AssistantOS brings together 140 documented use cases in 13 chapters – from research and writing to image, video and audio, collaboration, communications and administration. The scenarios come from the platform's real feature set and can be filtered by roles such as consulting, creative, marketing and communications.
Use cases for consulting, creative, marketing, communications and administration – from the real feature set of AssistantOS. Not a click through ten killer features, but the many small and large moments in daily work where the platform takes work off your hands.
Not every client has every capability enabled – cards with a badge show where a capability depends on plan tier or is currently being integrated.
A quick way in: click your role and the explorer above filters straight to the use cases that will change your day-to-day work the most.
Your leverage is research depth and synthesis speed. Start with: topic deep-dives in Agent Mode, SharePoint topic research, Excel analysis in conversation, a knowledge base with citations, project boards for switching between clients, and visibility and reputation analysis as a new consulting product.
View use casesYour leverage is the path from idea to asset. Start with: draft documents for iterative writing, image generation in brand style, recurring campaign characters in Character Studio, image and audio editing in the browser, live moodboarding in Canvas, sentence comparison Word vs. InDesign – and Memory that learns your tone of voice.
View use casesYour leverage is output per campaign. Start with: industry news and competitor monitoring, content repurposing into five formats, banner sets and carousels, campaign evaluation by question instead of pivot tables, social media integration, Visibility Tracker for GEO proof.
View use casesYour leverage is consistency across many channels. Start with: press releases in draft mode, interview transcription, Staffbase publishing with reaction simulation, Podcast Studio, Reputation Tracker as an early warning system, Duden QA before every approval.
View use casesYour leverage is the sum of small tasks. Start with: querying your inbox and drafting replies, calendar in chat, Word documents from templates, OCR for scans, cleaning and merging lists, reminder routines.
View use casesYour leverage is overview and accountability. Start with: project boards, proposal and report creation, shared chats for alignment, approval workflows, cost dashboard, workspace switching between clients.
View use casesYour leverage is enabling everyone else. Chapter 13 is yours – from the knowledge base over coffee to model selection and the feature toggles that bring your team along step by step.
View use casesThe capabilities behind every use case in this document – across all chapters.
Content, research, strategy, image, audio, visibility: you work with a catalogue of specialised assistants, each with their own knowledge, skills and stance. Pin your favourites and search finds assistants and chat content alike.
Agent Mode plans multi-step tasks itself: research, then write, then illustrate – in one request, with a live view of which step is running. What the market calls agentic AI is the normal way of working here.
Voice mode turns chat into a conversation – you speak, the assistant replies with voice and can use tools along the way. And with Marvin there is a dedicated voice assistant for quick requests in between.
Images, documents, tables, comparisons and dashboards appear as structured cards in chat – with actions like edit, download and pin right there.
Clarifying questions when things are ambiguous, approval loops for critical actions: autonomy yes, but with you in the loop.
Pin and group chats, full-text search; drafts are kept; titles are assigned automatically by AI.
The full interface in German or English – for international teams without friction.
GDPR-compliant, tenants strictly isolated, your data does not train models. Security is standard equipment, not an add-on.
Maybe you do not use Staffbase or an editorial plan in Airtable, but a different intranet or your own PIM, DAM or CRM.
The honest answer: probably yes, and faster than you think. AssistantOS is built around a plugin system that exists exactly for this: connecting new systems as sources and destinations. A new connector is not a menu click, but it is not a major project either. It starts with a conversation.
The best platform helps little if rollout is left to luck. That is why we deliver the path to AssistantOS too: with FlowCheck, our transformation platform that our consultants use together with your team. Process first, tool second.
In the readiness workshop you capture your department's core processes as visual process chains with our facilitators: who does what, with which tools, and where does it hurt? The result is a prioritised process backlog instead of a gut feeling.
Flow modelling in workshopsFlowCheck calculates AI readiness per process and shows in what-if scenarios what happens when a step runs with AI support. Rollout starts where the hurdle is low and the leverage is high.
Gap analysis and what-if scenariosThe skill matrix shows per role which competencies are missing for the new AI-supported processes, from prompting to confident review of AI output. That becomes a targeted enablement programme.
Skill matrix and role profilesThe AI Maturity Index condenses depth and breadth of AI usage into a score from 0 to 100, updated quarterly, with a dashboard for management.
AI Maturity Index with dashboardThe 100-day assistant supports implementation: it automatically records conversations and decisions in notes and generates an ongoing action plan from them.
100-day assistant with action planEvery agreed measure is tracked with status, priority and owners. AI-supported reports are generated at the push of a button and sent to stakeholders as PDF.
Measure tracking and report builderFlowCheck is a guided consulting offering: our facilitators lead workshops and analyses, your team works along and keeps full visibility. Talk to us when you are planning the rollout.
The use cases at assistantos.de/en/anwendungsfaelle describe 140 concrete work scenarios from the real feature set of AssistantOS – from research and writing to image, video and audio, collaboration, communications and administration.
Use cases are filterable by role: consulting, creative, marketing, communications, admin/office, agency management and team lead/admin. Many scenarios apply to several roles at once.
The explorer groups 140 use cases into 13 chapters, for example research and understanding, writing and editing, design and production, analysis, knowledge, collaboration and setup for team leads.
No. Not every client has every capability enabled. Use cases with the badge "Depends on plan tier" depend on the booked package; "Integration in progress" marks features currently being connected.
Yes. On desktop the sidebar filters by role; on mobile via the floating filter pill. Multiple roles can be combined. Clicking a role in the "Your starting point" section applies the filter directly.
Yes. The page is available in English at /en/anwendungsfaelle – with the same 140 use cases, chapters and role filters.
Book a personal demo and see AssistantOS live – no obligation, in 30 minutes.
You need the core message of a long company page, blog post, or annual report on the web. You paste the URL into the chat – AssistantOS fetches the text, strips navigation and ads, and summarises what actually matters.
Web scraper: text extraction
You want to know what the competition announces on their press page – but you do not want to check manually every day. You tell your assistant once: "Check this page on weekdays at 8 a.m. and summarise anything new for me." From then on, the briefing is waiting in your chat before your coffee is ready. The market calls this autonomous agents – for you, it is simply one less thing to worry about.
Web scraper + Schedule Manager
For the pitch, you need an overview of a prospect's full web presence – services, team, references, tone of voice. AssistantOS crawls the site page by page and delivers a structured summary instead of fifty open tabs.
Web scraper: website crawling
Competitor pricing, trade show dates, rankings – data sitting in a table on some website. AssistantOS extracts it in a structured form, and you can follow up immediately: "Where do we stand in comparison?"
Web scraper: table extraction
Modern websites often load their content only in the browser – ordinary tools then see nothing but an empty shell. AssistantOS renders those pages in a real headless browser and reaches the content others miss.
Web scraper with Browserless rendering
Numbers, names, events – before anything goes into a press release, it should be correct. Your assistant searches live on the internet and delivers up-to-date information with sources, instead of guessing from outdated knowledge.
Live web search via the AssistantOS gateway
"Research the state of the debate on topic X, compare it with our internal positions, and summarise the three most important tension points for me." The assistant searches the web, reads multiple sources, checks your knowledge base – and delivers a synthesis instead of a link list. The market calls this deep research; here, it is a single prompt.
Agent mode: web search + web scraper + knowledge base in one run
Trade media, competitor blogs, association announcements – everything has an RSS feed, but nobody has time for twenty feed-reader subscriptions. AssistantOS reads the feeds, filters by your topics, and bundles the relevant items into one overview.
Autonomous toolbox: RSS parsing
The competitor figures are on page 47 of a 120-page PDF – in a table you cannot copy. AssistantOS extracts tables and KPIs directly from PDF documents and makes them analysable: compare, calculate, visualise.
Autonomous toolbox: PDF table extraction
"What do our documents say about product line Y?" The assistant full-text searches your SharePoint sites, finds the relevant documents, and quotes the matching passages – you stay in the chat.
SharePoint connector: full-text search
For the strategy paper, you need everything the company has ever written on a topic – spread across dozens of documents. AssistantOS reads up to ten documents in combination and distils a topic overview with references.
SharePoint connector: topic research
Who has already blogged internally on this topic? What is in the wiki? What appeared on the timeline recently? AssistantOS searches your Haiilo intranet – blogs, wiki pages, timeline, workspaces, events – and delivers the results prepared in the chat.
Haiilo connector
You are looking for internal expertise for a project. The assistant searches employee profiles in the intranet and names people who fit the topic – including context on where that information comes from.
Haiilo connector: people search
The interview with management, the recording of the press conference, the voice note from the client: you upload the audio file and get a clean transcript – including who spoke when. What used to be an hour of typing becomes two minutes.
ElevenLabs transcription with speaker diarisation
Some information is not on websites but in open JSON APIs – event data, statistics portals, industry directories. AssistantOS calls those interfaces directly and makes the data usable in the chat.
Web scraper: JSON API fetch
You describe occasion, audience, and core message – AssistantOS delivers the newsletter not as a wall of chat text, but as an editable document right beside the conversation. You change sentences yourself or tell the assistant what to rewrite. No copy-paste into another tool.
Draft documents with live editing
Facts in, first draft out – then you iterate: sharpen the quote, rework the lead, attach the boilerplate. The draft stays as a document and grows with each round, instead of disappearing in the chat history.
Draft documents
The delicate decline, the friendly reminder, the outreach to a new client: you provide bullet points, AssistantOS writes – and because it remembers your tone of voice, it sounds more like you over time instead of like a machine.
Draft documents + Personal Memory
Before approval, the text runs through official Duden checking: spelling, grammar, style – directly in the chat, with concrete correction suggestions. This is not an LLM guessing; it is the Duden engine that knows.
Duden spell check
"Innovative" for the third time in the same paragraph? The assistant gives you Duden synonyms and alternatives in context – faster than any dictionary.
Duden synonym search
The expert article is done – now you need the LinkedIn post, the newsletter teaser, three social snippets, and an FAQ section. One brief, all formats, each in the right tone for the channel. Content repurposing without busywork.
Chat + draft documents
You correct once: "We do not say 'solution', we say 'tool'" – and AssistantOS remembers. In the next text, the rule is simply there without you repeating it. Never explain the same context twice again.
Personal Memory
Your colleague built the perfect prompt for product copy? It lives in the shared prompt library – searchable, tagged, copied into your chat with one click. Good prompts become team knowledge instead of solo secrets.
Prompt library
The board text in German, the version for the international team in English – translation and adaptation in one step, with tone in mind instead of word-for-word. The interface itself is also available fully in both languages.
Multilingual assistants, bilingual UI
Invitations to a client event, personalised by name, company, and topic focus: you supply the list and the template, AssistantOS fills them via a template engine – hundreds of variants, zero copy-paste errors.
Autonomous toolbox: template-based bulk content
"Turn these notes into an agenda for tomorrow" or "Format this as a briefing document with goals, audience, and messages." Structural work that normally eats 20 minutes happens on the side.
Draft documents
"What did we call that claim again?" Search goes through not only your assistants but also your chat content – and jumps straight to the spot in the old conversation.
Global chat search with jump to message
More and more people ask ChatGPT instead of Google. For your content to appear there, it needs structure: FAQ formats, Schema.org markup, clear entities. AssistantOS generates and validates those structured formats – generative engine optimisation, implemented in practice instead of as a buzzword.
Chat + autonomous toolbox: JSON-LD/schema validation
Your organisation has a values charter on file? Then those principles flow invisibly into every answer – and a downstream "conscience check" flags when a text contradicts your own standards. Positioning as a system setting.
Charter values system: Echo & Conscience
You need a visual quickly for the social post or concept sketch: enter the prompt, wait a few seconds, image in the chat. For the quick draft there is the speed model; for the big deliverable, the quality tier.
FLUX image generation
When it should be more than a draft: photorealistic images up to 4K resolution, with format presets for common ad sizes. From prompt to presentable visual, without stock-licence research.
GPT-Image 2.0: generation up to 4K
Not some generic AI look, but your look: with a style model trained on your visual world, AssistantOS creates images that fit your CI – recognisable across every channel.
FLUX with brand LoRA
For employer branding, internal campaigns, or a slightly lighter tone: images in a Pixar-like illustration style – approachable instead of sterile.
FLUX Pixar style
"Make the background lighter", "Swap the product in the hand", "Remove the person on the left": you upload an image and describe the change in plain language. Retouching without Photoshop skills.
GPT-Image editor, Nano Banana image editing
The only available photo is too small, too soft, too noisy? AssistantOS upscales, sharpens, denoises – and removes the background on request. From a phone snapshot to a usable asset.
Topaz image enhancement: upscale, sharpen, denoise, background removal
Crop, add text, filters, drawing: the built-in image editor opens any generated or uploaded image right in the platform. For the last 10%, you do not need a second tool.
Image editor in the browser
The animated product scene, the mood clip for the social reel: you describe the scene, AssistantOS generates the video – in high quality for delivery or in the fast variant for trying out ideas.
Google Veo 3 / Veo 3 Fast video generation
The explainer film needs a voice, the social video needs a speaker: text in, choose a voice, finished audio track out – with emotional shading from neutral to enthusiastic on request.
ElevenLabs text-to-speech with emotion control
The jingle for the podcast, the ambience for the trade show clip, the transition sound: AssistantOS generates sound effects and full music tracks from a description – royalty-free because freshly composed.
ElevenLabs sound effects & music
The assistant wrote the text – now it should speak it too. You split the answer into segments, choose a voice per segment, and merge everything into one audio file. From text to audio track without leaving the platform.
Audio composer in chat
Opening too long, level too quiet? The built-in audio editor trims, fades in and out, normalises volume, and exports as WAV or MP3.
Audio editor in the browser
Podcast Studio is its own workbench: you compose episodes from segments – AI voices, your own recordings, music, sound effects, pauses – maintain a pronunciation dictionary for company names and technical terms, master to broadcast loudness, and export the finished MP3. Corporate podcasts without a production company.
Podcast Studio
Medium rectangle, leaderboard, skyscraper: you describe message and look, AssistantOS generates the full banner set in standard sizes as HTML – ready as a ZIP for the campaign.
Google Ads banner generator
SWOT matrix, process timeline, comparison chart, org chart: you supply the content, AssistantOS builds the infographic – in your organisation's configured design on request, exportable as SVG or PNG.
Infographic generator with tenant theming
Not every chart has to be static: AssistantOS builds interactive HTML visualisations – dashboards, clickable charts, small widgets – that you explore directly in the chat and export as PNG for the presentation when needed.
Visualization Bridge + PNG export
From the long text to a multi-page image carousel in 4:5 format: core statements per slide, consistent layout, CTA at the end. AssistantOS handles text distillation and image assembly in one pass.
Autonomous toolbox: image assembly via code
The video as a GIF for the internal email, the podcast recording as MP3, the audio track extracted from the video: AssistantOS converts audio and video files between all common formats.
Autonomous toolbox: media conversion
The biggest problem with generated image worlds: the same character looks different in every image. In Character Studio, you define your figure once – with reference images, description, and a list of non-negotiable traits – and train a dedicated style model for it. After that, you create your mascot in any scene: trade show booth, social post, Christmas card – always recognisably the same character.
Character Studio: character management + LoRA training
The testimonial figure in the office, with the client, at the event – without reshoots and without a photographer: for quick variants there is the speed lane; for difficult poses and final visuals, the high-end lane with maximum face and outfit consistency up to 4K.
Character Studio: quick & advanced edits with anchor references
The strongest still becomes a reel: from every image of the character you generate short video clips – head turn, smile, camera move – optionally in cinema quality with native sound or in the fast variant. Successful clips can be extended and built into sequences.
Character Studio: image-to-video with Veo 3.1 / Kling, video extend
Career posts, recruiting stories, intranet visuals – with recognisable figures in a consistent look instead of random stock faces. And where real photos are legally sensitive or simply unavailable, consistent AI figures provide the alternative.
Character Studio
A result from the gallery is especially strong? You adopt it into the character's reference pool with one click – future generations orient themselves to it. Your characters become more precise the more you work with them.
Character Studio: adopt gallery hits as anchors
Not "looks like PowerPoint", but your master template: AssistantOS fills uploaded .pptx templates with content from the chat. Fonts, colours, slide layouts – everything stays on-brand because it is your file.
PPT generator from templates
When it does not have to be the company template but should look good fast: AssistantOS creates presentations via Gamma – fully designed, with a link for further editing and export as PDF or PPTX.
Gamma presentations
The letter, the concept paper, the project report – generated directly in your .dotx Word template, with logo, styles, and footers. Download, proofread, send.
Word generator from corporate templates
Markdown to PDF, HTML to Word, concept to EPUB: AssistantOS converts documents between all common formats. The research result becomes a ready-to-send attachment without you formatting it.
Autonomous toolbox: document conversion via Pandoc
The monthly report needs numbers, diagrams, and interpretation – AssistantOS assembles all building blocks into a finished PDF or Word document: chapter structure, embedded charts, consistent formatting.
Autonomous toolbox: report assembly
You describe services and effort in the conversation – out comes a formatted proposal PDF with deterministically calculated pricing. No Excel formula errors in the client document.
Proposal PDF generator
The assistant built a landing page section or an HTML snippet? The built-in preview renders it safely in a sandbox – you see the result the way the recipient would.
HTML preview in chat
Copy formatted for email, plain text for the CMS, download for filing – every answer, every image, every audio and video file can be exported directly from the chat.
Copy and download functions, universal download
The campaign evaluation, the budget list, the survey export: you drag the file into the chat and ask: "Which channels are below plan?" AssistantOS loads the data, calculates, and answers – with numbers, not guesses.
Data analytics: CSV/Excel analysis
"Show me revenue by region and quarter as a cross-tab." What is menu acrobatics in Excel is one sentence here. The cross-tab appears in the chat, as a chart on request.
Data analytics: crosstabs & charts
The raw export from the survey tool is a data swamp. AssistantOS describes the distributions, finds the notable differences between groups, and formulates the core messages – the basis for a report or press release.
Data analytics + chat synthesis
Sort 5,000 free-text answers by topic, categorise job titles, tag product names: the assistant reads the table, classifies the entries, and writes the labels back as a new column – as a finished CSV for download.
Autonomous toolbox: read_table + apply_labels
"Müller GmbH", "Mueller GmbH", "Müller Gesellschaft mbH" – obvious duplicates to you, three different companies to Excel. AssistantOS recognises similar entries via fuzzy matching and cleans your lists.
Autonomous toolbox: fuzzy deduplication
Sales delivers a spreadsheet, the event tool a CSV, the CRM an export – each with different columns. AssistantOS merges the sources, normalises formats, and delivers one consistent master list.
Autonomous toolbox: pandas data consolidation
Sometimes no tool fits: a special weighting logic, a custom score, an unusual transformation. Then the assistant writes the analysis code itself, runs it in an isolated sandbox, checks the result, and refines – an agentic loop where you only define the goal. Your data never leaves the protected environment: the sandbox has no internet access whatsoever.
Autonomous toolbox: Python sandbox without egress
Every analysis can end in a chart: bars, lines, distributions – exported as PNG and ready for the next slide.
Data analytics: chart rendering, matplotlib
Instead of ten static charts: an interactive dashboard in the chat where you filter, zoom, and explore data points – and export as an image when the state is ready to present.
Visualization Bridge
You have style guides, product manuals, and strategy papers in the knowledge base? Then the assistant answers from them – and shows for every statement which document and page it came from. Trust through footnotes, not through assertion.
RAG knowledge base with inline citations
Today you need the brand guidelines, tomorrow the product knowledge: while writing, you choose which knowledge collections the assistant should use for this message.
Collection selection per message
For complex questions, the assistant does not wait for what is automatically supplied – it formulates its own search queries in your collections, reads whole documents in full text, and searches for terms deliberately. Agentic RAG: the AI works with your knowledge like a good researcher.
Agentic knowledge search: search, full text, keyword
The scanned legacy archive, the layout-heavy brochure, the diagram in the attachment: with vision parsing, AssistantOS understands documents that classic text extraction fails on – and makes them searchable.
RAG vision parsing
Today's research should help everyone tomorrow: the assistant writes results – competitor profiles, FAQ answers, topic dossiers – back into a configured knowledge collection as documents, where they are immediately searchable and available to other assistants too. It can update, list, and remove them later. Knowledge compounds instead of leaking into chat history.
RAG write: persist knowledge in configured collection
AssistantOS remembers what you are working on, which projects are running, and how you like things – automatically, in the background. In the next chat, it is simply there. And you stay in control: view everything, delete individual items, export completely, or switch on incognito mode when something should not be remembered.
Personal Memory with GDPR export and incognito mode
Some things should not be left to chance: you mark a passage in the chat and save it deliberately to memory – or have the whole chat summarised and stored. Your personal project archive that the assistant actively uses.
Manual memory + chat summarisation
"I am a consultant in healthcare, I write for specialist audiences, informal address is off limits": stored once in your personal system prompt, it applies to all your conversations – your profile as the default setting.
Personal system prompt
Your company's stored strategy context – positioning, audiences, no-gos – flows into every assistant on the team. The market calls this context engineering; for you it means: every assistant knows the company before the conversation begins.
Strategy prompt at organisation level
Onboarding guides, internal rules of play, FAQ on AI use: your organisation maintains its own pages directly in the account menu – bilingual, without an external wiki.
Tenant custom pages
The newsletter draft is there, but you are unsure about the opening. Instead of copy-paste into an email: you share the chat directly with your colleague. She sees the full history, continues writing with the assistant herself – and in the team chat beside it you align without confusing the AI. Four eyes, one document, no version chaos.
Conversation sharing with team chat
The notification bell shows: a colleague invited you into their chat. One click, and you see the complete conversation including all attachments and results – with read or write access, depending on what they gave you.
Sharing invitations with notifications
On the real-time whiteboard you collect moodboard images, claims, sketches, and post-its, see everyone else's cursors live – and bring the AI assistant directly onto the board when a text variant or visual is missing. Workshop atmosphere, even remotely.
Canvas: collaborative whiteboard with AI chat
The client sits in the canvas with you – and so do the AI assistants: you generate image worlds in real time, have text variants written alongside, lay everything out, discard, refine. People and assistants work as one team on the same board; every result lands there with one click. At the end you export the state as PNG or PDF in the desired format – decisions happen in the meeting, not three days later by email.
Canvas + AI assistants on the board + image generation + export
Before the big rebuild round on the board: create a snapshot. Version history keeps every state – you can experiment without fear of losing the good intermediate version.
Canvas: auto-save, snapshots, version history
One personal pinboard per project: the relevant chats as cards, the important files, notes, the responsible assistant. Monday morning you open the board and are back in context within seconds – instead of searching twenty chats for the right one.
Desktop: project pinboards
The generated campaign visual is good? One click on "Pin to desktop", and it sits on the right project board – beside the briefing, the notes, and the chat it came from.
Pin-to-desktop from tool results
Frames for project phases, status badges for "in approval" and "final", notes for interim states: the board structures itself as granularly or broadly as your project needs – with free zoom and pan.
Desktop: frames, badges, notes, pan & zoom
Pin what matters. Build groups by drag and drop ("Client A", "Internal projects"). Full-text search across all histories with jump to the hit. Your chat history becomes an archive instead of noise.
Chat organisation: pin, groups, search
The team's best prompts – curated, tagged, usable for everyone. New colleagues work at the level of the experienced from day one because prompt knowledge is shared.
Prompt library with management
From briefing to finished Staffbase post without switching tools: the assistant writes the post in the right tone for the channel, attaches the matching image – and publishes to the employee app after your approval.
Staffbase connector: create and publish posts
The sensitive post about reorganisation: before it goes live, AssistantOS simulates reactions – a multi-agent simulation plays through how different employee groups receive the text, where misunderstandings loom, and which formulations create resistance. The market talks about synthetic audiences – you test with them already.
Reaction simulation via MiroFish
Which Staffbase channel fits this topic? What ran there recently, what landed well? The assistant analyses your channel structure and existing posts and recommends placement and preparation.
Staffbase connector: channel analysis
Create a new page, update an existing one, archive old posts: content maintenance in Staffbase runs entirely through the chat – including image search and media upload.
Staffbase connector: pages, media, archiving
What has already been published internally on sustainability? The assistant searches Haiilo – blogs, wiki, timeline, events – and gives you the basis for the next article without you clicking through.
Haiilo connector
The same questions about travel expenses, IT, or benefits over and over: an assistant with internal guidelines in the knowledge base answers them around the clock – with source citations to the respective document.
Assistants + RAG knowledge base
The management update as a podcast instead of a PDF: in Podcast Studio you produce internal audio formats with professional voices, music, and mastering – communication that also reaches people on the commute.
Podcast Studio
Plan, approve, and publish posts across all channels – in integrated social media management, via single sign-on directly from AssistantOS. One workspace per organisation, no extra logins.
Mixpost integration
You are writing the LinkedIn post and the text is not landing? The floating assistant button brings AssistantOS directly into the editor: generate variants, shorten, sharpen – and adopt text and media into the post with one click.
Mixpost bridge in the post editor
The product advisor, the career guide, the service helper: assistants can be embedded as chat widgets on any website – in your branding, with starter questions, file upload, and usage analytics. GDPR-compliant without cookies, and only on domains you approve.
Embed widget with analytics
The AI-produced audio guide or company podcast gets a public web player – shareable by link, without recipients needing an account.
Audio stream service
Develop text, define recipients, send – the assistant dispatches the email via your connected Microsoft account. And because sending is final, it asks once more beforehand: you approve, it sends. Human in the loop, very practical.
Microsoft 365: email sending with approval step
Your content should rank not only on Google but also appear in answers from ChatGPT, Claude, and Gemini. AssistantOS structures texts accordingly – and with the Visibility Tracker you measure afterwards whether it worked.
Content structuring + GenAI Visibility Tracker
"What important mail came from client X this week?" The assistant searches your emails by sender, subject, and content and summarises – you make decisions instead of scrolling.
Microsoft 365: email search and summarisation
"Reply politely that we need to reschedule the meeting and suggest two alternatives." The assistant drafts the reply – and waits for your okay before it goes out. Reply-all too, if needed.
Microsoft 365: email replies with approval
"What is on next week?" – "Schedule a meeting with the project team for Thursday at 2 p.m. as an online meeting." Read calendar, create appointments, invite participants: all in the chat.
Microsoft 365: read calendar and create appointments
"Get the current version of the budget plan from OneDrive and tell me what changed compared with the previous version." The assistant finds the file, reads the content, and continues working with it – without an Explorer odyssey.
Microsoft 365: file search and document access
The spreadsheet in the attachment, the briefing as a text file: the assistant lists attachments, reads them, and makes them usable immediately – summarise, evaluate, process further.
Microsoft 365: read attachments
The old contract exists only as a scan, the business cards from the trade show as photos: OCR turns them into searchable text – in German, English, or both at once.
Autonomous toolbox: OCR via Tesseract
What is planned this week, what is waiting for approval? The assistant reads your Airtable bases – editorial plan, asset database, campaign list – and writes its own suggestions back into the inbox table on request.
Airtable connector
The Word as PDF, the Markdown as DOCX, the presentation as a handout: format conversion is a side sentence in the chat, not its own work step.
Autonomous toolbox: document conversion
The brochure is 48 MB, the mail limit is 10: AssistantOS compresses PDFs to sendable size – without you resorting to an online tool with unclear data handling.
Autonomous toolbox: PDF compression
Up to ten files at 50 MB per message: briefings, tables, images – upload, and the assistant works with them. It understands images in content too: "What is on this flipchart photo?"
File uploads + image vision
The publishing classic: did everything from the approved Word file really land in layout – and nothing get invented? AssistantOS compares the Word original with the InDesign export and lists every deviation. Proofreading in minutes instead of days.
Document comparison Word vs. InDesign HTML
Two PDF versions, thirty comments in between in a correction round: AssistantOS compares the versions and checks comment by comment whether the note was implemented. The approval email almost writes itself afterwards.
PDF comparison with annotation check
The EU directive against greenwashing (EmpCo) makes green claims delicate. AssistantOS checks texts and visual advertising material for problematic environmental statements – before legal does, or worse.
EmpCo compliance check incl. image review
For the compliance check or rebranding audit you need all images from a website at a glance: AssistantOS collects a page's image material systematically – and checks it further directly, for example for greenwashing signals or outdated logos.
Web scraper: image extraction + vision analysis
Send email, publish post, delete document: for everything that affects the outside world or is irreversible, the agent pauses and puts the action before you for approval – with all details. You decide, it executes. Autonomy with guardrails.
Human-in-the-loop approvals
Two interpretations possible? Instead of choosing the wrong one, the assistant asks a follow-up question – with clickable options or free text. Less waste, less "that is not what I meant".
Ask-user follow-ups in agent mode
Before anything leaves the building: Duden checking as a fixed step in the quality routine – on request as a standing order for every text the assistant produces.
Duden spell check
Who configured which assistant, who approved what? Security-relevant actions are logged – governance that makes compliance teams happy without anyone feeling it in daily work.
Audit logging
"Every Monday at 9: Summarise the week's competitor news." – "On the first of every month: Create the visibility report." You describe the job in plain language, AssistantOS runs it and delivers the results automatically in your chat. That is the core of what the market calls agentic AI: systems that work for you while you do something else.
Schedule Manager: time-based jobs in natural language
Which automations are running for me, when do they fire next? The central job overview in settings shows everything – pause, change, delete, or jump straight to the related chat.
Scheduled jobs: central management
"Remind me every Friday about the team week-in-review." Pure thinking and writing tasks can be scheduled too – the assistant delivers on time, you only need to send.
LLM schedules without plugin
Millions ask AI instead of search engines – but what do ChatGPT, Claude, Gemini, and Grok answer when asked about your industry? The GenAI Visibility Tracker asks defined questions systematically to the major models and measures whether, where, and how your brand appears – as a reliable visibility index over time. Generative engine optimisation is on everyone's lips; here it gets a measuring instrument for the first time.
GenAI Visibility Tracker with LTVI index
For which questions is the competitor named and you are not? Which topics "belong" to you in AI answers, which have you lost? The analytics tools answer that in the chat – blind spots, competitive comparisons, citation analyses, trends.
Visibility analytics: 14 analysis tools in chat
Gemini loves you, Claude does not know you? The model comparison shows how differently the major AI systems talk about you – and where targeted content work has the biggest leverage.
Visibility analytics: model comparison
Visibility is one thing – tone is another. The Reputation Tracker evaluates how AI models talk about your strategic topics: Is sustainability linked positively with you? Does the risk topic from 2023 still appear? With traffic-light logic, sentiment analysis, and evidence snippets for every rating.
GenAI Reputation Tracker
AI models suddenly link your brand to a topic that was not on your radar? The emergent-topics analysis finds such new topic clusters automatically – an early warning system for the topic landscape.
Reputation analytics: emergent topics
Visibility results as a public dashboard or CSV/JSON export: stakeholders see the numbers without needing an account.
VT export and public dashboard
Your skills and assistants are also available via MCP – the open standard some call "USB-C for AI" – in Claude Desktop, Copilot Studio, or Cursor. And via the webhook gateway, n8n, Make, or Zapier connect. Not an island, but infrastructure.
MCP server + webhook gateway
Once an IT project, now a coffee break: you drag twenty documents – style guide, FAQ, product info – into a collection, assign them to your assistant, done. From then on it answers with source citations from your own materials. No ticket, no vendor.
RAG collections in tenant admin: create, upload, assign
Which assistants does your team see? With which system prompt, which knowledge, which skills? You configure that yourself in the admin area – per assistant, without programming.
Tenant admin: assistant management
Creative copy with one model, analysis with another, sensitive content with an EU-hosted open-source model: via the central gateway, assistants can run on different models – model-agnostic, without vendor lock-in. Switching is a configuration decision, not a migration project.
LiteLLM gateway: model choice per assistant, incl. open-source models
Is the new model really better for your texts? Set up two assistants with identical prompts but different models and let the team compare – evidence-based model choice instead of gut feeling.
Configurable models per assistant
Positioning, audiences, language rules, no-gos: the Strategy Prompt stores organisational context centrally, and every assistant on the team works with it. Consistency by system setting.
Strategy prompt at tenant level
Invitation by email (bilingual), assign roles (Owner, Admin, User, Viewer), done. And whoever works across several organisations – agency classic – switches workspace with one click in the header.
Tenant invitations, roles, workspace switch
Canvas, Podcast Studio, sharing, voice, Visibility Tracker: every function can be switched on and off per organisation. You start focused and grow feature by feature – instead of overwhelming the team on day one.
Feature toggles per tenant
Token usage, cost trend, forecast: the billing dashboard shows what AI usage costs – per period, transparent, without surprises at month end.
Billing and usage dashboard
Colours, fonts, style direction for all infographics in the organisation – stored centrally. Every generated graphic looks like it came from one source, no matter who creates it.
Infographic theming per tenant
Two-factor authentication, session control, isolated tenant environments, encrypted API keys, hosting in Germany: the governance baseline is built in – not "shadow AI" sprawl, but a framework in which the team can experiment safely.
2FA, tenant isolation, GDPR-compliant EU hosting