Contents
A week ago Agent learned to take prompts off your plate. Today it grows to version 2.0 — its biggest update yet: employees, photos in the chat, sound, a livelier dialogue, and noticeably more reliability. We keep improving it through real conversations, and your usage is the best way to make it sharper.
The headline is employees. One generalist head can't hold marketplace cards, product photography, copy, video, sound, strategy, and serial storytelling in deep focus all at once. Now you can hire one of eight roles: Card Designer, Product Photographer, Copywriter, Videomaker, Sound Designer, Prompt Engineer, Meaning Maker, and Creative Strategist. Each works in its own chat and knows its craft — under the same rules: a card with the exact price, launched only after you press Generate.
What else changed
Agent itself grew too. Attach photos right in the chat — and brush-select the exact area to rework. It makes music and voiceover through the same flow as images and video. It streams replies word by word, shows step by step what it's doing right now, and offers quick-reply buttons instead of retyping.
And the third layer is reliability and honest economics. No message is ever left hanging: even on a failure you get a plain explanation, not silence. If your dialogue model is overloaded, the turn completes on a fallback — with a visible badge. Reply spending and your daily message limit sit right in the chat. And the firm rule: platform failures are never on your bill — tokens don't burn.
Since the release we have added Seedream 5.0 Pro for controlled image work, Grok 4.5 for bold conversational ideas, and Fable 5 for the hardest agentic tasks. They are now joined by the GPT-5.6 family — not one model, but three distinct options: Sol, Terra, and Luna. Below is a real Seedream result, with no manual retouching or text replacement.
Why 2.0 — and how we keep improving it
We bumped the version honestly: employees, a multimodal chat, sound, and the new reliability make this a different Agent from the one that shipped a week ago. Quality work does not stop at release: every week we review real conversations where Agent misread a brief, where a prompt came out weaker than it could, or where an answer ran too long — and we fix it.
So the most useful thing you can do is simply use it. And if Agent answers oddly or misses the point — tell it right in the chat, or write to support. Every such case turns into a concrete fix.
A specialist keeps focus — you skip the context lecture
When one head answers for everything, every conversation starts with setup: "I need a marketplace card, vertical, with infographics, bigger text…". An employee has already been through that setup. The Card Designer thinks in slides and offers from the first message, the Videomaker in shot lists and first-second hooks, the Copywriter in keywords and buyer benefits. You just state the task.
The roster follows real demand: we looked at what marketplace sellers and small businesses order from freelancers most, and built employees for exactly those jobs. Each has its own professional techniques and knowledge base — including marketplace requirements checked against the official Wildberries and Ozon help pages.
The two latest roles widen that scope. Meaning Maker builds fictional character canon, conflict, season arcs, hooks, and vertical scripts. Creative Strategist turns the goal, audience, and available project materials into a brief, positioning, campaign directions, and testable hypotheses — separating supplied facts from assumptions.
An employee is not a separate plan or a subscription to a person. It's the same Agent working in its own chat within its specialty: the same tokens, the same limits, the same price card before every generation.
The gallery, a dedicated chat, and slash commands
The project chat list gained a "Hire an employee" button. It opens a gallery of role cards describing what each one does. Pick one — and the employee gets its own chat with its own starter suggestions; in the chat list it carries a role badge so it never blends in with regular conversations.
The fast path is a slash command right from the main chat: the specialist gets hired and your brief lands in its new chat immediately — together with any attached files. The dialogue model for each role comes with a sensible default, and you can change it in the same picker as Agent's. Hiring any number of employees costs nothing by itself — you pay, as usual, only for turns and approved generations.
Photos right in the chat — and sound through the same flow
In the previous announcement "photos in chat" honestly sat under "What's next" — that next has now arrived. Attach an image right in the conversation: as a reference ("make it in this style"), as a source ("here's my product — shoot it on white"), or for editing. For precise fixes there's the brush: open the image, select an area, and ask to redo only that part. Materials already uploaded to the project are available through the paperclip too.
Sound arrived through the same flow as images and video: ask for a jingle, a background track for your clip, or a voiceover — Agent picks the model, shows the price card, and launches after you press Generate. Voiceover reads your text verbatim; for music you set the mood, tempo, and duration to match your clip.
Streaming, visible work steps, and quick-reply buttons
Replies no longer land as one block after a pause — they stream word by word, like a live conversation. While Agent works you can see what it's doing right now: "Searching materials", "Picking a model", "Composing the prompt" — and who is working, when a specialist runs the turn.
When Agent offers options — "vertical or square?", "which of the three concepts?" — buttons appear under the message. One tap instead of retyping. And new chats now take their name from your first message, so the chat list stays readable.
Never leaves you hanging — and never loses your generations
We went through every place Agent could go silent or stumble — and closed them. No message is ever left unanswered: whatever happens inside, the chat gets a plain explanation instead of an endless spinner. If your dialogue model is overloaded or failing, the turn automatically completes on a fallback — with a visible badge showing which model answered.
Agent also sees its own work better now: ask "what's up with my video?" and it checks the status of launched generations and answers to the point. If a generation fails, it figures out the cause, rewrites the prompt, and proposes a new run as a card — not a blind retry on your dime.
Spending in plain sight — and never paying for failures
The chat now has Usage & limits: how much replies in this chat have cost and how many messages you've used today out of your plan's limit. With a paid subscription it also shows Bonus, a daily allowance for Agent replies that is spent before your balance. Each reply shows its model and the amount charged: hover over it on a computer or tap it on a phone. No trips to settings to know where you stand.
The proposal card got more patient: its lifetime is longer, and if it expired while you were discussing details, one click brings it back — no new paid turn. Type "Approve" as text and Agent highlights the card it refers to, but the launch still happens by button only.
And two honesty rules. First: if you keep approving cheap generations without edits, Agent offers — once — to auto-run the small stuff; your call, and it can be turned off. Second, the firm one: a platform-side failure is not your problem. Such turns aren't charged, and failed paid generations refund their tokens — with the status right in the error card.
Seedream 5.0 Pro for visuals, plus Grok 4.5, Fable 5, and GPT-5.6 for dialogue
Agent can now propose Seedream 5.0 Pro for photorealistic commercial hero images, product photography, or edits to materials, colors, and individual objects guided by annotated references. It is especially strong on natural lighting, skin, and material texture, with 1K/2K output and up to ten references.
This is not a new undisputed image-model leader. Early hands-on tests praise product photography, commercial posters, and local edits, while complex text, infographics, and UI still show errors. Nano Banana Pro remains the choice for 4K brand scenes, and ChatGPT Images 2.0 is the most proven general option for complex instructions and layouts. Seedream 5.0 Pro stands out for photorealistic commercial imagery and cheaper iterations.
Grok 4.5 was added to the Agent model picker for bold ideas, lively writing, and an unconventional point of view, along with Fable 5 as the highest tier for the hardest briefs and long agentic work. The GPT-5.6 family is now available there too. These are three separate models with different capability and price levels, not one generic GPT-5.6 option. The price of a reply depends on the model you pick.
Where Agent 2.0 grows from here
The roster and the abilities keep growing. On the roadmap:
Employees working together
Specialists will learn to hand tasks to each other: the Copywriter passes a script to the Videomaker, who calls the Sound Designer for the track.
A growing roster
New roles by the same principle: real demand first, then the employee.
Skill management
Turn knowledge packs on and off by hand, for a specific task.
Frequently asked questions about Agent 2.0
What if Agent misses the brief at first?
Clarify the task right in the chat — Agent can correct its course. Your money stays protected: every generation shows its price before launch, charges happen only after you press Generate, platform failures aren't billed, and failed paid generations refund their tokens. Every oddity you report helps make the next conversation sharper.
How is an employee different from the regular Agent?
Focus. An employee is a specialist with its own chat, its role's professional techniques, and a knowledge base to match: the Card Designer thinks in slides and marketplace requirements, the Videomaker in shot lists. The general Agent stays right where it was — a generalist that still does a bit of everything.
How much does hiring cost?
Hiring itself is free — there's no separate plan for employees. A chat with an employee costs the same as a chat with Agent: turns are priced by the dialogue model you pick, generations by the price card, and only after you press Generate.
Can an employee launch anything without my approval?
No. One rule applies to everyone: any generation is a proposal card with the model and price, launched only after you press Generate. Some roles deliver text rather than launching media: the Copywriter delivers copy, the Prompt Engineer a prompt, the Meaning Maker a legend or script, and the Creative Strategist a brief and concepts.
Do employees see my project memory?
Yes, memory is shared. The style, brand, and rules from "Project context" apply to Agent and every employee — no need to repeat yourself in each chat.
Will custom employees appear?
The current first-party roster has eight roles. We're considering custom employees but won't promise dates: first we want to prove that the current roles deliver useful results consistently.
Agent answered oddly or missed the point — what do I do?
Tell it right in the chat — it often corrects itself. If that doesn't help, write to support: we review such cases by hand and turn them into concrete quality fixes.
What's new, straight to your inbox
