Features

Everything an AI power-user needs

One clean workspace instead of five subscriptions and ten tabs — and almost every part of it is a setting you can change.

397AI models
9providers
265modules
32task templates
20built-in skills

The three that need a page of their own

Chat is the front door. These are what turn it into a workspace that produces things.

Everything included

On every plan, unless the feature line says otherwise.

A chat using several AI models in one conversationGPT-5ClaudeGemini+ 40 moreSummarise this in three pointsThree points, shortest first:Switch model mid-chat, keep the thread

Chat with any model

GPT, Claude, Gemini, Grok, DeepSeek and more — switch between them in the same conversation. If the one you picked is down, a similar model can step in automatically, at a similar price.

An image prompt and four generated picturesa fox in a neon city, poster artDALL·E · Flux · Stable Diffusion — four at a time

Generate images

Create artwork, logos, product shots and illustrations from text, or edit an image you upload.

A generated video clip beside an audio waveformPrompt or image → a short clipNatural speech

Video & audio

Turn a prompt or a starting image into short AI video, or generate natural speech and sound.

Search results with numbered citationswhat are they charging this month?Competitor A raised to $29/mo1Competitor B unchanged at $192Competitor C added a free tier3Every claim carries the source it came from

Web search & research

Answers grounded in real-time information with citations — and the AI decides when a search is actually needed rather than searching everything.

Uploading a file, and downloading generated Word, Excel and PowerPoint filesDrop a PDF, spreadsheet or imageand ask about what is insideDOCXWordXLSXExcelPPTXSlides↓ Generated for you, ready to download

Files & documents

Upload files for analysis, or have the AI generate Word, Excel and PowerPoint files to download.

A project workspace holding several chats and its membersACME rebrandown instructions · files · memory · modelLanding page copyLogo directionsTone of voiceMAJ3 in hereEvery chat inside starts already knowing all of it

Projects

Group chats into a workspace with its own instructions, notes, memory, files and default model — every chat inside it starts already knowing the context, and you can invite someone in.

Remembered facts, each with its own on-off switchWhat it remembers about youPrefers metric unitsEmails go out at 08:00Writes in British EnglishAnd a switch on any single chat, for the ones it should forget

Memory you control

The AI remembers what matters across chats — and a switch on every conversation turns that off for the ones you would rather it forgot.

A browser side panel open beside the page being reada-long-article.comAsk this pageGrab a tab, or type @remind without leaving the site

Chrome extension

A side panel that follows you around the web — ask about the page you are reading, grab a tab, or carry on a chat without leaving the site.

A share link with a Copy button and the people invited inaichatconsole.com/s/9fK2…CopySend a read-only link — or invite them in:MAJ+2 editors · 1 viewerThey see the conversation live as you work

Share & collaborate

Send someone a read-only link to a conversation, or invite them in to work on it with you.

A shield with Zero Data Retention switched onZero Data RetentionOnNever stored, never usedto train anyone’s model.Plus a memory switch on every single chat,for the conversations you would rather it forgot.

Private & secure

Zero Data Retention models available — conversations are never stored or used for training.

Prompt Library

Ready-made prompts with real examples of what they produce. Use one as it is, or save your own and reuse it with a /.

Automatic routing

Let the app pick the right model for each request — a cheap one for a short question, a strong one for hard reasoning — instead of choosing every time.

Ask for it in chat. It actually happens.

Most AI tools tell you what you could do. This one offers to do it — and waits for your yes before it does.

A reminder card and the notification it sends to a phone“Remind me tomorrow at 10 to call the bank”⏰Call the bankTomorrow 10:00 · notify + emailSnoozeDone⏰ ReminderCall the bank10:00Wherever you asked for it — phone, email or chat

Reminders that reach you

“Remind me tomorrow at 10 to call the bank.” It arrives as a notification, an email, or a message in your chat — whichever you asked for. Snooze it, or mark it done, from wherever it lands.

A week of calendar entries and the private feed linkThis week⋮MonTueDentistWedThuABC 11:00Fri⚠ Clash flagged before you agree to it🔗 Private link — adds to Google, Apple or Outlook

A calendar it keeps for you

“I have a meeting with ABC Company on Thursday at 11.” It goes in, clashes get flagged, and the private link adds it to Google, Apple or Outlook Calendar. Ask to be briefed and half an hour beforehand you get a prep note.

A scheduled task with its budget and its last three runs“Every Monday, check these pages and email me”Pricing watchMon 09:00budget 8 cr16 Sepdelivered📄9 Sepdelivered📄2 Sepno changesRuns on the server — your tab can be closed

Work that runs without you

“Every Monday, check these three pricing pages and email me what changed.” See Tasks for how the planning, the budget and the delivery work.

A drafted email waiting for approval before it is sentTo:client@example.comSubject:Your monthly summarySend itEditfrom your own domainNothing that emails another person is ever automatic

Email, written and sent

“Email me the summary at 9 tomorrow.” Or send it to a client from your contacts, from your own domain, with an unsubscribe link and the compliance footer already handled.

A watched page staying silent until the day it changesexample.com/changelog👁Monno changeTueno changeWed3 new entries🔔It only speaks up when something actually moved

Pages watched for changes

“Tell me when their changelog changes.” It checks on a schedule and only speaks up when something actually moved.

An action card showing what, when and what it costs, with Confirm, Edit and Cancel✉Email the summary to a clientWhatMonthly summary, from your domainWhenTomorrow at 09:00Costabout 2 creditsConfirmEditCancelNothing is saved, sent or spent until you choose

You are always asked first

Every action appears as a card: what it is, when it happens, what it costs. Confirm, edit or cancel. Something a web page asked for can never act on its own, and nothing that emails another person is ever automatic.

💡

Reminders and calendar entries are free. Only work that runs the AI — research, writing, scheduled tasks — uses credits, and the card tells you the budget before you agree to it.

Tuned to how you work

Almost everything is a setting. Change it once and it follows you across every device you sign in on.

Default model, creativity and reply length for new chatsEvery new chat starts withModelGPT-5 mini›Creativity0.7›Reply lengthup to 8,000›✓ Or let the app pick the right model per requestOverride any of it inside a single conversation

Your default model

Pick the model, temperature and response length every new chat starts with — then override any of it per conversation, or let the app route each request to the right model by itself.

A slash-command menu of saved prompts/rev/reviewCheck this against our checklist/replyDraft a reply in our tone/brief {client}Start a project brief{client} asks you to fill it in · chain several steps

Shortcuts

Save prompts you reuse and drop them in with a /. Give one variables to fill in, have it replace or extend your message, or chain several steps together.

Three AI personas, each for a different job🧐Code reviewerblunt, spots riskIn use🎓Patient tutorexplains slowly✒️Copywriterknows our voiceEach with its own instructions and model

Personas

Build AI personalities for different jobs — a blunt code reviewer, a patient tutor, a copywriter who knows your brand voice — each with its own instructions and model.

Memory recall setting with a per-conversation off switchMemoryon by defaultEverythingOnly relevantMinimalThis conversation: memory offA switch on every chat — that one neither readsyour saved facts nor learns anything new.

What it remembers

Decide what the AI keeps about you, whether new chats start with memory on, and how much of it gets recalled. Every conversation has its own switch.

Light, dark and OLED themes with accent colours and type sizesLightDarkOLEDaccent colourAaAaAafour faces, adjustable sizeComfortable or compact · reduced motion

Look and feel

Light, dark or OLED black, accent colours, four font families, adjustable text size, comfortable or compact spacing, and reduced motion if you prefer less movement.

Provider API keys, two in use and one not setYour API keysOpenAIsk-•••• 4f2ain useAnthropicsk-ant-•••• 91cin useGemininot setaddRuns on your own account instead of your creditsStored encrypted · never shown again after saving

Your own API keys

Bring keys from any supported provider and run on your own account instead of your credit balance — on the plans where BYOK is enabled.

Privacy switches and a monthly credit budgetPrivacyEncrypt stored chatsZDR models onlySave historyMonthly budget2,000 credits — spend cannot run away

Privacy controls

Encrypt stored chats, restrict yourself to Zero-Data-Retention models only, turn history saving off, or set a monthly credit budget so spend cannot run away.

The message box with composer preference switchesAsk anything…SendEnter sendsStream repliesTimestampsCitation linksSidebar, scroll behaviour and spacing too

Composer & layout

Enter to send or a newline, streaming replies or all at once, timestamps and citation links on or off, sidebar collapsed, and how the view follows a reply as it writes.

Illustrations of the real interface, drawn from its own markup and kept in step with it in code.

Your conversations stay yours

🛡️

Never used for training

Not by us, and not by the providers on the Zero-Data-Retention models — where nothing is stored on their side at all. The model picker marks which those are.

🔐

Encrypted storage

Turn encryption on for stored chats, or turn history off entirely and keep nothing. Deleting a chat deletes its attachments with it.

📉

Spend you can cap

Set a monthly credit budget, see what every request cost, and check the estimate before a module or task runs. See the billing guide.

Common questions

Do I need a separate subscription for each AI model?

No. One plan covers every model on the platform, and you can switch between them inside a single conversation. Credits are spent per request, so cheaper models stretch further.

What can it do besides chat?

Generate images, video and audio; search the web with citations; read files you upload; produce Word, Excel, PowerPoint and PDF files; run ready-made modules; work on background tasks on a schedule; and keep projects with their own instructions, files and memory.

Is my data used for training?

No. Conversations are never used for training, stored chats can be encrypted, and you can restrict your account to Zero-Data-Retention models only, where the provider stores nothing at all.

Where do I read about the rest?

The user guide covers every feature in depth, the billing guide covers plans, credits and invoices, and the fair usage policy covers the limits.

Try it before you decide

No credit card required. Credits never expire, and you can cancel any time.

🧩

Modules

Fill in a short form, get finished work — hundreds of ready-made tools.

Read more →
✅

Tasks

Give the AI a goal. It plans, works in the background and delivers.

Read more →
📘

Skills

Teach it how you want a job done once. It applies that by itself.

Read more →