Callmenta keeps evolving

The client card now displays the entire communication history New feature

• When you open the client card, you will see the communication history alongside files, hearings, and collections. Phone calls with their duration and outcome, WhatsApp and live chat conversations by channel, notes, and appointments are all listed in a single stream, sorted from newest to oldest. If a call recording is available, you can jump directly to its analysis; if there is a conversation, you can go straight to the chat itself.
• You no longer need to search across separate screens to find what was discussed with a client months ago.
• A Calls section has been added to the file screen. If a client has multiple files, you can tag which conversation belongs to which file; anyone opening that file will only see the conversations associated with it.
• On the petition screen, you can now select a client first. The selected client’s files are listed immediately below. You can also draft a petition for a client who does not yet have a file; name and address details are still auto-filled.
• A Clients box has been added to the main Law screen, allowing one-click access to the client list. From the law section of the customer card, you can also navigate directly to the client card and the appointment form.
• Phone call and conversation lines are only visible in offices using the relevant services; if you do not use them, the card continues to work in its simplified form.

Conversation analysis page is live, voice-based emotion recognition limitation clearly stated New feature

• Conversation analysis and sentiment analysis are among the most frequently asked topics, but the site lacked a dedicated page for this subject. Terminology was scattered across different pages, leaving users unsure of which page to visit.
• What’s new:
• The conversation analysis page has been launched. What is extracted from a call is detailed under eight headings: Turkish speech recognition and speaker diarization, sentiment journey, quoted critical moments, de-escalation success, twelve-dimensional quality score, score justification, operational risk, and coaching notes.
• Voice behavior metrics are explained in a separate section: interruptions, talk balance, dead air, and overlapping speech. These metrics are calculated directly on the recording, without querying the AI model.
• The rationale behind our decision not to perform voice-based emotion recognition has been published. The EU AI Act prohibits inferring emotions from employees’ biometric data in the workplace, while inferring emotions from customer voice falls into the high-risk category. Our sentiment analysis is derived from the transcribed text of the call, and every reading is backed by a specific sentence.
• We outlined how we differ from general-purpose sentiment services: report cards instead of labels, a Turkish-specific engine, integrated panel and workflow, and clear data residency.
• The page features a six-question FAQ section, with questions marked up in a format readable by search engines and AI assistants.
• Result: Organizations looking for conversation analysis or sentiment analysis for their call centers can see what we do and what we deliberately choose not to do, all on a single page.

Send from your own email account with your own branding New feature

• You can now send emails entirely from your own account and with your own branding.
• The email step in the setup screen has been refreshed. Simply choose one of Resend, Brevo, SendGrid, Postmark, or Mailgun and paste your API key. The screen provides instructions on how to set up an account for each provider, where to obtain the key, and what to pay attention to. After saving, you can verify your settings by sending a test email to yourself using the "Send test email" option.
• When you connect your account, only your identity appears in the survey, quote, appointment, and notification emails sent to your customers: your own logo, company name, brand color, and signature. Callmenta contact details do not appear at the bottom of the email, and replies go directly to your address.
• You can manage the logo, color, signature, and signature text from the email branding section. You must use an address belonging to a domain verified with the provider as the sender address; sending from an unverified address will be rejected.
• For companies that do not connect their own account, sending continues to be handled via Callmenta as before.

The Legal screen now works like an app Enhancement

• When you open the Legal section, you are now greeted by a single home screen instead of a long menu list. Petition, case law search, analysis, contract, deep research, deadline calculation, assistant, and colleague forum are displayed side by side with large icons; you find what you are looking for at a glance, without having to read.
• If you also use the Office side, files, hearing calendar, meetings, and other office screens are arranged in a single row at the bottom of the same page. Both sides open from the same place and share the same look.
• When you enter a tool, a thin strip remains at the top of the screen: it shows which section you are in, which tool you are using, and how much of your monthly generation quota you have used. The button next to it takes you back to the home screen with a single click.
• In the left menu, Legal now occupies only a single line. This way, if you have other sections, the menu becomes shorter, and you can reach the section you are looking for without scrolling.
• Icons and colors have been adjusted to be readable in both light and dark themes, and the same layout works on phone screens as well.

Select font independently from theme New feature

• Each panel theme previously came with its own font. If you liked a theme's typography but switched to a different color palette, you would lose that font. Now, the font is selected independently of the color.
• Open the theme button in the top bar and tap the Font line at the bottom of the list. The same box switches to the font list: six options are displayed with a sample sentence, each shown in its own typeface. When you select your favorite, the panel updates instantly, and since the menu remains open, you can try them one by one.
• Your selection is saved to your account. It will be waiting for you even if you log out and sign in from another device.
• The Theme font option at the top of the list maintains the previous behavior: if you keep this selected, the font will change along with the theme whenever you switch themes.
• On mobile, the same list is available in the top bar menu. This feature also applies to the offer portal.

Timeout and loading indicator issues fixed in petition generation Bug fix

• Generation of long petitions was previously interrupted and could be retried; it now completes in a single attempt, and the server timeout limit has been extended accordingly.
• The issue where two loading indicators appeared simultaneously on the page has been resolved; the screen now displays that generation typically takes 1 to 3 minutes.

My Saved Petitions New feature

• Law firms can save the petition they are working on under a name. The saved items appear in the "My Saved Petitions" list on the right side of the Lawsuit Petition Assistant page; you can search by name, type, court, or within the text.
• When you open a saved item, the text, court header, and signature block load into the editor; you can edit and overwrite the same record, or download it as UDF, PDF, or DOCX. Opening a firm file is not required for a saved record.

Lawsuit Petition Assistant: Petitions are now written as a single cohesive text, with case law cited within paragraphs Enhancement

• The Petition Workshop has been renamed to Lawsuit Petition Assistant.
• The explanations section is no longer divided into topic-specific subheadings; instead, it flows as a single narrative in numbered paragraphs, just like in actual petitions. Claims such as divorce, compensation, and alimony are addressed within their respective paragraphs rather than under separate headings.
• Instead of being appended as a list at the end of the petition, relevant Court of Cassation precedents are cited within the corresponding paragraph, establishing a connection with the specific case and including a brief quoted excerpt from the decision’s reasoning.

Fixed issue with returning to the login screen when signing in with Google Bug fix

• Fixed the issue that caused the page to return to the login screen when signing in with Google.
• When the login button was tapped multiple times in quick succession, the first attempt was invalidated. Upon returning from Google, session verification failed, and the user was redirected back to the login screen. Since tapping the button a second time while waiting for a response is common on phones, this issue was most frequently observed on mobile devices.
• Attempts initiated within the same session no longer invalidate each other; the login completes regardless of which attempt returns. There is no relaxation on the security side: each verification key can be used only once, is short-lived, and returns with unrecognized keys are still rejected. Failed verification returns are now logged.

Document vault, timeline, and deadline assurance in legal files New feature

• Case files and Legal Intelligence now come together in one place.
• Document vault for your file.** You can save text generated in the petition and contract workshop directly to the file with a single click, in UDF, PDF, and DOCX formats. You can also upload existing documents such as preliminary hearing minutes, notifications, expert reports, or evidence directly to the file. Documents are stored in a protected area specific to your company, and their links are not exposed externally.
• Timeline.** There is a new tab on the file screen: hearings, deadlines, documents, expenses, and UYAP movements are displayed in a single stream, in chronological order. You no longer need to check five separate tabs to answer the question, "What happened in this file?"
• Deadline assurance.** You can record the final day found on the deadline calculation screen to the file with a single button, or add it to your calendar (.ics). For deadlines recorded in the file, a reminder is automatically set three days before the final day. At the top of the Legal Dashboard, deadlines expiring within fourteen days are displayed on a separate banner; expired ones remain in the list.
• Analyze documents in the vault.** You can send expert reports, decisions, or contracts added to the file to the analysis screen with a single button; there is no need to re-upload the same file. Scanned documents are readable via OCR.
• File screen refreshed.** At the top, the file number, stage, and client are displayed in a single band; next to them are buttons for "write a petition for this file," file report, and UYAP transfer. A notification-to-deadline processing form has been added to the hearing tab.
• File report.** The document list and file movements have been added to the PDF report; when heading to a hearing, the entire file is available in a single printout.
• Additionally, the file selector on the petition and contract screens now also displays files in the pending, appeal, and cassation stages; contract outputs can also be taken as PDF and DOCX.

Actual road distance, access area, and vehicle arrival point in field routes New feature

• Field team daily routes are now sorted by actual road distance and duration instead of straight-line distance; estimated arrival times are calculated based on travel time. The metric used is clearly displayed on the screen.
• "Access Area" has been added to the task map: click a point on the map to view the 15, 30, 45, or 60-minute driving area and instantly see how many visits fall within this area.
• For tasks with resolved addresses, the vehicle's required arrival point is also saved; the directions on the task screen lead to the roadside stopping point rather than the center of the building.
• Address fields in customer and task forms now show suggestions while typing; selecting a suggestion automatically plots the location on the map. The "Find on Map" button also provides more accurate results at the street number level.
• These features are intentionally disabled for healthcare visits; patient addresses are not sent to external services.

Plugin description simplified Enhancement

• On industry pages, in the legal plugin introduction, and in presentations, the relationship between plugins and packages is now described as "added"; the previous "overwrites" terminology has been removed.

Fixed invisible icons on mobile and narrow screen layout issues Bug fix

• Some icons in the panel appeared blank or with incorrect symbols on both mobile and desktop; all icons outside the standard icon set now display correctly.
• On mobile, the workspace collapses into a single column, preventing widgets from being squeezed into narrow boxes. The "open phone in a separate window" warning no longer appears on mobile and tablet, and it no longer overlaps card content.
• The filter row in the CALLAII score dashboard, KVKK compliance report tables, game page header, and team profile header no longer overflow on narrow screens; the page no longer scrolls horizontally.

Broader candidate pool in legal search Enhancement

• The number of candidates re-ranked in case law and legislation search increased from 40 to 150, and the results list expanded from 8 to 12 decisions. Fewer relevant decisions are lost in the first pass; search time remains unchanged.

Legal Intelligence case law library is now updated daily New feature

• The case law library was previously a one-time loaded archive; the latest decision remained stuck at June 2026. Now, new decisions are automatically fetched every morning from the UYAP Emsal, Court of Cassation, and Council of State decision search portals, and they become available for both keyword and semantic search on the same day.
• Since Court of Cassation and Council of State decisions are typically published months after the ruling date, the feed scans with a retroactive window; a weekly broad scan completes any decisions that could not be retrieved temporarily. Previously fetched decisions are not re-downloaded, and personal data is masked before entering the library.
• A Legal Library page has been added under CALLAII in the admin panel: decision count per source and the latest decision date, number of chunks awaiting vectorization, scheduler plan, and recent runs are visible here. The feed, broad scan, and vector completion can also be triggered manually from this page; the job runs in the background and its progress is tracked on the page.
• Precedent decisions displayed in the petition and legal analysis screens are now sourced from this updated library.

Petition types increased from 46 to 61 New feature

• Fifteen new types were added: divorce due to adultery, abandonment and criminal act, separation, interim alimony, removal of alimony, annulment of marriage, permission to marry, termination due to absence, child custody, liquidation of property regime, interim injunction, interim attachment, prevention of interference, and constitutional objection.
• Each type comes with specific warnings for that case: which cases are filed without a defendant, which rights are lost if not reserved, and which deadlines are forfeiture periods. All legal bases were verified against the legislation database.

Texts citing repealed laws now trigger warnings Bug fix

• References to all fully repealed laws, such as the 818-numbered Code of Obligations, the 1086-numbered HUMK, and the 743-numbered Turkish Civil Code, are now detected and the replacement law is shown. This check runs on both your drafts and the opposing party's petition.
• Additionally, a gap in the Constitution text was fixed: because a section of the articles was registered with two separate abbreviations, it was not appearing in searches.

The grounds in the opposing party's petition are being verified New feature

• When you paste the opposing party's petition into the system, the decisions and law articles it relies on are scanned against the case law library. Non-existent case citations, repealed laws, and article references that do not match the subject are listed in a separate box.
• This allows you to respond with a concrete finding: while it is not possible to reliably determine whether a text was written by artificial intelligence, whether the cited case is current can be verified.

Petition drafts no longer write anything that was not explained Bug fix

• In petition generation, artificial intelligence used to fill in unmentioned gaps: a dependency, insult, or crime accusation that never appeared in the description could enter the text, and unwanted requests could be added. Now, if an accusation, date, or amount that was not mentioned in what you described enters the text, the system catches it, shows it to you by name, and stops that draft from being downloaded.
• Drafts now develop each request separately under its own heading. Relevant Supreme Court decisions are added to the draft with their citations; if no suitable decision is found, the reason is written on the screen. In similar cases, defenses raised by the opposing party are also presented as suggestions.
• Requests and missing information that were not included in the description, but left incomplete, are collected in a separate lawyer note without mixing into the text sent to the court.

Legal assistant runs on its own server Bug fix

• When the legal assistant's own AI server was not in use, requests ended with 'response could not be generated' for petitions, contracts, analysis, and chat. Now, the request is preserved until the server wakes up; if the server cannot be reached, the response is generated by the general CALLAII model, and this situation is clearly reported to the lawyer.
• The server's startup time is displayed on the screen as the actual duration; the information you enter is not lost, and the request is automatically retried.

Case law search scope expanded Enhancement

• Case law search has two branches: keyword search and semantic search. When a long question is asked in natural language, the keyword branch required every word in the question to appear in the same decision fragment. Since it is rare for all eight words to co-occur, this branch returned empty results and the search effectively ran on a single branch.
• Now, distinctive words are identified from the question. Terms that are rare and carry meaning, such as "dressed" or "eviction undertaking," are selected for the search; common words that appear in nearly every decision, such as "considering" or "due to," are excluded from narrowing the results.
• Measurement: across five real legal questions, the number of candidate decisions found by the keyword branch increased from 7 to 516. Because both branches now operate, results are both broader and more accurate.
• Additionally, the "searchable fragment" ratio on the Legal Intelligence dashboard is now genuinely measured. Previously it was displayed as a fixed 100 percent; going forward, if library preparation lags behind, the dashboard will reflect that.

Tenant approval for customer data access is now enforced on every screen Enhancement

• Previously, Callmenta team members could only view your company data with your explicit approval. This rule is now applied without exception across the entire panel.
• In the past, certain screens fell outside this gate: HR and payroll records, time tracking, shift schedules, evaluation forms, knowledge base documents, announcements, and customer cards. If a screen opened without a company selected, the scope widened. Now, if no approval is in place, these screens display no data.
• The approval flow remains unchanged: a request is sent to you as a notification, you approve or reject it from the Organization > Security screen, and access automatically expires when the time limit is reached. Data from a company that has not granted approval does not appear in lists, searches, reports, or exports.

Company administrator can manage records in health screens Enhancement

• The company administrator could access the health module screens but was unable to add or edit records. Now they can manage their company's health records: clinic, physician, and service definitions, forms, field teams, ambulance vehicles, and patient score recalculation.
• The permission list for the entire health module is maintained in a single location; the button displayed on the screen and the underlying operation always read the same permission.

Resolved permission conflict on the ambulance vehicle screen Bug fix

• On the ambulance vehicles screen, some users could open the add/edit vehicle form, but saving was rejected. Viewing the form and saving permissions are now determined from a single source, so the two can no longer be separated.
• Supervisors can still view the vehicle list and assign vehicles to cases; they cannot add or edit vehicle records. Vehicle assignment to cases also goes through the same single gate, so users without permission cannot perform this action through any other path.

Only your own line items on the quote screen Bug fix

• For companies with the Quote module enabled, the line-item picker now shows only your own registered service items. Previously, Callmenta's own packages also appeared in the list; they have been removed.
• Your quote template, item list, and numbering are entirely yours. The items you enter in your first quote are saved automatically and will appear in the selection list for subsequent quotes.

Proposal document fits on a single page whenever possible Enhancement

• The proposal PDF now measures whether it fits on a single page. Proposals that overflow to a second page due to an increased number of line items are condensed onto one page by slightly reducing font size and spacing.
• Proposals that already fit on a single page remain completely unchanged and look exactly the same as before. Proposals that would only fit on one page if shrunk to an unreadable size are left as they are; readability is never compromised.

Cards on the sales board are easy to drag Bug fix

• Previously, you had to grab only the small handle to move a card on the sales board; dragging from the card body itself did nothing. Now you can pick up and drag a card from anywhere on it. The buttons, links, and stage selectors on the card continue to work as before.
• Dragging also works the same way on touchscreens, and the card won't slip out of your hand while you scroll through columns.

CRM list redesigned from scratch Enhancement

• The boxes in the customer list are now uniform in size. The summary boxes at the top are displayed with equal width, and the stage and temperature filters have been converted into single-size chips, each showing the record count next to it.
• Search and filter fields have been standardized to a single font size and height. Tag selection is now done via a searchable menu rather than a scrolling list. The stage, temperature, and tag badges in the table have also been aligned to the same visual language and remain readable in both light and dark themes.
• Company cards are laid out in equal sizes; even if the card content expands, the buttons stay aligned on the same baseline.

Color themes have been simplified Enhancement

• The panel's color themes have been streamlined. Instead of sixteen themes, there are now four palettes, each available in light and dark variants.
• Palettes**
• Callmenta**: brand yellow, light and dark
• Clay**: warm terracotta tones
• Cocoa**: cream and brown tones
• Steel**: metallic gray with red accents
• When choosing a theme, you now select the palette first, then whether to use the light or dark variant. Each card in the menu displays a small preview rendered in that theme's actual colors.
• What changed**
• Charts, tables, and badges now use the colors of the selected theme. The separate chart palette option has been removed, so there is no longer a need to pick one independently.
• All text is rendered in the panel's own font, including chart labels.
• In every theme, text-to-background contrast is kept above the accessibility threshold.
• The color-blindness-friendly mode continues to work as before.
• If a theme you previously selected has been removed, it has been migrated to the closest new palette. No action is required on your part.

Queue dashboard displays the real-time status of the switchboard Bug fix

• The agent counts on the queue dashboard are now read directly from the switchboard.
• Previously, the dashboard only recognized agents who sent a pause notification to the switchboard. Agents who did not send a notification did not appear in the list at all, making the queue seem to have fewer members than it actually did. A paused agent might not even show up in the "paused" row.
• Three things have been fixed as a result:
• An agent whose phone is not registered on the switchboard is no longer counted as available. Such an agent is not on break, but their phone will not ring; the dashboard now displays this separately as "phone unregistered" and raises an alert if the queue is actually unstaffed.
• When no one remains in the queue who can take a call, a notification is sent to the supervisor. Previously, this was only noticed when someone went on break; now it is also caught when the last phone's switchboard registration drops. The notification is only sent during shift hours.
• When the switchboard is unreachable, the dashboard does not go blank — it displays the last known state and notes below the row that this is not current.

Legal Intelligence: Search accuracy, citation validation, and speed improvements Bug fix

• Relevance scoring in case law search is now calculated correctly. Searches based on fact descriptions and precedent scans return far more accurate results; the "no source found" warning now appears only when a source is genuinely unavailable.
• When searching by decision number (e.g., 2019/1234), decisions with that number appear at the top of the list.
• Search time for long fact narratives dropped from 6–9 seconds to under 1 second; petitions, analyses, and deep research now complete faster.
• Articles from the Labor Code, Civil Code, and Constitution are now correctly cited in petition and contract grounds; references written with legacy abbreviations are also recognized and validated.
• Repealed articles have been removed from search results and citation lists.
• A new warning has been added to the audit card: articles and decisions mentioned in the text but not provided as grounds for this output are flagged separately, so the lawyer can see exactly what needs verification.
• The Legal Intelligence dashboard previously took over a minute to load; it now opens in under one second.
• Formatting artifacts in contract and analysis outputs have been cleaned up; text now arrives plain and readable.

Live support inbox view has been redesigned Enhancement

• The chat list, message area, and visitor panel on the live support screen have been made more readable. Chat cards now stand out clearly when selected, the input box has been consolidated into a single frame, and visitor information along with live coaching metrics are displayed in separate cards with neatly aligned rows. Colors in both light and dark themes are consistent with the theme.

PBX connection test now covers all providers Bug fix

• The Connection Test button in Company Settings now works when SIP Trunk, Asterisk, Twilio, or Avaya is selected; previously it would show an incorrect warning for these providers. For SIP Trunk, a real validation is performed: the server address is resolved, connectivity is attempted over the selected transport, and if Register is enabled, credentials are tested and the registration target is verified. Each step's result is shown individually, so you can see exactly where it fails.

Files and questions can now be sent to the CALLAII assistant in a single step Enhancement

• The requirement to upload a file first and then type your question in the assistant chat has been removed. When you select a file, it is added to the chat as a small tag; when you type your question and press Send, the file and the question are sent together. You can attach multiple files and remove any of them before sending. If you send only a file, the assistant will review it and provide a summary. You can also paste an image copied to the clipboard directly into the chat box.

Top bar layout redesigned for phone and tablet Bug fix

• On small screens, the theme, language, shortcut, and admin filter buttons in the top bar were overlapping each other and covering the menu button.
• These options are now grouped in a drawer that opens from the three-dot button in the top-right corner on phone and tablet. The drawer lists theme, chart palette, accessibility, language, shortcuts, and bug report in a single list; selecting an option updates the view instantly.
• The top bar now retains only the company selector, live support, messages, notifications, and profile; no element overflows anymore.

Outbound call pool, post-sale handoff, and company-specific metrics New feature

• Three new capabilities.
• Outbound call pool: An admin can place a call list into the pool without assigning it to anyone. Users in the company see the pool in their workspace and claim a record themselves; the same record is never routed to two people. The number of records each person can hold simultaneously is configured per list. A claimed record can be returned to the pool, and the admin can also retrieve it from the list view.
• Post-sale handoff: When an agent closes a call as a sale, the record is automatically routed to the team defined by the company: field visit, request (web transaction, invoice, etc.), or sales opportunity. Rules are defined in the Organization menu; which outcome, which target, to whom, and with what title. The same call is never handed off twice. A "Sale" option has been added to outbound call outcomes.
• Company metrics: An admin defines their own dashboard figures: source (calls, outbound call items, opportunities, requests, field tasks), filter, and period. For example, "Calls converted to sales" or "Records where Site = Ankara." The defined metric is then selected as a tile in the dashboard layout.
• Custom fields for outbound call lists: Custom fields defined in the CRM (site, company-specific code, etc.) can now be mapped during outbound call list import, are visible on the agent screen, can be filled in during the call, and are written to the customer card.

Law firm: case law search, legal analysis, and petition workshop New feature

• Three new sections have been added to the law firm module.
• Case law search: Search across Supreme Court, Council of State, Constitutional Court, and local court decisions. Each result includes the case and decision numbers; filters for court and year are available. The default search covers the last ten years and can be expanded with a single click.
• Legal analysis: From a case description, it performs precedent screening, decision analysis, contract review, and identifies potential objections the opposing party might raise. Precedent screening displays favorable and unfavorable decisions separately. Every decision number in the output is verified against the database; any reference that cannot be found is flagged in the text with a warning.
• Petition workshop: Petition text is downloaded in UDF format and can be uploaded to the UYAP lawyer portal. After the file is generated, it is read back and compared with the text; if the comparison fails, the file is not delivered. The downloaded file is unsigned; e-signature and submission are the lawyer's responsibility.
• Date and deadline calculations are not performed in these sections; the deadline calculation screen continues to work as before.
• Each of the three sections can be enabled separately. Law firms that wish to use them can contact us.

Have reports delivered on the day and time you choose Enhancement

• Report subscriptions now let you pick the delivery day and time: weekly reports go out on the day of the week you choose, monthly reports on the day of the month you choose, and daily reports at the time you specify. If no selection is made, the previous schedule remains in effect. The subscription list shows when each report will be sent.

Call Criteria Scorecard Report New feature

• A new report has been added to the Report Center: an aggregated view of criteria measured in calls and chats. The report shows which agent fell short on which criteria and how many times, the most frequently violated criteria, weekly trends, and a list of violations with evidence. It can be filtered by business line and branch, and exported as PDF, Excel, or CSV. If a manager rejects a result, the corrected version is reflected in the scorecard.

Sensitive record views are now logged Enhancement

• Access to customer cards, call records, policies, and complaint screens is now written to the audit log, recording who opened what and when. If the same user opens the same record again within one hour, it is not counted as a separate event. When a user is deactivated, their open session closes on the next click.

Two-factor authentication can be made mandatory Enhancement

• Company administrators can enforce two-factor authentication for admin roles; users who have not set it up are redirected to the setup screen on their next login. Password rules have also been strengthened: a minimum of 12 characters, a combination of letters and digits, and passwords found in breached password lists are rejected.

Support access is now subject to your approval New feature

• The Callmenta support team can view your company's data and settings only with your approval, and only for the duration you specify. Requests appear on the Organization > Security page; you can approve, reject, or revoke open access at any time. All access is logged and automatically expires when the time limit is reached.

Free trial guidance on the purchase page Enhancement

• When card payment is not yet available, the purchase page no longer leaves you hanging. A link to open a 14-day free trial account now appears below the order summary, and you can still reach out to us if you prefer. The copy has also been simplified: instead of explaining who defines the package and how, it now focuses on what you can do.

Per-site embedding, brand and campaign tagging, and listing linkage Enhancement

• The service appointment form and live chat are embedded per site with a separate token; the source and brand are populated from the moment the request arrives, and the branch is automatically assigned based on the brand. If a campaign tag is added to the site URL, the appointment and chat are recorded under that campaign. A used-vehicle listing status and listing address field have been added to the vehicle inventory. Multiple WhatsApp numbers are consolidated into a single inbox, replies are sent from the number the customer wrote to, and the waiting time for a response is displayed in the list. A video vehicle presentation option has been added to the appointment types.

Customizable dashboard New feature

• Customizable widgets have been added to the admin dashboard, tailored by company and role: open alerts, criterion violations, open complaints, SLA-exceeded complaints, policy renewal rate, policies expiring within thirty days, today's calls, and open work orders. In the Dashboard Layout screen under the Organization menu, you can select which widgets each role will see; a maximum of four widgets can be placed on a single row. Widgets for closed modules cannot be selected, and the layout can be reset to the default.

Insurance module separated from automotive New feature

• Insurance is now its own module. A policy is linked to the customer; vehicle, residential, or business can be added as an optional subject. All lines are now available: comprehensive, traffic, personal accident, health, supplemental health, life, residential, DASK, business, travel, cargo, liability, agriculture, and individual pension. Commission rate and amount, payment schedule, endorsement, claim file, quote record, and previous policy link are all present; when a renewal is opened, the old policy is automatically marked as renewed. The renewal rate is displayed at the top of the list. Existing vehicle policies have been migrated to the new structure, and the insurance-expiring list on the service dashboard remains unchanged. For health, life, and individual pension lines, only the policy record is stored; notes, declarations, and documents are not recorded.

Unclear lines are re-listened from the audio Enhancement

• Lines flagged as suspicious in the transcript are now verified by re-listening to that section of the recording. If the audio is clear, the line is corrected or confirmed; if it remains unclear, the line is marked as "[unintelligible]" and the quality score does not rely on those lines. Announcements leaking from the speech recognition system are filtered at the source, and a confidence value is stored for each speech segment; segments with low confidence are automatically flagged. A configuration issue that caused long analyses to restart in the queue has also been fixed.

Campaign code in subscription New feature

• A campaign code field has been added to the subscription screen. If you have a code, simply enter it at the payment step: the discount is applied to the amount immediately, and the payment is processed at the discounted rate.
• Codes can be defined with separate rates for monthly and annual payments, restricted to specific packages, and closed based on usage count. The discount remains valid on renewals for as long as the subscription is active.
• For visitors arriving via a campaign link, the code bypasses the account creation step and is already ready in the payment screen.

Complaint and Reputation Management Module New feature

• The new Complaints module does more than surface complaints — it closes them. Low satisfaction scores, complaint-risk labels from call and chat analytics, negative support requests, and unresolved live-support sessions are automatically scanned daily and consolidated into a single list; the same record is never opened twice. Şikayetvar and Google reviews flow into the same list once a corporate API key is connected; manual entry and CSV import are also available. The system suggests the CRM customer based on the phone number, e-mail, plate number, work order, or policy number in the complaint; a manager approves the suggestion, and the confidence score and rationale are retained in the record and can be reversed. Complaints are linked to work orders, call recordings, opportunities, and requests; an owner and branch are assigned, SLA timers run according to priority, root cause is tagged, and the record is closed with a closure note. Unassigned, unanswered, repeatedly root-caused, or owner-concentrated complaints trigger alerts from the alarm engine. CALLAII drafts the public response; no response is sent without approval. Complaint text containing health-related expressions is visible only to the manager role, and closed records are deleted once the retention period expires. A Complaint and Reputation report has been added to the reporting center.

Call Criteria: What Should and Should Not Be Said New feature

• The company now defines its own call criteria: required statements such as whether the KVKK announcement was given, whether the service fee was disclosed in advance, and whether an appointment confirmation was obtained; and prohibited statements such as guaranteeing a definitive repair time, making absolute claims about the warranty, or making negative remarks about competitors. After each call and chat analysis is completed, it is scored against these criteria; the outcome is recorded as pass, fail, or indeterminate, with a transcript quote and timestamp. No decision is made without evidence. A keyword match yields a definitive result, while the model's assessment serves as a secondary view; if the two conflict, the result remains indeterminate and no alert is generated. In case of a violation, a single email and dashboard notification are sent to the selected managers. An agent may appeal the result, and the manager either approves or rejects it; a rejected outcome is not written to the record. If the IVR system delivers the announcement on an inbound call, the agent is not marked as deficient. Results are generated in batch after analysis, not in real time.

Alarm engine: alerts that catch what is missing, not just what is present New feature

• The new alarm engine raises alerts not only when something exists, but also when something that should exist is missing. It triggers when a policy renewal offer has not been issued with 45 days remaining until expiry, when a claim has been unanswered for 48 hours, when a vehicle on the books still carries a policy, when a customer with comprehensive insurance has no DASK policy, or when a vehicle arriving at the service centre has an expired comprehensive policy. Rules are defined in the Organization menu; the time threshold, priority level, and line of business are all configurable. Notifications are sent only to the company's own users, once per alarm, and do not repeat until the suppression window has elapsed. When the underlying condition is resolved, the alarm closes automatically; managers can also close it with a note. Records are checked in real time as they change, and additionally in a batch sweep every fifteen minutes.

Business line distinction applies to all records Enhancement

• Business line filtering now applies not only to the customer list but also to vehicles, sales opportunities, work orders, appointments, branches, and the service catalog. A user assigned to the insurance line cannot see vehicles, work orders, or appointments belonging to the service line. Records created by a user with a single business line are automatically assigned to that line. For companies that have not defined business lines, nothing changes.

Very short calls are no longer scored Bug fix

• Calls whose transcripts contain only one or two words were sometimes receiving a quality score. Calls with fewer than twenty words are now flagged as short calls, are not scored, and are displayed separately in reports. Certain fixed phrases originating from the speech recognition system that do not constitute actual speech are also being removed from the transcripts.

Unintelligible lines in transcripts are now flagged New feature

• When a phone line is faulty, the speech-to-text conversion can produce nonsensical sentences that look like real dialogue. Now, every transcript line is reviewed; unintelligible lines are clearly marked with a "speech not understood, text unreliable" warning and displayed in a muted style. If the ratio of such lines is high, the transcript's confidence score drops and the call is routed to the queue of records requiring review. These lines are prioritized in the transcript correction screen; the corrected text is used to improve the speech recognition model.