Workspace
Your AI Assistant
Meet your personal AI companion — always available, always learning, and deeply connected to your workspace. Not a chatbot bolted onto the side of a productivity tool, but a partner that understands your work, remembers your conversations, and can take real action.
Everything below is illustrative. Connectors, retention, and approvals follow your organisation’s rules.
Capabilities
What the assistant can do
Six pillars cover everyday work — from open-ended chat to scheduled automations that run on rails while you sleep.
- Manual
Chat about anything
Writing, research, analysis, brainstorming, problem-solving — ask whatever you need and get context-aware answers grounded in your work.
- Semi-auto
Remember your conversations
Carries forward what you have discussed across sessions. No re-explaining context every time you come back; the assistant builds on what it has learned.
- Manual
Search your knowledge base
Connect documents and shared drives so the assistant pulls accurate answers from your team’s actual content — with citations back to the source.
- Manual
Manage files
Upload documents, reference them in conversation, and let the assistant work with your files directly — read, summarise, extract, draft.
- Automatic
Run on a schedule
Set up recurring tasks (cron jobs) so the assistant handles routine work automatically, even when you are not around — and only pings you on exceptions.
- Manual
Reason through complex questions
Watch the assistant work through multi-part problems with extended thinking — visible reasoning that auto-collapses to a clean answer.
A companion that remembers
Two sessions, one assistant. The second conversation picks up exactly where you left off — without re-explaining anything.
Opisense Assistant
Personal session · long-term memory active
Memory is selective by design — preferences and decisions persist; throwaway chatter does not.
How it works
Three sources of intelligence, one grounded answer
The assistant blends what you have said, what you have shared, and what you are doing right now into a response that actually fits.
Your conversation history
Past discussions, preferences, and decisions — extracted into long-term memory and referenced when relevant.
Your knowledge base
Documents and files you have uploaded — searchable with citations back to the original passage.
Your workspace context
What you are working on right now — projects, calendar, open threads, and the modules you have permission to see.
Grounded response
A reply that uses your voice, cites your sources, and respects what you are doing in the moment.
Retrieval and actions inherit your permissions — the assistant cannot improvise access beyond what you grant.
Knowledge base
Answers grounded in your team’s actual content
Connect vaults, shared drives, and folders. The assistant searches across them, quotes the passages it used, and links back to the source so you can verify in one click.
- Inline citations next to every claim — no opaque summarisation.
- Permissions-aware retrieval: the assistant cannot read what your role cannot.
- Snippet previews show why a source was selected before you open it.
Opisense Assistant
Knowledge base · 3 sources cited
Opisense Assistant
Pulled from procurement and logistics vaults:
- The master agreement allows a 14-day force-majeure window before penalty clauses activate. [1]
- The Q1 review already flagged this carrier for renegotiation, so the conversation is not new. [2]
- On-time delivery is 88% versus a 95% target, with freight variance +2.1% — material but not critical yet. [3]
Sources
- [1]
Supplier-Master-Agreement-2026.pdf
PDF · 4.1 MB · Procurement vault
Section 7.2 — Force-majeure window of 14 days for transit disruptions, with mandatory weekly status updates from the supplier.
- [2]
Q1-Procurement-Review.docx
Doc · 612 KB · Finance shared
Vendor scorecard places Lot 14 carrier at 78/100 — flagged for renegotiation by end of Q1.
- [3]
Carrier-Performance-Q1.xlsx
Sheet · 1.3 MB · Logistics warehouse
On-time delivery at 88% (target 95%); freight cost variance +2.1% versus committed rate card.
Citations link to the exact passage retrieved — never to vague summaries.
File management
Upload, reference, and let the assistant work with your files
Drop documents into the conversation. The assistant reads them, holds them in context, and helps you draft, extract, and summarise without leaving the thread.
- Reference any uploaded file as a chip in the composer.
- PDFs, Office docs, sheets, slides, and images are all in scope.
- Files stay scoped to your workspace — the assistant respects retention rules.
Opisense Assistant
Files · referenced in conversation
Your files
Upload- Referenced
Supplier-Master-Agreement-2026.pdf
PDF · 4.1 MB · Today
Q1-Procurement-Review.docx
Doc · 612 KB · 2 days ago
Carrier-Performance-Q1.xlsx
Sheet · 1.3 MB · Yesterday
Board-Pack-Draft.pptx
Slides · 7.8 MB · Today
Floor-plan-line-B.png
Image · 980 KB · Last week
Opisense Assistant
- Read 84 pages · 12 minutes saved versus reading manually.
- Eight penalty clauses extracted, grouped by trigger (delivery, quality, confidentiality).
- Suggested next step: draft a renegotiation memo using clauses 4.1, 7.2, and 9.3.
Uploaded files inherit your retention policy — nothing leaks beyond your workspace.
Scheduled runs
Recurring tasks that run on rails
Set the cadence once. The assistant takes care of routine briefings, monitors, and recaps in the background — and only surfaces something when there is an exception worth your time.
- Standard cron expressions plus human-readable cadence labels.
- Per-job enable / pause toggles with audit trail.
- Output preview docks under each job so you always see what last shipped.
Opisense Assistant
Scheduled runs · scoped to your role
Your scheduled jobs
cronLatest output
Weekdays · 07:00 Europe/OsloBrief preview · Today
- Throughput recovered to 98.4% of plan (Line B back online).
- Two CAPAs closed overnight; one new ticket on Lot 14 carrier.
- Decision needed: alternate supplier sign-off before noon.
Reads overnight signals, drafts a one-page narrative, and posts it to your inbox before 07:30.
Scheduled runs inherit identity and retention from your workspace — no shadow automations.
Extended thinking
Watch the assistant reason through hard questions
When a problem needs working through, the assistant’s reasoning streams into a collapsible panel above the answer. Expand to follow along — collapse to keep the thread clean.
How it appears
- A brain icon with a shimmer animation and a "Thinking…" label appears above the response.
- The panel auto-expands so you can follow reasoning as it streams in.
- When done, the label switches to "Thought for X seconds" and the panel auto-collapses after about a second.
Interacting with thinking
Click the panel to expand or collapse it at any time. If you manually expand it while the assistant is still thinking, it stays open — the auto-collapse will not override your choice.
When it activates
Complex questions
Multi-part asks that require weighing several factors before answering.
Analysis and reasoning
Comparing options, evaluating trade-offs, or working through logic step by step.
Nuanced topics
Questions where the answer depends on context and several angles need to be considered.
Opisense Assistant
Extended thinking · live preview
Extended thinking does not appear on every response — only when the question needs it.
Memory
How your assistant remembers — and what you control
Memory is the difference between starting from scratch every time and a companion that actually knows you. Two layers, both governed.
Session memory
Everything in the current thread
The conversation you are having right now — every message, every follow-up, every file you have shared. Resets when you start a new conversation.
Long-term memory
Persists across sessions
Important details extracted from past chats — preferences, recurring topics, key decisions — so future conversations build on what came before.
What it retains
Preferences
How you like things formatted, your communication style, your go-to tools.
Key decisions
Choices you have made that might come up again.
Recurring context
Projects you mention often, people you work with, topics you care about.
Explicit instructions
Things you have directly told it to remember.
What it forgets
- Casual small talk and one-off exchanges.
- Throwaway questions with no signal value.
- Anything that does not match a useful, recurring pattern.
Managing your memory
- View memory — see what has been stored about you in assistant settings.
- Forget a single item — remove an extracted detail without affecting the rest.
- Clear long-term memory — start the assistant’s view of you fresh whenever you want.
Privacy
Your memory is private to you. No one else on your team can see what your assistant remembers about you — each person has their own separate memory space.
Memory
Preferences
Tone, format, and tooling defaults.
Concise board summaries
Three slides max, decision ask at the end, executive voice.
Inferred · 2 weeks ago
Prefers Markdown over rich text
Drafts default to bullets and short paragraphs.
Explicit · last month
Key decisions
Choices that shape future answers.
Phased carrier move
15% → 25% → 40% across two quarters; contractual notice filed first.
Inferred · this week
Recurring context
Projects, people, and topics you care about.
Project: Supplier diversification
Owner: you. Stakeholders: procurement, finance, ops.
Recurring · 12 mentions
Topic: Lot 14 carrier risk
Discussed in five threads since last quarter.
Recurring · 5 threads
Explicit instructions
Things you told the assistant to remember.
Always cite sources for legal claims
No exceptions — even on quick answers.
Explicit · 6 weeks ago
Clear long-term memory
This does not delete your conversations — only the extracted memory.
Only you can see what the assistant remembers about you.
Quick start
Up and running in four steps
You do not need a rollout plan to feel the value — start a thread, share a file, and let the assistant learn from there.
- 1
Open the assistant
Click Assistant in the sidebar or use the keyboard shortcut. A clean conversation is ready to go.
- 2
Send your first message
Try "What can you help me with?" or "Help me brainstorm next quarter’s priorities."
- 3
Explore what it can do
Upload a file, ask follow-ups, or try "Summarise the key points from my uploaded document."
- 4
Customise your experience
In Settings, tune personality, connect knowledge sources, schedule jobs, and review memory.
Privacy and trust
Built for organisations that take governance seriously
Every conversation, file, memory, and scheduled run inherits your workspace identity and policy — so manual exploration and automation defend equally well.
Private by default
Your memory and conversations are not shared with the rest of your organisation — each person has their own scoped space.
Permissions travel with retrieval
Knowledge base search and file references obey the same access rules as the underlying systems — no silent escalation.
Auditable runs
Scheduled jobs and assistant actions are logged with structured traces back to inputs, models, and policies.
Let's find the right solution for you
Start with the pilot program and see real usage — then finalize scope and commercials with our team.