GPT Enterprise: The AI Business Suite

Technical Overview of GPT Enterprise

GPT Enterprise is a subscription AI suite for small and mid-sized businesses, professional firms and agencies. It began as a set of AI writing and chat tools and has been rebuilt around a single problem every company has: the way it really operates lives in a few people’s heads. The founder’s companion book Building an AI-Powered Company calls that person Deb, and GPT Enterprise is the product that writes Deb down before she retires.

At the centre is the Digital Brain, a company knowledge base that accepts pasted notes, uploaded PDFs and Word files, web pages, whole websites and meeting transcripts, organises them into collections, and answers questions with numbered citations back to the exact source. A browser extension, the Companion, saves pages and asks the Brain from any tab. Custom assistants are built on top of it: a Proposal Writer grounded on past proposals, an Onboarding Guide grounded on the playbook, each with its own instructions and a chat that keeps its persona for life. Alongside them sit a curated prompt library, a Data Scientist that analyses uploaded spreadsheets and draws charts, and an AI Researcher that investigates a topic and reports back with sources.

The Content Studio turns that knowledge into work: a document editor with rewrite, expand, tone and fact-check tools and exports to PDF, Word and HTML; a presentation builder; batch article creation; a brand kit; AI image generation, editing and compositing; a lead-magnet ebook wizard; and scheduled publishing to WordPress and other channels. Project Copilot runs the work that follows, with projects, boards, an approval queue, open commitments and promises, client portals opened by magic link, and summary emails. Edith, the AI meeting agent, joins video calls, records and transcribes them, answers by voice when addressed, and emails a recap with action items and decisions to the people who were there.

Everything is multi-tenant and team-ready: users and quotas, usage analytics per person, a white-label portal for agencies, credit-based metering of every AI action, and a self-service Your Data page that exports or deletes everything a company has stored.

Key Features

• Digital Brain
  • Add knowledge by pasting text, importing a web page or an entire website, uploading PDF, Word, text or Markdown files, or saving from the browser Companion.
  • Hybrid retrieval: vector and keyword search combined, with inline numbered citations and a resolved Sources list on every answer.
  • Collections and server-side search across names, content, notes, sources and tags; a “Use my Brain” toggle grounds ordinary AI Chat in company knowledge.
  • Brain Chat for conversations with the knowledge base itself, including meeting transcripts Edith has filed.

• AI Assistants
  • Custom assistants: name, description and instructions, optionally grounded on the whole Brain or on one collection; a chat opened with an assistant keeps that persona and its sources.
  • Prompt library of curated business prompts by department (Sales, Marketing, Support, Operations, HR, Finance, Strategy, Writing) plus the user’s own saved prompts.
  • Data Scientist for uploaded data sets with answers and charts; AI Researcher for sourced reports on any topic; streaming AI Chat with templates.

• Content Studio
  • AI document editor with versions, rewrite, expand, tone, SEO and citation tools and export to PDF, DOCX, HTML and Markdown.
  • Presentations and batch creation: build decks from an outline and generate many pieces from one brief; document templates and a brand kit keep everything on-brand.
  • AI images: create, edit, composite and overlay, with a searchable image library.
  • Lead-magnet ebook wizard from topic to outline, chapters, cover and PDF or EPUB, with narration and translation.
  • Post Automation publishes drafts to WordPress and other connected channels on a schedule, with a review queue.

• Project Copilot
  • Projects and boards with tasks, issues, notes and decisions; an approval queue for anything that needs a human yes.
  • Open commitments and promises: what the team owes clients and what clients owe the team, surfaced on the home page.
  • Client portals by magic link with per-file visibility; summary emails and file-live notices; a connector so approved AI assistant apps can read and update a project.

• Edith, the AI Meeting Agent
  • Schedules herself onto Google Calendar events with video links, joins, records and transcribes.
  • Hybrid voice: answers by voice when addressed during the call, grounded in the project’s files.
  • Emails a recap with summary, decisions and action items; recordings expire on a retention schedule.

• Teams and Account
  • Add users with per-feature quotas, usage and resource analytics, top-user leaderboards and white-label portal customisation.
  • Credit-based plans with annual options, API keys and documented REST endpoints.
  • Free business tools from the book (succession scorecard, knowledge-gap calculator) as public lead magnets.
  • Your Data: export everything as one archive or delete the account and all its data from Settings.

Underlying Technology

1. Models and AI layer: large language models for chat, writing and summaries, abstracted behind the product rather than exposed by vendor name; embedding models and a hybrid vector-plus-keyword index for the Digital Brain, with optional reranking; speech-to-text and text-to-speech for Edith’s live voice.
2. Interface: a responsive web application with guided tours, a browser Companion extension, magic-link client portals that need no login, and a standards-based connector so AI assistant apps can work inside a project.
3. Data security and privacy: strict per-tenant isolation on every query, encrypted storage of calendar and mailbox tokens, owner-checked file downloads, recordings purged on schedule, and a self-service export and delete-everything flow. Company knowledge is never used to train models.
4. Scalability and infrastructure: a Django and ASGI stack with WebSocket streaming for live meetings, a PostgreSQL database, background workers for ingestion, publishing and recaps under a self-monitoring watchdog, and credit metering so every AI action is accounted for.