Skip to main content

Management Hub

The Management page is where you upload and organize the supporting materials the bot uses in conversations — master plans, training documents, project images — and where you decide who has access to the project.

Master Plans​

Management → Master plans. Master plans are PDF documents — site plans, route maps, layout overviews — that the bot can send to a client when asked. Common examples:

  • A real-estate development's overall site map
  • The floor plan for a building or unit
  • A service area map

For each master plan you can:

  • Upload new PDFs
  • Drag-to-reorder them so the most important plans show first
  • Delete outdated ones

The bot recognizes phrases like "can you send me the master plan?" or "do you have a layout?" and replies with the master plans you've uploaded.

Training Files​

Management → Training. Training files are documents you keep on the project as a shared reference library — brochures, FAQ sheets, spec documents, internal manuals.

Supported formats: .pdf, .docx, .txt, .json, .eml, .md, .html.

For each training file you can:

  • Upload training file — add a new document
  • Queue for processing — send a file through processing again
  • Train / Untrain — include or exclude the file's processed content

Two things you might expect are not available: a training file cannot be deleted once uploaded, and downloading the original is currently switched off — the button reports that downloads are unavailable. The same applies to master plan downloads. Keep your own copy of anything you upload.

Uploaded files are processed in the background, and a file that fails to process is retried automatically. Large documents take longer than short ones. Each file shows its state as Pending, Processing, or Trained.

Training files do not currently shape the bot's replies

Files uploaded here are stored and processed, but the assistant does not currently draw on them when it answers a client. A file can reach Trained and still have no effect on what the bot says, and the page's own wording about the AI learning from these documents overstates what happens today.

What genuinely shapes answers today is:

If something in a training file needs to reach clients, put it in one of those three places.

Project Images​

Management → Images. Project images are the photos clients see when they ask for visuals — "can you send me some pictures?" or "show me images of the project."

For each image you can:

  • Upload new photos
  • Reorder the gallery (the order is the order the bot sends them in)
  • Delete old or low-quality ones

For per-product images (the photos of a specific apartment unit, for example), use the Catalog instead.

Image Categories​

An image category names what's actually in a photo — Master bedroom, Entrance, Pool, Lobby — so the bot can match a photo to what a client asked for instead of sending the whole gallery every time.

Each category has two fields:

  • Name — a short label for the place or subject, up to 100 characters (e.g., Master bedroom)
  • Synonyms — the words a client might actually use for it, in any language, up to 2000 characters. At least one word is required, because this is what the matching relies on. Be generous: main bedroom, principal, suite, dormitorio principal.

Create them with New category, revise them with Edit, and open the full list with Manage categories. There are two ways to tag your photos:

  • The batch picker on upload — "Tag every photo in this upload with" applies the same categories to everything in that upload at once
  • Edit categories on an individual photo — for correcting or refining a single image afterwards
There is no separate Image Categories page

Tagging happens where the images are. You create and assign categories inline here, and in the product image uploader inside the Catalog. Nothing else needs to be configured elsewhere.

What it changes in conversation: when a client asks for something specific, the bot sends only the photos matching what they asked for, then offers to send the rest. Without categories, it has no way to tell one photo from another. See Media Sharing for how the bot decides what to send.

Deleting a category removes it from every photo and file it was tagged on, and that can't be undone — the images stay, but they lose that tag and stop matching requests for it.

Image categories are not AI Categories

They sound alike and do completely different jobs. Image categories describe what's in a photo so the bot sends the right visual. AI Categories classify leads — hot, cold, investor, and so on — so your team knows who to chase. Changing one has no effect on the other.

Team​

The Team card controls who can access this project and who answers as an agent. Use Add member to search for someone by name or email and add them — this is the only place a person is granted access to the project.

Each row carries independent switches rather than a single role dropdown:

  • Agent — agents can pick up chats and reply on behalf of the project
  • Available — agents only receive new chats while available
  • Remove — takes them off the project, and they're notified by email

Roles shown here are Owner, Admin, and Member. For what each role can do, and for the availability and contact-phone settings each person manages themselves, see Agents & Team.

They need an account first

There's no email invitation flow. The person must already have a Remo account before the search will find them, so ask them to sign up first and then add them.

Tips​

  • Keep master plans and project images up to date — outdated visuals are worse than no visuals.
  • Give every photo at least one image category before you rely on the bot sending photos. An untagged gallery can only be sent whole.
  • Write synonyms the way clients speak, not the way your team writes. Add the words in every language your clients use.
  • Group photos in a sensible order — clients often look at only the first few.
  • When product or pricing details change, update the catalog and your project characteristics. Re-uploading a training file will not change what the bot says.