AgencyKey — What’s New
65 subscribers
6 photos
20 videos
5 links
Product updates and short videos about AgencyKey — the CRM for creators and agencies.

AgencyKey.com · Support: @AgencyKey_bot
Download Telegram
.
🔍 AI Fan Recon

In most CRMs, a chatter walks into a fan's chat blind.

Doesn't know what the fan bought from other creators. Can't see which openers the fan reacts to and which make him bounce. Doesn't know if he's Silent, Warm, or Talker. The first message becomes a guess, not a tactic. And every early mistake eats into conversion.

AgencyKey works differently.

The moment a chatter opens the chat, they get a tactical briefing on the fan — AI analyzes the fan's available conversations from the permitted pool of AgencyKey creators. It sees which approaches led to purchases and which didn't. It knows which phrases the fan responded to with money.

Here's what it looks like in practice:

Mike, 37, New York. Office job. Has a girlfriend but turns to OF for what his real relationship doesn't offer: dom/sub roleplay and foot content. UFC fan, hits the gym 4 times a week, loves traveling. Buyer type: Warm — buys after 8–15 messages. What worked: talking about workouts, soft tone, PPV in the $15–25 range, and custom videos on request. What didn't: pushing expensive PPV early, without warm-up. Trigger phrases before buying: "show me more," "i want to see that," "sounds good" — any of them usually lead to a purchase.

The chatter sees all of this in seconds, before writing a single word.


👤 About the fan

• name, age, location, job, relationship status
• interests, hobbies, conversation topics
• kinks, fetishes, and limits
• emotional profile — what hooks him, what he responds to


🎯 What works with him

• buyer type and how many messages before the first purchase
• sequences that already led to payments
• approaches the fan ignored
• the fan's own phrases right before buying


🔐 Access and sharing

Viewing and generating AI Recon use separate permissions. Both are enabled by default for every standard role; review them separately for custom roles. By default, only Owner and Admin can change the sharing mode.

Chat sharing itself is fully optional. Without it, AI Fan Recon doesn't work — but every other AgencyKey feature runs as usual.

The agency chooses the sharing mode in settings:
• share only within your own agency
• or across the full CRM with other participating creators

The decision is always the agency's — and the setting can be changed at any time.


AgencyKey — the key to every fan. From the first message.
1👍1🔥1
🔍 Chat Summary

One and the same feature covers several tasks in an agency at once.

Chat Summary is a quick recap of a specific conversation with a fan: what you talked about, what they bought, where they refused, what’s happening right now, and what the obvious next step is.

Where this is especially useful:


🔄 Shift handoff

One chatter logs off, another enters the chat with no context. Instead of “read the last 300 messages,” the new one gets the key points and continues the conversation without a drop in tone or sales logic.


Returning to an old chat

You haven’t written for 3 days, 180 messages piled up, and the fan has already taken the conversation in a new direction. The summary restores the full picture in seconds: what changed, where the money is, where the interest is, and where to be more careful with the offer.


💸 Preparing for PPV

Before an important sendout, it helps to understand what they’ve already bought, what the fan responded to, where there was a refusal, and whether it’s too early to push the next offer. The summary helps you avoid shooting blind.


👀 Manager control

A manager doesn’t need to read the full history to check what’s happening in the chat. It’s enough to open the conversation and immediately see the short, medium, or detailed picture.


This is a recap of one specific conversation, not a fan profile. AI Fan Recon looks at the fan more broadly, using available conversations from the permitted pool of creators. The task here is different — to quickly restore the context of one chat.


🔐 Access

Viewing and generating Chat Summary use separate permissions, both granted by default to Owner and Admin. Generate Chat Summary is also required for manual and automatic generation and for changing summary settings. Configure these permissions separately for custom roles.


⚙️ What can be customized for each chatter

• analysis depth: up to 500 messages
• result length: Short / Medium / Detailed
• your own prompt with the right focus points (upsell, triggers, leak risks)
• auto-generation when opening a chat (optional)


💡 Bonus

• notes on the fan are automatically pulled into the context
• reopening it again — the summary loads instantly from cache
• translation into your selected language when Chat Summary translation is enabled separately


AgencyKey — the key to every chat. From the first click.
1👍1🔥1
🤖 AI in AgencyKey does what's impossible in other CRMs

When something starts slipping inside your agency, you need to catch it before it hits revenue. If that takes an hour buried in dashboards, you don't have a tool — you've got dead weight.

AgencyKey gives you that tool. Through MCP, Claude and ChatGPT handle your requests in a single command.


💬 What AI can do beyond the interface

1. Team anomalies
Find chatters who had a high message volume over the last 14 days, but low revenue per conversation. Sort by deviation from the average

In the UI, this means manually piecing together multiple slices — AI immediately gives you a review list, flagging likely activity imitation.

2. Causes of model decline
Show me creators with a decline for 3 weeks in a row and compile likely reasons based on CRM data: traffic, team shift changes, conversion, content

Not just a decline chart, but a list of creators with hypotheses on what exactly went wrong.

3. New model setup in one command
For creator @name — set up a proxy, copy smart lists from @other, assign shifts with Anna and Kate

5+ actions across different CRM sections — in a single command.

4. Report for a call
Build a report: revenue, traffic, ARPU, TOP marketers, WoW dynamics. Format it in HTML for the client

Not a generic dashboard, but a ready document for a specific request.

5. Whale segment purchase patterns
Analyze conversations with the whale segment after large purchases and show the trigger phrases that appear most often before high-ticket spends

AI finds patterns across thousands of messages — you can't pull that together manually.

6. Whale retention segments
Whales whose spending dropped sharply over the last 14 days — sort by past LTV and comeback potential

Not a template segment, but a retention sample built for a specific task.


📤 Export and import

Your data stays yours — any slice on request. If you're moving, AI parses scripts and settings from another CRM and imports them into AgencyKey.


AgencyKey — the key to management. From the first command.
👍21🔥1
What’s new in version 2026.5.10

🔗 Tracking Links — the table has been redesigned: it is now more compact, clearer, and easier to understand. Performance columns have been combined, groups have been added, and a campaign details modal has been introduced.

📨 Mass Mailings — the modal UX has been redesigned: the name and note can now be edited for non-DRAFT campaigns, duplication has been fixed, and numeric field filters have been improved.

📊 Analytics — the default range is now 30 days, with date selection limited to 90 days.

💬 Unanswered Chats — gradual scanning has been added, with Stop and Scan more buttons.

👥 Search in Accounts/Team — search has been added; employees are now displayed in two lines, and sizes have been generally revised.
👍1🔥1
What’s new in version 2026.5.11

🔔 Account notifications page — a new separate /my/notifications page with expanded functionality.

PPV/Tip badges in the chat list — now update in real time.

📨 Mass mailings with media and text — sending has been fixed.

Scripts — automatically marked as “Sent” after insertion, editing, and sending.

💸 Scripts — when sending a script, the text is no longer marked as paid.

🎧 Audio upload — now works.

🖼️ Media drag-and-drop — fixed media order changes when sending a message.

🧭 Sidebar — right-click, Ctrl-click, and middle-click now work correctly in the web version.

🤖 AI Assistant — the email code field on claude.ai no longer breaks.

💬 Chat — removed the “free” marker from message text, as it was redundant.
🔥2👍1
What’s new in version 2026.5.12

⌨️ Custom hotkeys — users can now configure any keyboard shortcuts for themselves.

🌍 Translation directly in the chat list — the translated message is now visible in the preview without opening the chat.

🚧 Hint for platform restrictions — if the platform temporarily blocks actions, a clear message appears with an automatic solution.

🔔 Notifications are more convenient — clicking a paid-message notification now scrolls directly to that message in the chat.

🖱️ Drag-to-scroll tabs — chat tabs can now be scrolled by dragging with the mouse.

Translations load faster — repeated translations are taken from memory, without waiting for a server response.

🧠 Chat sidebar remembers its state — when collapsed and expanded, the selected tab is preserved: Insights / PPV, scroll position, and loaded data. The next time the chat is opened, the panel opens on the same tab where it was closed.

🖼️ Gallery — fixed display after the first opening.

Scheduled messages — after sending, the message immediately disappears from “scheduled.”

🔐 Flexible access rights in Mass Mailings — permissions in the Mass Mailings section can now be configured in more detail: separately allowing creation, editing, deletion, and sending. Buttons are automatically hidden from employees who do not have the corresponding permissions.
👍3🔥2
What's new in version 2026.5.13

🌐 Built-in proxies — models without their own proxy now automatically connect through a built-in one, works more stable and faster

📁 Vault — all media formats — you can now upload any file format supported by the platform. A watermark with @username is added automatically

⚠️ Send errors no longer get lost — previously the platform could silently reject a message, now you'll see the error right away

📋 Support logs — if something goes wrong, you can send the error log file to support and the issue will be resolved faster

🚧 Clear message when the platform fails to load — if the page doesn't load, a banner appears with a "Retry" button

🔔 Accurate unread counter — now matches the account counter

🛑 Mailing speed limit — configure the maximum sending speed

✏️ Correct name substitution — in mailings, {name} now always inserts the name of the exact fan the message is being sent to
👍4
What's new in version 2026.5.14

Instant status updates for delayed messages — if a delayed message is canceled, sent, or fails — you see it right away, no waiting

Delayed messages cancel correctly — the «cancel when new message is sent» option now works reliably. If you try to cancel an already-sent message, you get a clear hint

⚠️ Clear error on send failure — if the platform rejects a message, the reason is now shown directly on the failed bubble in the chat

📊 Correct progress for large video uploads — the percentage no longer jumps around, it now shows real progress

🛑 Canceling large file uploads works — previously, removing a file mid-upload left «ghost» placeholders; cancellation is now clean

📋 Screenshots from clipboard no longer duplicate — pasting a screenshot via Ctrl+V creates one file, not two

📎 More stable media upload in chat — other media attachment bugs fixed

🎯 Scripts switch only on click — switching between scripts, steps, and tags no longer triggers accidentally on hover
🔥3
What's new in version 2026.5.15

🔐 Two-factor authentication (2FA) — you can enable sign-in via an authenticator app (TOTP) or email code. Recovery codes are available in case you lose access

📁 Faster Vault publishing — media upload is now parallel and faster, cancel-related issues fixed

🔍 Search in chat list dropdown — search in the chat header dropdown now works correctly in the creator chat list
2
What's New (versions 2026.6.0 — 2026.6.2)

🔄 Forced update for outdated versions — if the app is badly out of date, a window appears prompting you to update

💬 «Contact support» button — requests are tied to your account, so the operator instantly sees who's writing

🆔 Convenient Face ID login — the verification link is shown in a dedicated panel over the app

🔌 Fewer false «session lost» messages — shown only on a real creator-account session loss

♻️ Auto-recovery after updates — if part of the interface didn't load after a new version, the app reloads itself

🛡️ Protection against crashes when the disk is full — the app no longer crashes when you run out of space

Permission for delayed messages — you can configure which employees can schedule messages for delayed sending

📁 Vault is more convenient — uploaded media is properly ready to send, and unnecessary system categories are removed from the selector
👍4
What's new in version 2026.6.3

✏️ «Paid text» checkbox off by default — when creating a mass message, the text is free by default; you enable paid manually

🔒 Proper field locking in mailings — fields that can't be edited are now correctly disabled

📁 Vault — gentler with the platform’s daily limit — uploads run in batches of 20, and on hitting the limit you get a clear error instead of a failure

📨 Fixed mass send with Vault media — such mailings could previously fail to send; now they go through correctly

Delayed send via the native calendar — scheduling now works correctly
👍3
What's new in version 2026.6.4

🚫 Exclude recent chats — in priority mailings, you can skip people you've recently talked to. The period is configurable

🔍 Search by link in Tracking Links — you can now find a campaign right by its link, with all filters visible at a glance

🌍 Time zone tied to your account — your selected time zone is now saved to your account and applied consistently across all devices and windows. You can set it in Settings → Preferences →
Date & time. If the time looks off, just re-select your time zone there

🔐 More flexible employee permissions — you can now fine-tune access to Collections, Queue, and Settings, giving each employee exactly the capabilities they need

⌨️ Keyboard shortcuts work everywhere — your custom key combinations apply across the entire app
2🔥2👍1
What's new in version 2026.6.5

📊 Response time in employee analytics — a new «Response time» column with a detail popup: see how quickly employees reply to fans

🤖 Configurable data sharing for AI — you can now fine-tune which data is used for AI Recon and CRM statistics

📁 Faster bulk upload in Vault — large media batches upload faster, with cancellation support and better video handling
🔥2👍1
What's new in version 2026.6.6

🤖 Automated (trigger) campaigns — a big new feature! You can create mailings that send automatically based on an event:
- 👋 Welcome new subscriber — when a fan just subscribed
- 💰 After a purchase — when a fan buys a PPV
- 🔄 Resubscribe — when a fan renews their subscription
- 👀 When a fan comes online — the message is sent automatically as soon as the subscriber appears online

(triggers «Expiring soon», «Expired», and «Birthday» are coming later)

🚫 Flexible settings for trigger campaigns — you can exclude recent chats and set a period so the same fan doesn't get a trigger too often

📊 Clearer response-time breakdown — improved popup with employee response-time details

🎥 Camera and microphone in streams — camera and mic now work in the embedded creator workspace for live streams
👍2🔥21
🧪 AgencyKey Beta — new chat workspace

We've redesigned the workspace for chatting with fans and released it in a beta of the desktop app.
Download: agencykey.com/download/beta

What's new:

🗂️ Tabs are now on top — creator tabs moved from the left menu to the top, the familiar browser-like way. And CRM sections can now be opened in separate tabs too —
previously not possible

📊 Redesigned analytics panel — Notes, AI Summary, and AI Recon moved into a single analytics panel on the right. You can reorder the blocks and collapse the ones
you don't need, and the blocks themselves are now much more compact

💬 More room for the chat — these three blocks used to sit above the conversation and take up space. Now they're in the right-side panel, always in view, leaving
more room for the chat itself

📌 Models sidebar — quick access to all your models in one place

🔜 Coming soon — a redesigned chat list and the blocks above it: new design, badges, and notifications window

👉 This is a beta — try it and share your feedback; it'll help us polish the interface for release.
Download: agencykey.com/download/beta
👍2🔥21
🧪 Beta 2026.6.3 — what's new

We keep polishing the new interface. Here's what landed in the latest beta:

🆕 Refreshed chat list header — reworked the filters and lists above the chats: filter chips and a handy dropdown (system, smart, and custom lists). The redesign of the chat cards themselves is still in the works

🖥️ Workspace alignment — on wide monitors you can pin the creator workspace to the left or right edge, or keep it centered

🌐 Notes translation — notes in the analytics panel can now be translated, and brief blocks are no longer height-capped — they expand to fit the content

💬 Support via QR code — support links now open with a QR code, so you can quickly switch to your phone

🧹 Removed the «Hide answered» filter — it was just getting in the way

👉 This is a beta — try it and share your feedback; it'll help us polish the interface for release
🔥2
🧪 Beta 2026.6.3 — what's new

🆕 New toolbar above the chat list + sorting — refreshed the panel above the chats and added sorting: show chats newest-first or oldest-first

🕒 Easier time zone picker — instead of a huge list of hundreds of zones, they're now grouped by UTC offset (e.g. «UTC+2 — Kyiv, Helsinki, Athens…»), and
search by city still works. Settings → Preferences → Date & time

🔗 Cleaner marketing links table — switching models now shows a loading state instead of stale rows, and the delete confirmation clearly states what's
being deleted

🔌 More reliable login and faster startup — fixed some auth failures on cold connections and after the computer wakes from sleep

🌐 Proxy: the «Host» field accepts a plain hostname — e.g. isp.decodo.com is no longer rejected as invalid
2