The Content Agent: Questions Into Content | DigiKat

Your customers are already telling you what to write. They ask it in support tickets, in website chat, on sales calls and in the notes your team leaves on deals. The same question shows up dozens of times and nobody writes the article, because nobody has time to count. The Content agent counts.

This post continues our series on the management agent team we build on HubSpot. So far we've covered the Sales agent, the Support agent and the Marketing agent. The Content agent sits behind all three. Its job is to notice what people keep asking, work out what's missing and draft the content to fill the gap.

Where the questions come from

The Content agent doesn't start with a keyword tool or a content calendar. It starts with the questions already logged in HubSpot:

• Tickets. Subjects, descriptions and the replies your team wrote. A question that keeps coming back to support is a question your website and knowledge base aren't answering.
• Chats. Website chat and portal chat transcripts, including the ones the Support agent handed off because it had no approved answer.
• Deals. Notes, logged emails and closed lost reasons. If prospects in one industry keep asking the same thing before they buy, that's a sales asset waiting to be written.
• Sales calls. Call recordings and summaries logged against contacts and deals. What objections come up? What did the prospect not understand?

Because all of this sits in one CRM and is associated to contacts, companies, deals and tickets, the agent can group questions by who asked them. That matters. "How long does setup take?" from a prospect at the proposal stage is a different piece of content from the same question asked by a new customer in week one.

Finding the gaps

Collecting questions is the easy part. The useful part is comparing them to what you've already got. Each week, the Content agent:

1. Groups similar questions into themes, so dozens of slightly different wordings become one topic.
2. Checks the knowledge base for an approved article, email or asset that answers it.
3. Flags the gap when there's nothing, or when what exists is out of date or keeps getting rewritten by the team before they send it.
4. Ranks the gaps by how often the question comes up, which stage it shows up at and how much time it costs the team to answer by hand.

The output is a short list, not a 200-row spreadsheet. For example, an illustrative weekly report might say: "Twelve tickets this fortnight asked how to update payment details in the portal. There's no article. Three prospects at the proposal stage asked about data residency. The current answer is in one person's sent folder."

That's the kind of list a business owner can act on in five minutes.

Drafting the content

Once a gap is approved as worth filling, the Content agent drafts the piece. It drafts from what your team has already said, not from the internet:

• The best existing replies to that question from tickets and emails.
• Any approved material already in the knowledge base on related topics.
• Your brand voice and style rules.

Depending on the gap, the draft might be:

• A knowledge base article for the help desk and portal.
• A sales email or one-pager for a particular deal stage.
• A nurture email for the Marketing agent's sequences.
• A blog post outline when a question is common enough that people are probably searching for it too.

Each draft is tagged with where it's meant to be used: which deal stage, which ticket type, which audience. That tagging is what lets the other agents find it later.

What it won't write

The Content agent drafts admin and how-to content. It does not write clinical, legal or financial advice. In a health business, a recurring question about medication or treatment becomes a flag for the clinical team to answer, not a draft. Claims about results, pricing promises and anything contractual get flagged for a person to write or sign off line by line.

From draft to knowledge base, with approval

Nothing the Content agent writes goes live on its own. The flow looks like this:

1. The agent drafts the piece and attaches the evidence: the questions it's answering and where they came from.
2. It passes the draft to the knowledge base for review, with a task for the person who owns that topic.
3. The owner edits and approves, or rejects with a note.
4. Approved content goes into the knowledge base, tagged by stage, ticket type and audience.
5. From then on, the Support, Sales and Marketing agents can use it.

That last step is the payoff. Once an article is approved, the Support agent can answer that question in chat instead of handing it off. The Sales agent can send the one-pager at the right deal stage. The loop closes, and the same question stops costing your team time.

This is what we mean when we say agents work as a team. We described the operating model in more detail in from workflow to agent team.

How we set it up

Same method as every agent: Context, Action, Coordination, Result.

• Context: check that tickets, chats, deals and calls are actually logged against the right records. If sales calls live in someone's head and support happens in a personal inbox, there's nothing for the agent to read. Pick one job, usually "find the top five support gaps", and measure how long those questions take to answer today.
• Action: HubSpot's own content and AI tools first. Custom work only where they stop, such as grouping questions across tickets, deals and calls together. We test it on a past quarter of real questions before it drafts anything new.
• Coordination: people approve, agents act. Every piece of content is approved by a person before it reaches the knowledge base, and anything with legal, clinical or financial weight stays with a qualified person.
• Result: a monthly scorecard: gaps found, drafts approved, questions that dropped once content went live, what it cost and the net return.

If the same questions keep landing on your team's desk, the content you need is probably already written, scattered across tickets and sent emails. Build my solution and we'll show you what a Content agent would find in your HubSpot, or start with an AI Readiness Audit. You can read more about how we combine HubSpot's agents with our own.