PROJECTS

One space per client, niche or campaign — with memory that doesn't leak.

Every Project has its Knowledge Base (30 files, 300 MB), its Brand Voice, its history and its connectors. Multi-client agency work becomes routine — not juggling.

01 / ANATOMY

What lives inside a Project.

Knowledge Base

Up to 30 files (PDF, Excel, MP3, MP4) totaling 300 MB.

Brand Voice

Tone and vocabulary profile applied automatically.

History

Every thread in the project becomes context.

MCP Connectors

GA4, Google Ads, Meta Ads tied to the project.

Custom specialists

Profiles trained on the project files.

Entity Tagging

Semantic tags extracted from the files.
02 / KNOWLEDGE BASE

4 formats, 30 files, 300 MB.

📄

PDF

Briefs, decks, past reports.
📊

Excel (.xlsx)

Metric spreadsheets, media plans.
🎙️

MP3

Recordings of interviews and meetings (transcribed).
🎥

MP4

Product videos, past campaigns.

The AI consumes everything and turns it into permanent project context.

03 / EXTRACTOR

Built-in scraper — 300k chars per extraction.

300k chars
Per extraction — entire site in a single call
JavaScript rendering
SPA, Next.js, React — renders before extracting
Anti-bot bypass
For sites that block generic scrapers
Image extraction
Along with text, ready to use in another artifact

Ideal for: auditing competitors, extracting reference LP copy, mapping category arguments.

04 / WORKFLOW

Multi-client agency workflow.

  1. 01

    Create Project

    New project "Client X"

  2. 02

    Upload KB

    10–20 files: brief, reports, brand guide

  3. 03

    Scraper

    Extract client site + 2 competitors

  4. 04

    Voice + MCP

    Set up Voice (5-min Copilot) + connect GA4/Meta

  5. 05

    Inheritance

    Every conversation in this Project inherits all context

05 / SECURITY

Isolation across projects — by design.

  • Between projects, nothing is shared — no file, no Voice, no history
  • Data encrypted in Google Cloud buckets (AES-256)
  • TLS 1.2+ between client and server
  • MCP connectors use encrypted OAuth; tokens revocable per project
  • Your data does not train third-party public models

Frequently asked questions

Can I move files between projects?
Yes, via duplication/re-upload. By default, projects are isolated to prevent context leakage.
Does Knowledge Base have OCR for scanned PDFs?
Default is native text extraction. For scans, check with support.
Are MP3 audios transcribed automatically?
Yes. Audios become indexed transcripts for search.
Does the scraper work on large sites?
The scraper is per-URL. For large-scale crawl, use the integrations roadmap.
Can I share a Project with my client?
On the Teams plan, yes — with read/edit permissions.
If I delete a project, are MCP connectors disconnected?
Yes. OAuth tokens are revoked along with the project.
Ready?Ready?

Stop stitching 5 tools together. Run marketing from a single system.

11,000+ professionals in 40+ countries. First GEO audit free. No credit card.