AI Assistant Studio & Vector RAG Engine
Construct, train, and deploy private OpenAI vector store AI assistants in minutes.
Create purpose-built AI assistants for your team, give each its own private knowledge base, and control exactly who can use them.
Ready to automate your own business?Create Your Business Automation
Already a customer? Sign in to this product.Start Using
- Private per bot
- Knowledge base
- Role + person
- Access control
- Tokens & cost
- Usage ledger
AI Assistants
The Problem It Solves
Teams already ask AI for help, but each person writes their own prompts and the answers depend on what they happen to paste in. Company knowledge sits in documents the assistant has never seen, and nobody knows how much it is used.
Everyone prompts differently
The same question gets different answers depending on who asks and how.
Answers that ignore your documents
General answers do not reflect your policies, price lists or procedures.
Sensitive material with no boundary
It is unclear who may use an assistant that has seen confidential files.
Usage with no visibility
Nobody can say how much AI is being used or what it costs.
Assistants that need engineers
Setting one up for a team turns into a development project.
No record of changes
Changes to an assistant's instructions or files are not tracked.
Without it, and with it
The same everyday work, before and after.
Without it
Each person writes their own instructions.
With it
A bot captures its instructions and model once, so everyone gets consistent answers.
Without it
Answers come from general knowledge.
With it
Answers draw on passages found in the bot's own knowledge base, built from the files you upload.
Without it
Anyone with the link can use it.
With it
A sensitive assistant can be limited to specific people or roles, with manager oversight of all chats.
Without it
AI spend is a surprise.
With it
A usage ledger records executions, tokens, estimated cost and failures.
What it does
AI Assistants captures a prompt setup once so everyone gets consistent, grounded answers. A bot is configuration, not code: write its instructions, choose its model, upload its knowledge files, decide who can use it, and it appears in their sidebar.
Each bot's knowledge lives in its own private vector store that only that bot searches. Chats keep their own conversation context, and a usage ledger records executions, tokens, estimated cost and failures.
Primary purpose
Build, manage, and audit custom organizational AI agents with private company knowledge bases.
Product tour
A preview of the interface with sample data. Select a screen, then hover or tap the markers.
Create your own automated workspace
Register your company and set up AI Assistants with the rest of the platform: one sign-in, shared roles and automations that connect your work.
Key features
What you can do with it, day to day.
No-code bot factory
Name a bot, write its instructions, pick a model and publish it to the people who need it.
Private knowledge bases
Upload PDF, Word, text, CSV and Excel files. Spreadsheets are converted so they can be searched, and each bot's files are indexed separately.
Role and person access
Limit a sensitive assistant to specific people or roles, with manager oversight of all chats.
Chat workspace
Every bot has its own chat page and chat history.
Usage dashboard
Chats, executions, token usage, estimated cost and failures at a glance.
Audit log
Every bot, file and chat change is recorded.
AI capabilities
What the AI does in and around this product — concretely.
Retrieval from your files
Answers draw on the passages found in the bot's own vector store, so replies are grounded in your documents.
Instructions and model per bot
Each bot has its own system instructions and OpenAI model, so a proposal writer and a policy assistant behave differently.
Spend visibility
Every execution is written to a usage ledger with tokens and an estimated cost, and AI usage counts toward the company's plan limit.
Automation & workflows
How work moves from start to finish without hand-offs and re-typing.
Launch an assistant
From idea to available in the sidebar.
- 1
Create the bot with instructions and a model
- 2
Upload knowledge files; each is processed into the bot's vector store
- 3
Choose which roles and people can use it
- 4
It appears in their sidebar and chats can start
Keep the knowledge current
Replace documents without rebuilding the bot.
- 1
Upload the new version of a file
- 2
Processing status updates until it is ready
- 3
Disable or remove outdated files so they are no longer searched
Use cases
Real tasks, real roles.
Proposal drafting
Sales uploads past proposals and the service catalogue, then asks the bot for a first draft for a new client.
Policy answers
HR uploads handbook files so staff can ask about leave or conduct and get answers from the real policy.
Requirement analysis
A business analyst feeds in specification files and asks for gaps and questions to raise with the client.
Meeting follow-ups
A team uses a bot configured to turn rough meeting notes into action lists.
Supported business scenarios
- A sales team that writes proposals from past wins
- An HR team answering repeated policy questions
- A delivery team that wants a specification reviewer
Benefits & business outcome
What changes for your business.
Staff stop re-pasting company context into a general chatbot, and managers can see what the assistants are used for and what they cost.
Consistency
Everyone gets the same grounded answer instead of their own prompt experiment.
Privacy by structure
A bot searches only its own knowledge base, and the API key never leaves the server.
Cost control
A ledger shows who used what, so spend is not a surprise.
Speed to launch
A new assistant is configuration, so teams build their own without engineering work.
Who it's for
AI-minded teams in sales, support, HR, business analysis and engineering who want assistants grounded in their own documents.
Departments
- AI & Engineering
- Business Analysis
- Sales & Support
- Product Management
Typical users
- AI Engineer
- Business Analyst
- Product Manager
- Team Staff
Works with
The other products and integration points it connects to.
OpenAI
Responses, Conversations and vector stores power the assistants.
Frequently asked questions
Each file is stored in a vector store that belongs to one bot and is searched only by that bot. The application keeps metadata, such as file records and chat pointers.
Request a demo
Ready to put AI Assistants to work?
Create your company workspace, or ask for a walkthrough with your own scenarios.
- Each company's data is kept separate from every other company's.
- One sign-in and one role model across every product.
- AI and automations that work across the products.