AI for content production, described as the ten people it replaces

What an AI content system actually is

An AI content system is a group of AI workers who between them cover the whole job of producing content, from deciding what is worth saying to storing what worked. Writing is only one part of it, and it is not usually the part that is broken. What breaks is knowing what to make, making enough of it, keeping it sounding like you, and remembering what already did well.

Most businesses that buy an AI writing tool find the same thing. The words arrive quickly and they sound like everybody else's words, because the tool was given a topic and nothing else. A system fixes that by giving the writer a brief, a voice guide and a reason the piece exists, all produced by other workers before the writing starts.

The other half of content is visual, and it is the half that stalls. Adverts need images. Products need photographs. Short videos need a script and a hook before anyone films anything. Those are separate jobs and they get separate workers.

The team, and what each one does

Ten jobs, described the way you would describe a member of staff.

• The researcher. Watches what competitors publish, what is being talked about in your category and which questions keep coming up, then reports the gaps worth filling.
• The planner. Keeps the calendar. What goes out, on which channel, in what format, on which day, and why that piece rather than another.
• The voice keeper. Holds your tone and your rules. Words you use, words you never use, how formal you are, whether you use Urdu and English in the same sentence the way your customers do.
• The writer. Turns a brief into a draft. Posts, captions, emails, guides and product descriptions, all written against the voice guide rather than from a blank page.
• The static ad maker. Produces the flat image adverts in volume, so a campaign can test many angles at once instead of betting the month on one picture.
• The video concept writer. Writes the first three seconds, the scenes and the script, which is the part that decides whether a short video works at all.
• The creator brief writer. Writes the briefs you hand to real people filming demos, reviews and testimonials, so they know what to say and what not to promise.
• The product scene maker. Produces product images in settings: on a table, in a hand, in a room, in a catalogue grid, without booking a studio for every variant.
• The repurposer. Takes one long thing and makes many short ones. A recorded talk becomes clips, a guide becomes a week of posts, a customer question becomes an email.
• The librarian. Records what worked. Which opening line, which format, which angle, so the same lesson is not learned twice and a new worker starts with your history rather than a blank slate.

A Tuesday inside an AI content team

8:30am. The researcher finishes overnight and reports two things: a competitor has started posting about warranty terms, and the same question about sizing has appeared four times this week in your comments. Neither is urgent. Both go to the planner.

10:00am. The planner rearranges next week. The sizing question becomes a short video and a post, because a question asked four times in public is being asked forty times in private. It moves a less useful piece back a fortnight rather than cramming.

11:30am. The writer produces three drafts against the briefs. Each one goes through the voice keeper first, which is a checking step rather than a person. Two pass. One comes back because it used a phrase your business does not use about pricing.

1:00pm. The static ad maker builds six versions of one advert. Same offer, six different opening lines. They are not published. They go to the marketing team's tester, which will run them properly rather than all at once.

2:15pm. The repurposer takes last month's recorded customer webinar and cuts it into four clips, one email and a written summary. This is the highest return job on the whole roster and almost nobody does it by hand, because it is tedious.

3:00pm. You approve. Six items, about ten minutes. You rewrite one headline yourself because you know something the system does not, and that rewrite goes into the voice guide so it is not a correction you have to make again.

4:30pm. The librarian closes out last week. One post did unusually well. It records the format and the angle, and tags it so the planner reaches for that shape again rather than guessing.

The work that happened today is the work that normally does not happen. Not the writing, which somebody always finds time for eventually, but the planning, the checking, the repurposing and the recording of what worked.

What stays human in content

The story stays human. Why you started, what you refuse to do, the thing that went wrong in 2019 and what you changed afterwards. Those are the parts people remember, and no system can produce them because it was not there.

Anything involving a real person stays human. A photograph of your staff, a customer's words, a name, a face. Consent is a conversation, not a setting, and using someone's likeness without asking is a problem that no amount of good content fixes.

Humour and cultural reference stay human. Jokes travel badly across languages and across cities, and the failure mode is not being unfunny, it is being offensive by accident. If a piece depends on a joke, a person signs it.

Final approval stays human. Everything published carries your name. The step takes minutes and it is the step that keeps the whole system trustworthy, because it means nothing goes out that you would not have said.

How content connects to other departments

Content is a supplier to two other departments and a customer of a third. It receives briefs, it produces material, and it gets told whether the material worked. When those three links are missing, content becomes a stream of posts nobody can judge.

The link that surprises people is with support and sales. The questions customers actually ask, in the words they actually use, are better content briefs than anything a planning session produces, and they are free.

• The marketing team says which angles need creative and reports which ones sold, so the librarian records something real rather than what got likes.
• The search team briefs the pages worth writing and reports back on what got found.
• The support team hands over the questions it answers all day, which become the clearest content you will publish all year.

What a content system costs and what changes the number

An audit, and then one of two things rather than four, because content is recurring work permanently. Either the system is built and your team trained to run it, a one time build fee with a small retainer afterwards, or Wobble runs it and you rent the output as an agreed monthly list of pieces. Static posts and carousels are made with AI either way.

Short video is the split, since a reel is made with AI, filmed with a crew, or both, depending on the product rather than on a policy, and where a shoot is involved that part is a retainer. The four ways to own this explains why these two sit outside the grid.

The audit matters more here than people expect, because the first month is mostly spent capturing things that currently exist only in your head: how you talk, what you sell, what you will not claim, which photographs you are allowed to use.

Video and photography move the figure most. Written content and flat adverts are quick. Anything filmed involves real people, real time and real scheduling, and no system removes that. Be clear at the start about which of the ten workers you are actually buying.

• How many channels are in scope, and how often each one publishes.
• How many languages, and whether the same piece has to work in both.
• Whether video and product photography are included or handled elsewhere.
• Whether a brand voice and a set of rules already exist in writing.
• How many people have to approve a piece before it goes out.
• Whether Wobble operates it monthly or hands it to your own team once the handover is done.

The case for publishing less, not more

If nobody engages with what you publish now, publishing four times as much makes the silence louder and costs more. The problem is usually the offer or the audience rather than the volume, and both are cheaper to test with five good pieces than with a hundred average ones.

If your voice is not defined anywhere, the output will read like everybody else's. That is not a failure of the writer, it is a missing input. A business that cannot say in a paragraph how it wants to sound is not ready for volume, and the fix is a conversation rather than a purchase.

If your product information is wrong or missing, content spreads the confusion faster. Sizes, materials, delivery areas, what is included in a price. Get those correct in one place first, because every worker on the roster reads from it.

If nobody will approve, nothing ships. Content is the department where an approvals queue backs up fastest, because there is more of it than anything else. One named person, ten minutes a day, or do not start.

Who runs this, what it costs to stand up, and where it stops

Ali owns marketing at Wobble and takes the first pass on a content system, because the decision that matters is not which worker generates the draft. It is which subjects your company is entitled to have an opinion about, and that is a positioning judgement rather than a production one.

What it costs is set by two things, and neither is volume. The first is whether a voice already exists in written form somewhere a system can learn from, or has to be pulled out of a founder's head across several sessions. The second is how checkable your subject matter is. A business writing about its own process is cheap to support. A business writing about regulated advice needs a reviewer on every piece, and the reviewer is the cost.

Nothing publishes itself. Every piece waits for a person to approve it, and that person is named rather than a queue clearing on a timer. Where a draft makes a claim about a client, a figure or a competitor, it is held until somebody can point at the source. A wrong published sentence is expensive in a way a late one is not.

Timing is short at the start and long at the end. The audit takes the first week, and the first workers, usually research and repurposing, are running inside a fortnight. What takes months is the part nobody can compress: enough published work for a pattern to be visible. Wobble is answerable for the system and stays month to month, so the month it stops earning is the month you stop paying for it.

Common questions

What is an AI content system?

An AI content system is a team of AI workers covering research, planning, writing, ad creative, product images, repurposing and record keeping, all working from one guide to how your business sounds. The point is that content gets planned and judged, not only written.

Will AI written content sound generic?

It does when the only input is a topic. It stops sounding generic when it is given your voice guide, a brief with a reason the piece exists, and your own facts to work from. The generic feeling comes from missing inputs rather than from the writing itself.

Can AI make images and video for adverts?

It can produce flat image adverts, product scenes and the scripts and hooks for short video. Actual filming with real people is still filming. The sensible split is that AI covers volume and variations, and a camera covers anything that has to feel real.

How much content should a business publish?

Enough that each piece is worth reading, which for most small businesses is less than they are told. A weekly piece that answers a real customer question does more than daily posts nobody finishes, and volume only helps once you know which shapes work.

Who owns the content and the brand voice guide?

You do. The voice guide, the briefs, the library of what worked and the published material sit in your own accounts. On every engagement model except renting the output, if the work ends you keep all of it and Wobble removes its access.

What is the first content worker to install?

The repurposer, in most businesses. You already have material that nobody has cut up: recorded calls, a long guide, an old video. It produces something useful in week one without needing a new idea, which makes it the easiest way to see whether this suits you.