Chats
All conversations inside this project. Clearly structured, searchable. Every chat starts with the full project context.
One project.
One context.
Many chats.
Your AI workspace for everything that takes longer than one chat.
A mandate, a client, a research topic, a product launch – anything you work on with AI over weeks or months belongs inside an anymize project. Define the context once (instructions, memory, up to 30 knowledge bases) and every new chat in that project starts fully informed. Share the project with your team, and everyone works from the same knowledge base – with individual, private chats.
What is a project?
Context
that stays.
A project is a bundle of chats sharing a common, recurring context. Instead of re-explaining your background in every new chat (“I'm working on mandate X, contract type Y, problem Z …”), you define the context once in the project. Every new chat in the project starts with full context access.
The result: you don't start from zero. The AI knows what this is about. You ask what you want to know – without a warm-up round.
Projects are ideal for anything that is recurring or grows over time: an ongoing mandate, a client relationship, a product development cycle, a research direction, a long-running theme in your work.
The four context pillars
Four building blocks
you define per project.
Every project stands on four building blocks. You decide which ones you need for this project:
Instructions
A project-specific system prompt. Tone, format, role description, style rules, abbreviations – everything you would always pass to the AI.
Project memory
A knowledge store that grows over time. The AI remembers important facts from your chats; you can add or delete entries manually.
Knowledge bases
Link up to 30 knowledge bases per project. Contracts, policies, research archives, case files – your company knowledge is instantly available in chat.
Instructions
Your standard briefing
for the AI.
Instructions are the project-wide system prompt. Written once, they apply to every chat in the project. You use them to define, for example:
01
Role of the AI
"You act as a contract lawyer with a focus on European IT law.”
02
Answer format
"Answer in bullet points, no more than 300 words, with references to the clauses.”
03
Tone
"Formal, but not stiff. Address me informally.”
04
Rules
"Never mention the company name in plain text. Only ever use the label 'the client'.”
05
Abbreviations / terminology
"When I write 'AV', I mean the Anlagenverordnung. When I write 'BEM', I mean Betriebliches Eingliederungsmanagement.”
In practice: Write once, benefit forever. Instructions save you the first five minutes of context-building in every new chat.
Project memory
A memory that learns
per project.
Project memory is a growing collection of project-specific facts. After each chat, anymize extracts information that might matter for future chats in this project: names of people involved, technical boundary conditions, agreed processes, intermediate results, open questions.
That knowledge is available from the very next chat onwards. After ten chats the AI has a solid picture of the project – without you ever having explicitly told it.
Two layers, one answer
Base layer
Personal memory
Active in all your chats.
Project layer
Project memory
Adds on top inside this project.
The AI combines both layers automatically – it knows you and the specific project context.
Two memory layers that work together
Important: project memory does not replace your personal memory – it complements it.
- Personal memoryFacts about you: profession, working style, communication preferences, favourite formats. Considered in all your chats.
- Project memoryFacts about the project: participants, background, intermediate status. Additionally considered whenever you chat inside this project.
Personal memory can be switched off per chat
You stay in control. Per chat, you can deliberately disable your personal memory. Useful when:
- You work on a sensitive project and do not want your personal working habits to flow in
- You act for a colleague inside a shared project and should stay as neutral as possible
- You want a "clean" analysis, independent from your usual habits
Project memory itself stays untouched.
Granular control
Every memory entry is visible and editable. You see what the AI knows about you and about your project – and can delete or adjust individual entries. No black box.
Up to 30 knowledge bases
Your project archive,
always at hand.
Per project you can link up to 30 knowledge bases. Every database is instantly available to every chat in the project – no manual activation per chat.
Typical compositions
Legal mandate
Client file · Contract templates · Case-law archive · Firm standards
Consulting project
Past analyses on the client · Industry research · Methodology handbook
Product launch
Product documentation · Customer feedback · Competitor analyses · Brand guidelines
Clinical case
Patient file · Treatment guidelines · Internal SOPs · Specialist publications
Research topic
Literature collection · Experiment data · Notes · Partner documents
Databases are searched automatically per question (Retrieval-Augmented Generation). The AI only cites the passages that match your question – token costs stay low, answer quality stays high.
Use cases
Projects in
the real world of work.
Six prototypical project types – drawn from the real working contexts of our customers:
Ongoing mandate (lawyer)
Instructions
Role as case handler, §203 sensitivity, citation style
Project memory grows with
Key participants, deadlines, line of argument
Knowledge bases
Client file, case law, firm templates
Due diligence (consulting)
Instructions
Structure by financial / legal / tech DD, executive-summary format
Project memory grows with
Identified risks, open questions, decision paths
Knowledge bases
Reports, interview transcripts, comparable deals
Product launch (PM)
Instructions
Tonality for customer communication, positioning, KPIs
Project memory grows with
Milestone status, team names, key blockers
Knowledge bases
Product documentation, competitor analyses, brand guide
Clinical case (physician)
Instructions
Language for patient records, abbreviations, documentation standard
Project memory grows with
Anamnesis, findings, course of therapy
Knowledge bases
Patient file, guidelines, specialist literature
Dissertation / research
Instructions
Academic style, citation style, argumentation structure
Project memory grows with
Research question, hypotheses, intermediate results
Knowledge bases
Literature, data sets, notes
Customer account (sales)
Instructions
Role as account manager, CRM language, negotiation guidelines
Project memory grows with
Names of decision makers, past interactions, objections
Knowledge bases
Proposals, emails, product data sheets
The pattern
Wherever you need context more than once – and especially wherever a team has to stay on the same page.
What you should know about projects.
Frequently asked questions
A bundle of chats sharing a common context. Every project can have instructions (a project-specific system prompt), a project memory (a growing collection of facts), up to 30 knowledge bases and a participant list. Every new chat in the project starts with the full context.
We stand behind anymize. And we know – when an AI tool touches client, patient or employee data, a demo video isn't enough. That's why we give you 14 days of full access – all models, all features, no credit card. Enough time to be certain, before you trust us.
Your AI workplace awaits.