New: an AI agent for scripts and video
Create it once.
Spread it everywhere.
One post, twelve networks. Your AI agent drafts the captions, writes the script and produces the video.
Free to start. AI comes with every plan.

Publishes to
- Threads
- X
- Bluesky
- Mastodon
- TikTok
- YouTube
- Tumblr
- Google Business
Composer
One editor. Every network at once.
Write the caption, attach the media, pick the accounts. Ask the assistant when you want a first draft or a second take, and override just the network that needs something different.
Per-network versions
A different caption for LinkedIn, a title for YouTube, a link for X. Or let the assistant adapt the caption to one account's length limit and suggest its first comment.
An assistant that suggests, never decides
Write from a brief, then improve, shorten, lengthen, fix the spelling, change the tone, translate into seven languages or suggest hashtags. It offers a few versions and you add the one you want.
Checked before it goes out
Every network publishes its own rules: caption length, how many files, which formats. A post that breaks one is caught in the editor, not by the provider.
Nothing is ever lost
Drafts save themselves while you type, one save at a time, and again the moment you leave the page.


Scripts and video
From what you know to a narrated video.
Give SPREVA your documents, links and notes. Write a script that keeps track of its facts as you go, then produce it: narration, scenes and captions, ready to edit in Studio.
Scripts and Studio come with the Pro plan.
01
Bring your context
Attach up to five documents from your library, up to five links and your own notes. The writer takes its facts from there, and none of it appears in the video.
- Every fact comes with the source it was taken from.
- Links are kept as references for you. Their pages are not opened.

The context a script is written from 02
Write the script, checked as you write
A hook of a few words that works without sound, then the lines: the text on screen, what the voice says and up to two pictures each. Write it yourself, or let the AI draft it from your context or improve what you wrote.
- Curiosities, news, promo, how-to, story or free, in English, Spanish, Portuguese from Portugal or Brazil, and German.
- The Checks panel flags a long hook, a line with nothing to say, a fact without a source or a length off target. A script with errors cannot be marked ready.

A script, line by line 
Checks, updated as you type 03
Produce it in Studio
Choose a voice after hearing a preview. SPREVA records the narration and builds the video: scenes, your pictures and captions timed to every word.
- Change the script later and the video can follow, scene by scene.
- Record the narration again whenever the words change.

Or start from your photos.
Write a brief, choose photos and clips, and set the format, the length and the pace: calm, balanced or fast. Create with AI makes an editable first cut.
Then ask for changes in plain words, like “make the intro faster” or “shorten to 15 seconds”. Every change can be undone.

Everything stays editable in Studio
- Layers, a timeline, templates and your brand kit
- Scenes: hook, context, evidence, payoff, call to action
- Zoom, pan and a focus point on pictures and clips, and clip speed from 0.5x to 2x
- Narration, music and clip sound, each with fades
- Captions that highlight the word being said
- MP4 and PNG renders, saved straight into your library
Media library
A library that knows what is in it.
Every upload is described and tagged as it arrives, so you find a picture by what it shows and a video by what is said in it. New images and the documents behind a script live here too.
Described as it arrives
Each file gets a description and tags, and a warning when its name does not match the picture. Describe older files up to fifty at a time, or switch it off for the workspace.
Search in plain words
Type what you remember seeing. The library matches what is in the picture, and also file names, tags, alt text and folders.
Videos you can search by what is said
One frame per shot is looked at and the sound is transcribed, so each video gets a summary, its key quotes and a transcript you can search.
Alt text, suggested
SPREVA suggests alt text from the picture. You check it and save it on the file, and every post that uses the file gets it.


Make the picture you need.
Generate an image from a prompt and a shape, or open the Agent: describe what you want, attach up to four pictures from your library as references and ask for changes round by round.
Saved where you will look for it
Keep the results you like in Media, with a folder and tags, ready for the next post.
Documents are media too
The library takes PDF, text, Markdown, CSV and JSON files, and Keep as context turns an Assistant conversation into one. A script can use them as its sources.

Scheduling
Schedule it. Or stop scheduling entirely.
Give a post a date, or hand it to a queue and let the queue decide. Your whole month is one calendar, and changing a date takes two taps.
A calendar, not a list
See what goes out on which day, per account, and reschedule without opening the post.
Reusable queues
Define the slots once, say Tuesday 09:00 and Thursday 18:00, and every post you queue takes the next free one.
Time zones that behave
Slots are stored in the workspace time zone, so the hour you chose survives the clock change.


Results and recovery
Know what went out, and what did not.
Each network is published to independently and tracked independently. When one fails you get the network's own reason, in plain words, and a way out.
Per-account and per-platform
Follow reach and engagement by account or by network over any range, and export the numbers when you need them elsewhere.
The real reason it failed
Rate limited, media rejected, token expired: the failure is named, and so is the fix.
Retry the ones that failed
A post that reached four of five accounts is not republished to the four. Only the failed target is retried.


Everything else
The parts you only notice when they are missing.
Bulk upload and bulk schedule
A launch is rarely one post. Bring in a batch of videos or images at once and take the whole set through in a single pass.
01
Drop the files
Videos and images together, as many as you have.
02
Caption each one
And choose which accounts it belongs to.
03
Send the batch
To a date, or to a queue that spreads them out.
It fits in a phone
Every screen is built for a 390-pixel window first. Write, approve, reschedule and retry from wherever you happen to be.

Six roles, not two
Someone can write without being able to publish, or read the numbers without touching the content. Every action is recorded with who did it and when.
- Owner
- Admin
- Publisher
- Editor
- Analyst
- Viewer
Faster from the keyboard
Jump anywhere, create a post or search the screen you are on without reaching for the mouse.
- Search everything
- ⌘K
- Create a post
- C
- Go to the calendar
- GC
- Search this screen
- /
- All shortcuts
- ?
Act from the calendar
Duplicate, preview, check the numbers or delete a post right where it sits. Drag it to another day, or move it with the keyboard.
Posts, the way you think
A list or a board by status. Open a post or a file in a panel beside the list and keep your place.
Campaigns
Group the posts of one launch under a campaign, then filter drafts, the calendar and analytics by it.
Separate workspaces
One login, several companies. Accounts, media, queues and members never cross from one workspace into another.
For developers
Everything you publish, your code can publish.
The dashboard is one client of the API, not a wrapper around it. Same endpoints, same rules, same validation.
Keys with scopes
A key is created for one workspace and one set of permissions. A read-only key cannot publish, and every call is rate limited and recorded.
An OpenAPI document
Posts, media uploads, queues, social accounts, analytics and AI: write a caption, generate an image, suggest and set alt text. The whole surface is described by a spec you can generate a client from.
Signed webhooks
Thirteen events, from post.published to connection.revoked. Each delivery is signed, retried on failure, and replayable from the dashboard.
MCP for agents
Twenty-four tools over MCP, from scheduling a post to writing a caption, suggesting alt text or generating an image, so an assistant can work for you. Agents never see a provider token.
- 120 / minute
- Rate limit per key, with the wait in the error.
- 10 seconds
- To answer a webhook before it is retried.
curl -X POST https://spreva.app/api/v1/posts \
-H "Authorization: Bearer POST_API_KEY_..." \
-H "Content-Type: application/json" \
-d '{
"content": {
"caption": "Ship day. New build is live.",
"mediaIds": ["c47e2a90-0000-4000-8000-000000000004"]
},
"targets": [
{ "connectionId": "e81b3f52-...-005", "placement": "feed" },
{ "connectionId": "a31d9f07-...-006", "placement": "reel",
"override": { "caption": "Ship day. Full notes in bio." } }
]
}'curl -X POST https://spreva.app/api/v1/posts/$POST_ID/schedule \
-H "Authorization: Bearer POST_API_KEY_..." \
-d '{ "scheduledAt": "2026-10-01T09:00:00.000Z",
"timezone": "Europe/Lisbon" }'POST /hooks/post HTTP/1.1
x-post-event: target.published
x-post-signature: v1=6f1c… # HMAC-SHA256 of "timestamp.body"
{
"id": "7f3c2d10-0000-4000-8000-000000000000",
"type": "target.published",
"createdAt": "2026-10-01T09:00:02.412Z",
"data": {
"type": "TargetPublished",
"postId": "5d2a7c3e-0000-4000-8000-000000000002",
"provider": "instagram",
"remoteUrl": "https://www.instagram.com/p/EXAMPLE/"
}
}How it works
Three steps, then it is just writing.
01
Connect your accounts
Sign in to each network once. SPREVA keeps the connection alive, refreshes it on its own, and tells you the day a network asks for a reconnect.
02
Write the post
One caption, one set of files, the accounts you want. Ask the assistant for a draft when the page is blank, and override the wording where a network deserves its own.
03
Send or schedule it
Publish now, pick a date, or hand it to a queue. Each account is published to on its own, and you can see exactly how each one went.

Plans
Three plans, told apart by what they include.
Every plan gets the composer, the calendar, queues, the media library, the developer platform and a monthly AI allowance. What changes is how many accounts and people you bring, how much AI you use, and whether bulk upload, Studio and Scripts are switched on.
Free
Enough to run a couple of accounts properly.
- 3 social accounts
- 2 members
- 5 GB of media
- 100 publishes a month
- Analyticsincluded
- API, MCP and webhooksincluded
- Bulk uploadnot included
- Studio and Scriptsnot included
AI each month
- 25 AI suggestions
- No AI images
- No AI videos
- No narration
- 50 media descriptions
- 60 minutes of transcription
Starter
For a brand that posts every day.
- 5 social accounts
- 3 members
- 50 GB of media
- Unlimited publishes
- Analyticsincluded
- API, MCP and webhooksincluded
- Bulk uploadincluded
- Studio and Scriptsnot included
AI each month
- 300 AI suggestions
- 30 AI images
- No AI videos
- No narration
- 500 media descriptions
- 600 minutes of transcription
Pro
Most completeFor a team running several brands at once.
- 15 social accounts
- 10 members
- 500 GB of media
- Unlimited publishes
- Analyticsincluded
- API, MCP and webhooksincluded
- Bulk uploadincluded
- Studio and Scriptsincluded
AI each month
- 1,000 AI suggestions
- 100 AI images
- 60 AI videos
- 60,000 narration characters
- 3,000 media descriptions
- 3,000 minutes of transcription
Questions
Before you start.
The short version: one post, every network, and you always find out what happened to it.
Which networks can I publish to?
Facebook, Instagram, Threads, X, LinkedIn, Bluesky, Mastodon, TikTok, YouTube, Pinterest, Tumblr and Google Business. Each one has its own adapter, so it publishes the way that network expects rather than through a lowest common denominator. Mastodon works with any server: you enter your server's domain when you connect. Each Tumblr blog connects as its own account.
Can the same post read differently on each network?
Yes. The caption, title, link and media you write are the default; any account can override them. Before anything is sent, the post is checked against that network's own rules for length, number of files and accepted formats, so you find out in the editor instead of afterwards.
What does the AI do, and is it included?
It suggests and rewrites captions, adapts them per account, translates, suggests hashtags and first comments, suggests alt text, describes your uploads so you can search them, makes images, and turns a script into a narrated video. Every plan comes with a monthly allowance, shown under Plans: images start on Starter, and scripts, narration and AI video on Pro.
Where do a script's facts come from?
From the context you give the script: up to five documents from your library and your own notes, with up to five links kept as references. The AI takes facts only from there and names the source of each one. Every fact is listed under Facts to check, with its source or marked as an estimate, and the Checks panel flags one without a source. None of the context appears in the video.
Does anything get written or posted without me?
No. The assistant offers suggestions and you add the one you want. Alt text is suggested and saved only when you save it, and nothing is published until you publish, schedule or queue it. The one thing that runs on its own is the description of new uploads, which makes the library searchable, and a workspace setting switches it off.
What happens when one network fails?
Each account is published to independently and tracked independently. A post that reached four of five accounts stays published on those four; the fifth shows the network's own reason, whether that is a rate limit, rejected media or a withdrawn permission, and you retry only that one.
How do posting queues work?
A queue is a set of slots: weekday and time, in your workspace's time zone. Hand a post to the queue and it takes the next free slot. Because the slots are stored as a weekday and an hour, the time you chose still means that time after a clock change.
Can I work with other people?
Invite them into the workspace and give them a role: Owner, Admin, Publisher, Editor, Analyst or Viewer. Someone can draft without being able to publish, or read the numbers without touching the content. Every action is recorded with who did it and when.
Do you see my social account passwords?
No. You authorise SPREVA from inside each network, and what comes back is a token, stored encrypted. You can disconnect an account at any time, and when a network withdraws its permission SPREVA tells you rather than failing quietly.
Is there an API?
The REST API under /api/v1 covers posts, media uploads, queues, social accounts, analytics and AI writing, images and alt text, described by an OpenAPI document. Keys are scoped per workspace, webhooks are signed with HMAC-SHA256 and retried, and 24 tools are available to agents over MCP. Scripts, Studio and the Agent are used in the app.
Stop posting twelve times.
Connect your accounts, write the thing once, and get your afternoon back.
- Threads
- X
- Bluesky
- Mastodon
- TikTok
- YouTube
- Tumblr
- Google Business














