HubSpot Content Remix: Scale Content Only With a Review Process

What is HubSpot Content Remix — and what isn't it?

Content Remix is an AI feature inside HubSpot Content Hub that turns up to six source materials into up to six new content assets per run — a blog post becoming a podcast episode, three social posts and a landing page at the same time. It's not an autopilot: every result lands as a draft, never as published content.

That's the real distinction from a generic AI text generator: Content Remix doesn't start from an empty prompt, it starts from real material that already exists — blog posts, landing pages, website pages, pasted text, uploaded documents (audio, video, images) or external URLs. The AI compresses and translates between formats; it doesn't invent the starting point. Technically, it runs on Breeze, HubSpot's own AI layer that also powers other Content Hub and Marketing Hub tools — Content Remix is one Breeze application among several, not an isolated feature.

A concrete comparison makes the difference clear: asking a general-purpose chatbot to "write a social post about HubSpot migrations" produces something plausible but generic, disconnected from anything the company has actually published. Running that same social post through Content Remix on top of an existing, approved blog post produces a condensation of sentences already vetted in-house. The difference isn't the AI itself — it's the reviewed material it's allowed to start from.

Availability isn't uniform across the product:

Teams running Content Hub alone can remix content, but can't carry a result all the way to a usable social post without also having Marketing Hub — the same product split already covered in smart content rules.

Which sources and target formats does Content Remix actually support?

Sources include existing HubSpot content, pasted or typed text, uploaded files, and — in beta — Google Drive files and external URLs, combinable up to six sources per run. Target formats include blog posts, social posts, images, SMS, ads, video/audio clips, podcast episodes, landing pages, marketing emails, and freeform content.

The multi-source combination is the part most short overviews of this feature skip: a single run can draw on a whitepaper, a webinar recording and an existing blog post all at once. That doesn't automatically make the output better — it just gives the AI more raw material to choose from.

For a B2B SaaS team, three combinations come up constantly in practice: a product update blog post that becomes three social posts and a short announcement email at the same time; a webinar recording that becomes a podcast episode and a blog post for readers who'd rather not watch the video; and an existing landing page that becomes ad copy for the same funnel without restating the page's core claim differently.

The beta and external source types deserve their own caution: Google Drive files are explicitly still in beta according to HubSpot, and external URLs pull in content that never went through your own editorial process — a competitor's press release, a news article, someone else's blog post. Both source types work better for research and inspiration than for content meant to go live without further review.

What needs to be settled before the first content remix run?

A super admin has to have turned on the "Give users access to generative AI tools and features" toggle, or the feature won't start at all. The person running it also needs the Content Remix permission itself, plus permission for every individual target tool — an SMS won't get created just because someone clicked generate, if the SMS permission is missing.

• Content Hub edition confirmed, plus Marketing Hub if social posts are planned (see table above)
• Super admin toggle for generative AI features switched on — usually the first blocker teams hit without realizing it
• Content Remix and target-tool permissions checked individually for the person running it, not just their general editor role
• Source content actually editorially approved — remixing an outdated blog post just spreads the same outdated claims across more channels
• An approval process for the generated drafts defined before the first run, not after

One question belongs in this phase too, even though HubSpot's own documentation doesn't take a position on it: who on the team is allowed to submit uploaded documents — an internal sales deck, a customer recording — as a remix source in the first place? Not everything shared internally was ever meant to become six new, independently published assets just because it happens to be uploadable.

The same caution applies to customer material: a webinar recording with a customer quote, or a case study draft with a number that hasn't been finalized yet, make tempting remix sources precisely because they're already narratively complete — and risky for the same reason, if approval only ever covered the original publication, not six new standalone assets built from it.

How does a content remix run actually work, step by step?

The path runs through Content → Remix in the HubSpot menu: pick or upload a source, optionally add more sources, select target formats via checkboxes, configure attributes like template, length, style and objective, generate, then edit inline or push each result into its target app.

1. Navigate to Content → Remix in the HubSpot menu (under "More" if it's not visible directly)
2. Choose a starting point: select existing content, paste text, or upload a file
3. Optionally add further sources via "Add content," up to six total
4. Select target formats via checkbox, up to six assets per run
5. Configure attributes per target format — template, length, style, objective. For a social post that means platform and tone, for an email it means subject-line focus and call to action, for a podcast clip it means target length in minutes
6. Click "Generate" and wait for results
7. Edit each result inline, or use "Save and edit in app" to move it into the relevant target tool

Step 3 is worth pausing on: because up to six sources can be combined, it's worth deciding deliberately before starting whether more raw material genuinely improves the result — or just gives the model more surface area to split the difference between old and new claims that don't actually agree.

Walked through in practice: a single, current blog post about a new feature needs to become a LinkedIn post, a short ad and an announcement email. Three target formats, one source, one attribute set per format — no combination risk, because nothing can contradict a claim that's only made once. That simple case reliably produces the most usable output with the least rework, which makes it the right first test run for any team still learning the feature, before more complex multi-source combinations even come up.

What do you do when the output isn't usable?

The most common causes are missing permissions, source content that's too thin or internally contradictory, and target formats that technically generate but can't be used further without the right Hub license attached. None of these are bugs — every one of them can be fixed before the next attempt.

• A target format can't be selected, or generates nothing usable: permission for that specific target tool is missing, not just for Content Remix itself
• The result reads generic or contradictory: multiple sources with different currency or tone were combined — trim the source selection, don't just tweak the prompt attributes
• A social post output can't be carried further: Marketing Hub Professional or Enterprise is missing on top of Content Hub
• The feature doesn't show up in the menu at all: the super admin toggle for generative AI features was never switched on
• The SMS target format stays greyed out: the separate SMS add-on is missing at the account level — a permission alone isn't enough, the add-on itself has to be purchased
• The result repeats outdated numbers or claims: the source itself was outdated — Content Remix reproduces what the source material says, it doesn't check it against current reality
• Multiple sources produce an indecisive middle-ground result: the model averaged between contradictory phrasing across the individual sources instead of favoring one — keep only the most current source and regenerate

How do you undo a remix without putting the original content at risk?

Content Remix never touches the source material — the blog post a remix was built from stays exactly as it was. Every generated result is a new, independent draft; discarding one simply means deleting that draft, not "undoing" anything against a shared original.

That reduces the risk structurally: a bad remix run costs, at worst, some unused drafts — never the source content itself. The real caution isn't about rollback, then, it's about what happens before it: once a draft has been pulled into its target app and developed further there, treat that version through the target app's normal version history, the same as any other content draft — not through Content Remix.

A second, easily missed point: a blog post already live, later used as a remix source, stays published and unchanged while new drafts get generated from it in the background. There's no moment where the live page goes offline or becomes inconsistent for the duration of a remix run — all of the risk sits with the new drafts, none of it with the existing page.

How do you confirm a remix result is actually correct before publishing?

Every generated asset deserves the same scrutiny as anything written from scratch: check facts and numbers against the source, verify brand voice, and actually read it in its target format — an SMS draft that looks fine in the editor can read as cut off on an actual phone screen. None of that can be delegated back to the AI.

Numbers and customer quotes deserve particular attention: when a source contains a metric, a customer quote or a specific experience, a remix run can subtly shift or compress it while translating between formats — "over 50 projects" can quietly become "hundreds of projects" because that reads punchier in an SMS. It's the exact same risk your own editorial process runs without a proof register, just harder to spot here, because the original sentence really was accurate.

Organic traffic adds a second check: Google has stopped grading content by whether a human or an AI wrote it, and grades it instead by whether it actually serves a real search intent — a pure reformulation with no added substance won't reliably pass that test. A remix result that repeats a blog post's argument in a different order, without a new angle, a new example or a new layer of context, competes with its own source instead of complementing it — and can end up undercutting the discoverability of the original page it was built from.

Brand voice deserves a separate check from the fact check: a blog post in running prose and an SMS message follow completely different registers, even when both come from the same source. A remix result that reads convincingly in blog form can suddenly land like a hard sales pitch once compressed into SMS length — exactly the kind of phrasing that would never survive in long-form copy, but slips through unnoticed in an automatically generated short format.

Which B2B content combinations are actually worth remixing — and where is AI not a shortcut?

Remix earns its keep most on pure format translation with no new facts involved — an existing blog post becoming social posts, a webinar transcript becoming email copy blocks. It earns its keep least when the target format needs new, specific claims that were never in the source material to begin with.

The pattern in that table: the closer a target format stays to a pure format swap, the more reliable the result. The more a target format expects genuinely new substance — a fresh number, an updated customer quote, a new framing — the more human rework stays mandatory, regardless of how convincing the model looks in any single case.

The logical takeaway for a B2B team running a HubSpot implementation: Content Remix doesn't replace a content strategy, it speeds up executing one that already exists. Teams that turn the feature on before deciding which claims and numbers are actually approved aren't scaling their editorial output — they're scaling the risk that one outdated number shows up in six channels simultaneously instead of one.

• Start with one source and one target format, not the maximum combination of six and six
• Only allow editorially approved content as a source, never internal rough drafts or unreviewed recordings
• Check numbers and customer quotes in every result against the source individually, regardless of target format
• Check brand voice separately for each target format, not just once for the source
• Define the approval process before the first run, so drafts don't drift unreviewed into target apps

A practical starting point: begin with a single, clean source and a single target format, check the result fully against the source, and only then attempt combining multiple sources and multiple target formats. That holds whether the underlying HubSpot implementation has run Content Hub for years or is only just being evaluated — the governance questions are the same before the first run, not just after the first visible mistake.

Book a free Launchpad and we'll map which content combinations actually save time against your existing material, and where an editorial step stays mandatory regardless of the tooling.