values: `ai-agent-replied` & `human-handover-requested`
Trigger name: Assign AI Handover Ticket
Tag value: `human-handover-requested`
These triggers (or order of triggers) may need to be adjusted to match your organisation's workflows and avoid any conflicts.
**Name**: Pre-Escalation User Info Collection
**Description:** Collect the customer's email and phone number before passing them on or ending the interaction. 2. Turn the toggle on for "Run task on every escalation / handover (live chat only)
3. Describe the steps you want the AI agent to go through and the data you want it to collect and what format it will be in. ### Does it happen on every escalation/handover? If you have created an escalation data collection task and enabled it, then it will run on every handover/escalation. ### Does it work for any helpdesk? The escalation/handover data collection task works on all helpdesk integrations. ### What happens with the data once it has been collected? Once the data is collected by the AI agent, it will reside in the conversation or ticket history but will only be added to any meta data or other fields if you create an integration to allow this, e.g. an [API tool](/features/ai-actions-tools) that posts the collected data somewhere. ### How does escalation data collection work in Zendesk Messaging? [Zendesk Messaging](/features/channels/zendesk/zendesk-messaging) has a native integration for passing collected data into the escalated ticket. When a conversation is escalated, any email address and name collected by your AI agent will automatically be attached to the Zendesk ticket created for the handover. To enable this, go to your **Zendesk Admin panel > Messaging > Messaging Settings** and enable 'Use both verified and unverified emails'. Once enabled, the visitor's email and name collected during the conversation will appear on the ticket, so your agents have the customer's details ready when they pick it up. This means you can require customers to provide their email on escalation in Zendesk Messaging so you can follow up.
| Live chat provider | Human handover? | Email handover? | Conversation summary |
|---|---|---|---|
| Intercom | true | true | Automated |
| Zendesk (Messaging + Tickets) | true | true | Automated |
| HubSpot | true | true | Automated |
| Freshchat | true | false | Automated |
| Freshdesk | true | true | Automated |
| Gorgias | true | true | Automated |
You can then upload a logo, like our [My AskAI logo](https://7ac07acbae116a5155080d295197d8d6.cdn.bubble.io/f1709308979339x243661440310076960/Logo.png). You can always find your Slack app if you need to make changes within '[Your Apps](https://api.slack.com/apps/)' in Slack. If you get stuck, there is an [additional guide from Slack](https://api.slack.com/messaging/webhooks) on how to create an app with webhooks. # Languages and Localization Source: https://support.myaskai.com/features/languages-and-localization You can change the language or localize your AI support agent to the languages your customers speak. ### How do I change the language and localize my AI agent? To change the language and localize your AI agent: 1. Log into your Dashboard. 2. Go to **AI Agent setup > Customize.** 3. Scroll to the bottom of the page, to where it says "Language". 4. Toggle the "Enable non-English support" switch. 5. Select your AI assistant's language from the drop-down. 6. Refresh your widget conversation and refresh the page. 7. Your AI support agent will now respond in, and all elements of your widget will be localized to your chosen language. ### Can I upload non-English content and websites? Websites and content can be uploaded in any language, regardless of which language you want your AI agent to respond in. ### Can I change the language and localize Zendesk, Intercom or HubSpot? You can change the language and localize both the Intercom and Zendesk widgets, but not currently HubSpot (get in touch via chat if you need HubSpot in other languages). ### Which languages do you support? We currently support 95 languages, including localization for: * Arabic * Bengali * Bulgarian * Chinese * Czech * Danish * Dutch * English (British) * English (US) (*default*) * Finnish * French * German * Greek * Hebrew * Hindi * Hungarian * Indonesian * Italian * Japanese * Korean * Malay * Norwegian * Polish * Portuguese * Romanian * Russian * Spanish * Swahili * Swedish * Turkish * Ukrainian * Vietnamese We also support other languages by request, if you need an additional language, sign up for a Pro plan and get in touch with us via chat and we will add your language right away. ### Do you support multilingual responses? We do support multilingual AI responses and can also auto-detect languages for localization. ### Can I localize my AI support agent? Yes, you can localize your agent, [follow these instructions that will show you how](/features/languages-and-localization#how-do-i-change-the-language-and-localize-my-ai-agent-askai). # Multibrand Source: https://support.myaskai.com/features/multibrand Many businesses may have different products, regions or brands that they want to have separate AI agents for so they can control the knowledge used and the way the agent will respond, here's how. ### Can I have multiple AI agents? Yes, you can have multiple agents. You should only have one 1 AI agent per brand, knowledge base or account, agents can also be separated by user group or product where necessary. ### Can you support multiple knowledgebases for separate products in a single AI agent? We are able to support multiple knowledgebases for separate products. ### How do I set up my AI agent(s) for multiple brands? How you set up your AI agent for multiple brands will vary depending on your helpdesk: * [How to set up multiple brands or agents in Intercom](/features/channels/intercom#i-have-multiple-brands-users-or-email-accounts-in-intercom-can-i-have-the-ai-agent-only-respond-to-o) * [How to set up multiple brands or agents in Zendesk Tickets](/features/channels/zendesk/zendesk-tickets#i-have-multiple-brands-users-or-email-accounts-in-zendesk-can-i-have-the-ai-agent-only-respond-to-on) * [How to set up multiple brands or agents in Zendesk Messaging](/features/channels/zendesk/zendesk-messaging#can-i-have-multiple-ai-agents-responding-to-my-zendesk-messaging) * [How to set up multiple brands or agents in Gorgias](/features/channels/gorgias#i-have-multiple-brands-users-or-email-accounts-in-gorgias-or-shopify-can-i-have-the-ai-agent-only-re) * [How to set up multiple brands or agents in HubSpot](/features/channels/hubspot#i-have-multiple-brands-users-or-email-accounts-in-hubspot-can-i-have-the-ai-agent-only-respond-to-on) * [How to set up multiple brands or agents in Freshdesk](/features/channels/freshdesk#i-have-multiple-brands-users-or-email-accounts-in-freshdesk-can-i-have-the-ai-agent-only-respond-to) * [How to set up multiple brands or agents in Freshchat](/features/channels/freshchat#i-have-multiple-brands-users-or-email-accounts-in-freshchat-can-i-have-the-ai-agent-only-respond-to) # Private (Internal) mode Source: https://support.myaskai.com/features/private-internal-mode In addition to the visitor or customer facing mode of your AI agent, you can also have "Private" mode, that you can use for your team, trained on material that isn't public.
| Industry | Use Case | Task |
|---|---|---|
| SaaS | Plan change & proration | Guides users to the right plan based on usage and entitlements, checks current contract term, calculates prorated charges/credits, confirms billing contact, executes the plan switch via billing API, updates seats and feature flags, and emails a new order form or confirmation. |
| SaaS | Seat management & access requests | Identifies the workspace and admin, verifies requester identity, checks seat availability vs. limits, creates/assigns new seats, provisions roles/SSO groups, and logs the audit trail. If policy requires manager approval, pings approver and completes automatically when approved. |
| SaaS | Data export & account deletion | Authenticates the requester, lists available data scopes, schedules an export to the secure bucket, notifies when ready, confirms deletion cooldown, queues deletion across services, revokes tokens/SSO, and issues compliance confirmation once data is purged per retention policy. |
| eCommerce | Refunds & returns orchestration | Looks up the order, confirms item condition and policy (window, category exceptions), offers refund/exchange/store credit, generates label, books carrier pickup if eligible, triggers warehouse inspection workflow, and issues refund/exchange when scan or inspection passes. |
| eCommerce | Missing parcel / delivery issues | Checks courier tracking and GPS delivery proof, verifies address and safe-place preferences, applies loss/theft policy, issues replacement or refund, files carrier claim with evidence, and updates inventory. If fraud signals are high, routes to review with a prepared evidence pack. |
| eCommerce | Size/fit exchange & recommendations | Pulls prior purchases and fit feedback, recommends sizes based on brand-specific charts and returns history, creates exchange order with variant availability, reserves stock, issues return label for original item, and updates loyalty points. |
| eCommerce | Warranty & damage claims | Collects photos/videos, validates serial or order, checks warranty coverage, decides repair vs. replace, schedules service or dispatches replacement, produces return materials for defective unit, and closes with confirmation and tracking. |
| Marketplaces | Order mediation & partial resolutions | Retrieves buyer and seller messages, validates evidence (photos, timestamps), applies marketplace policies for SNAD/INR, proposes partial refund/return labels, holds funds as needed, automates final settlement and feedback reversal rules, and updates metrics for both parties. |
| Marketplaces | KYC re-verification & payout holds | Detects payout block reasons, guides seller through KYC doc capture, validates against provider response, refreshes tax forms, recalculates withheld payouts, and releases funds when all checks pass. Escalates only on watchlist or mismatch flags. |
| Marketplaces | Delivery SLA breach remediation | Checks promised vs. actual ship/arrival times, identifies responsible party, applies fee rebates or shipping label upgrades, communicates the remedy to both parties, and updates performance scores and dispute counters. |
| Fintech | Transaction disputes | Collects disputed charge details, verifies device/location, checks eligibility windows and merchant category rules, files chargeback or inquiry with correct reason code, updates provisional credit status, and keeps the customer informed through each stage until resolution. |
| Fintech | Changes to personal details | Verifies identity (2FA/KBA/doc scan), captures new legal name/address, validates against sanctions/PEP rules, updates KYC profile across systems, reissues cards or statements, and confirms changes with an audit log. |
| Fintech | Limit increases & risk checks | Assesses income and payment history, checks outstanding obligations, runs risk engine, proposes a responsible limit, obtains consent to pull additional data if needed, updates credit line, and provides new repayment schedule and disclosures. |
| Healthcare | Appointment triage & scheduling | Collects symptoms and urgency markers, checks provider availability and insurance eligibility, recommends the right visit type (in-person/telehealth), books the slot, collects co-pay, shares prep instructions, and sets reminders. Flags red-flags for immediate escalation. |
| Healthcare | Prior authorization automation | Pulls diagnosis/treatment codes, validates payer rules, compiles clinical notes and attachments, submits prior auth, tracks status, requests additional info when required, and notifies patient and clinic when approved/denied with alternatives. |
| Travel & Hospitality | Flight change & disruption handling | Reads fare class rules, and waiver policies, proposes rebooking options, handles ancillaries (seats/bags), issues residual credit or collects fare difference, and pushes updated itinerary and boarding passes. |
| Travel & Hospitality | Hotel overbooking recovery | Detects oversell risk, locates comparable properties, secures walk arrangement with rate protection, handles transport vouchers, moves the reservation, and updates loyalty entitlements and compensation per policy. |
| Travel & Hospitality | Refunds for add-ons & services | Validates unused services (tours, transfers), checks supplier cancellation windows, initiates refunds/credits, updates vouchers, and notifies all parties with new confirmations. |
| EdTech | Course access & enrollment fixes | Confirms learner identity, checks LMS enrollments/licensing, resolves access errors, switches cohorts or schedules, updates billing if price differs, and sends orientation materials. |
| EdTech | Academic integrity review intake | Collects incident details and evidence (plagiarism report, proctoring flags), maps to policy, proposes outcomes (warning, retake), triggers instructor approval, communicates decision to student, and logs the case. |
| EdTech | Certificate issuance & name corrections | Verifies completion status, confirms certificate name, processes name change if needed (with doc proof), issues digital certificate, and updates public credential registry. |
| Insurance | FNOL (first notice of loss) | Captures incident details with guided questions, checks policy coverage and deductibles, books inspection or directs to preferred repairer, issues temporary accommodations if covered, and provides claim number and timeline. |
| Insurance | Policy changes & endorsements | Verifies identity, simulates premium change for endorsements (drivers, address, valuables), applies underwriting rules, collects additional premium or issues refund, and emails updated policy docs. |
| Insurance | Proof of insurance / ID cards | Confirms policy status, generates digital ID cards or COIs for landlords/DMV, delivers via email/SMS/wallet pass, and logs distribution for compliance. |
| Real Estate | Maintenance requests & vendor dispatch | Confirms lease and access permissions, triages issue severity, checks warranty/service contracts, dispatches approved vendor with SLAs, tracks completion, and collects tenant sign-off. |
| Real Estate | Lease renewals & rent adjustments | Surfaces renewal window and guidelines, proposes terms based on market data and history, collects tenant intent, generates renewal docs for e-signature, and updates property management and payment systems. |
| Real Estate | Application screening & deposits | Guides applicants through documentation, runs screening (credit, income, background) per jurisdictional rules, communicates decision, collects holding deposit, and sets up the move-in checklist. |
| Processing Activity | Purpose | Legal Basis |
|---|---|---|
| Creating, accessing, managing and using your account | To grant you access to My AskAI, administer and manage your account, and allow you to use our service | LI, CD, C |
| Payment & billing management | To process payments and subscriptions transactions | CD, C |
| Adding content to a My AskAI (AI training) | To allow you to add content to your My AskAI (bot) to enable it to answer questions | LI, CD, C |
| Conversational interaction with a My AskAI 'bot' | To allow the user to interact and 'chat' with their My AskAI and get responses to their questions | LI, CD, C |
| Customer support | To provide customer support | LI |
| Bug and security monitoring | To prevent and investigate system abuse | LI |
| Website audience measurement | To gather analytics on our website traffic | LI, C |
| Service improvement | To maintain and optimize the performance of My AskAI and understand how individuals use it | LI |
| Newsletter subscription management | To send a regular newsletter and gather statistics | LI, C |
| Marketing communication | To send marketing communications about updates and promotions about My AskAI | LI, C |
| Marketing communication (Customers' end users) | For customers to collect email addresses of end users via chat forms. | LI, C |
| B2B Lead management \* | To interact with prospective individuals about My AskAI by email and follow lead management | LI |
| Testimonial collection \* | To gather and display testimonials on our website | C |
| Virtual demo session \* | To organize a demo session that you sign up for | C |
| Feedback collection \* | To gather and display feedback on our Public Roadmap | C |
| Affiliate and referral programs management \* | To offer a reward program to people promoting My AskAI | LI, CD |
| Sub Processor | Location | DPA |
|---|---|---|
| Twillio (SendGrid) | US | Data Processing Agreement\* |
| Bubble | US | Data Processing Agreement\* |
| Stripe | US | Data Processing Agreement\* |
| Paypal | US | Data Processing Agreement\* |
| Hotjar | US | Data Processing Agreement\* |
| Google Analytics | US | Data Processing Agreement\* |
| Amplitude | US | Data Processing Agreement\* |
| CloudFlare | US | Data Processing Agreement\* |
| Amazon AWS | US | Data Processing Agreement\* |
| Render | US | Data Processing Agreement\*\* |
| OpenAI | US | Data Processing Agreement\* |
| CarbonAI | US | Data Processing Agreement\*\* |
| Papertrail (Solarwinds) | US | Data Processing Agreement\* |
| Qdrant | US | Data Processing Agreement\* |
| US | Data Processing Agreement\* Part 1 Part 2 Part 3 | |
| Slack | US | Data Processing Agreement\*\* |
| Microsoft Teams | US | Data Processing Agreement\* |
| Plan | Support Options |
|---|---|
| Pro |
|
| Scale |
|
| Enterprise |
|
5. If, after inspecting the answer you still don't understand the response or think it may be incorrect, you cam get in touch with us via our chat widget and ask to "Talk to a person". # I can't see any billing information for my account? Source: https://support.myaskai.com/troubleshooting/i-cant-see-any-billing-information-for-my-account If you are unable to see billing information or details for your account, it is likely because you aren't the owner of the account. You will need to either logout and then login as the owner of the account, or contact the account owner to provide you with the billing information for the account. You can find the account owner by going to Account Settings and looking under "Team access". # Login Source: https://support.myaskai.com/troubleshooting/login Having login issues for My AskAI? Here's what you can do. ### Can't Login When logging in from some work email accounts occasionally magic links don’t work. This is because they are often automatically “clicked” by email providers to check for spam or viruses. If you are experiencing issues: 1. Try again then; 2. Try [setting up a password](/account-management/user-access/setting-a-password-for-your-account) for your account. # Oops there has been an error Source: https://support.myaskai.com/troubleshooting/oops-there-has-been-an-error If you see the response "oops there has been an error" then first check that you are using the correct "Mode" for your AI support agent, if you have uploaded a file or used a connector, by default that information is private, and so you will need to toggle your AI support agent to ["Public + Private" mode](/features/private-internal-mode) to get responses from it. Also check that you have content added to your AI support agent, if it doesn't have anything to answer with, it won't be able to answer! If this doesn't resolve your issue please get in touch with us via chat. # Slow responses Source: https://support.myaskai.com/troubleshooting/slow-responses You can generally expect response times from My AskAI to be just a few seconds however there are rare occasions when this is not the case. Response times can vary based on API response times and also by the size of the answer being returned. The longer and more detailed the answer being given the longer it will take to produce. Unlike some other (faster) AI agents, our AI agents have a number of guardrail and quality assurance steps in place before responding. This can mean that a single question may go through 10+ processes before a response is returned. This ensures the highest quality answers for your users and customers. Also, bear in mind that answer speeds are always improving with the pace of AI development, so if they aren't quick enough now, give it a month or 2! ### What do you/can I do to try and make responses "feel" faster? Depending on your helpdesk integration your AI agent can display something different while the user is awaiting a reply, for example on: * Zendesk Messaging: an animated "..." icon will be displayed * Intercom chat: a "just thinking..." message will be displayed ([this can be turned off if required](/features/channels/intercom#can-i-turn-off-the-just-thinking...-or-just-having-a-think...-messages-in-intercom)) * HubSpot chat: a "just thinking..." message will be displayed ([this can be turned off if required](/features/channels/hubspot#can-i-turn-off-the-just-thinking...-or-just-having-a-think...-messages-in-hubspot)) * Freshchat: a "just thinking..." message will be displayed ([this can be turned off if required](/features/channels/freshchat#can-i-turn-off-the-just-thinking...-or-just-having-a-think...-messages-in-freshchat)) * Gorgias chat: a "just thinking..." message will be displayed ([this can be turned off if required](/features/channels/gorgias#can-i-turn-off-the-just-thinking...-or-just-having-a-think...-messages-in-gorgias)) # Website not added Source: https://support.myaskai.com/troubleshooting/website-not-added Very rarely we might miss individual pages or be unable to add a website, here's what you can do. ### Why can’t my AI agent access or add my website? The main reasons we see for bots not being able to access or add websites is if they are either behind password protection or if they have some sort of technology to prevent them from being scraped. If you are sure it isn’t one of these things then get in touch via chat and we’ll help you get your content uploaded. ### It's stuck training on my website If you're stuck adding knowledge for more than 24 hours, get in touch with us via chat and we will help. # Why can't it answer? Source: https://support.myaskai.com/troubleshooting/why-cant-it-answer What to do if your answers aren't correct or if your AI support agent isn't answering how you would expect. ### My AI agent answers are mostly correct, there are just a few answers I'd like to update In this situation, it is best to first try using the [Custom answers](/features/improve) feature to correct the issue. ### My AI support agent isn't answering a question that I have added content for #### Check you have uploaded the content required to answer the question The first thing you should check is that the content piece (website, file, or page) has indeed been uploaded to your AI support agent by locating the specific piece in the Knowledge section, you would be surprised how often we investigate these issues and find the AI agent didn't have the knowledge to begin with! #### Check the content itself will enable the AI to answer Secondly, when you ask the question, check any of the links given in the response to the answer and see whether any of them have the information that would be required to answer the question: * If the relevant information isn't there this suggests that the AI was unable to find a relevant answer in your Knowledge, in this case, you could either: add a [custom answer](/features/improve) or try and improve the language of your documentation to make it more specific and detailed. * If the relevant information is there, first ask whether you would expect a person, given the question and the information would be able to answer, if not then it may be that your documentation can be improved, if yes then please share details of the question and answer with us via chat and we will investigate. #### Ask yourself whether the question is something a user would likely need to ask AI agents are optimized to answer questions based on knowledge uploaded, they are not "character avatars" for people to have long, personal conversations with, therefore they may not give friendly answers to questions about where they are from or what their name is. Doing this would not only eat very quickly into your conversations, but would also create far more noise in your insights data, so our AI agents are designed to be "to the point" with questions and conversations. Our experience also shows us that when most users interact on chat, their main focus is on a speedy answer to a question and so that is what they give. #### AI agents can't answer *every* question Our AI agents are designed to help you instantly answer customer support and internal knowledge-based questions from your customers and team. They are not however designed for: * *Document comparison* - your AI support agent will not be able to compare 2 or more documents * *Data analysis* - Your AI support agent is unable to perform data analysis. This is due to the way the technology is set up and designed, it is made to answer questions using multiple (but a limited number of) pieces of context from a document instead of being able to aggregate, summarise or analyze and entire data set at once. * *Setting up or scheduling appointments* - your AI agent is unable to set up or schedule appointments for you in your calendar. * *Maths* - remember it is a *language* model first and foremost so we would always check its arithmetic if we were you and wouldn't recommend using it for maths. * Summarization - they will not be able to summarize documentation. ### My answers were working fine before but now are not We are always working on improving your AI agents, that may be through improving prompts, moving to the latest models, or making some infrastructural changes to improve speeds or answer quality. This can mean that occasionally things that were answered previously are now not, this is incredibly rare and often a temporary issue. We always strive to use the latest OpenAI models, as they are the most powerful and most actively developed, the sooner we can use them the longer we have to improve outcomes with them. In addition, OpenAI deprecates (removes) models over time - often within a few months of a newer model being released, so "sticking with the old" is rarely an option. While some questions right now may not be being answered, there may also be far more that are now being answered that weren't before. Answer quality is our utmost concern and priority and we monitor a range of statistics closely whenever we make changes (and before making these changes), and if it makes sense, we are not afraid to roll back changes where it is beneficial to the overall user base. In short, if you feel answers are not performing as well as they were previously, it is likely something that will resolve itself within a few weeks and the overall path of answer quality and ability is forever improving to the extent that we don't believe this will even be a consideration in the very near future. If you would like to track the changes we make, then you can see them here, in our algorithm change log. ### I've tried everything and I still can't get the answers I want If none of the above resolves your issue then please get in touch with us via chat.