Northstar BI sent a 12% renewal uplift. Keep Finance, Procurement, and Legal in one thread, and do not let anything go to the vendor without approval.
Northstar BI renewal packet.pdf
Product
AI work belongs in a
AI work belongs in a
shared workspace.
SoftworkerAI turns requests, approvals, evidence, and agent activity into visible, governed business execution.
Product demo
A governed workspace, not another chatbot window.
Start with a business request. Softworker routes the right agents, keeps policy checks visible, pauses for approval, and leaves evidence behind for every action.
What are you working on?
Start a governed thread. Nothing leaves Finance unless policy allows it.
Workspace / Vendor Evaluation
Northstar BI Renewal Owner Brief
Medium-risk renewal detected. I routed the run to Vendor Evaluation and invited Finance, Legal, and Procurement reviewers.
Finance Review joined Legal Review joined
Usage supports 88% of current seats, but the quote adds a 12% uplift. Ask for a concession or a seat-level cap before renewal.
Usage summary Pricing schedule Budget policy
The renewal notice date is valid. External outreach should include the updated security clause before anything is sent.
Legal clause delta
Approved to draft the owner brief. Keep the vendor response private until Procurement confirms the concession target.
Owner brief is ready. Approval is required before vendor outreach.
Medium risk 7 events logged
Route Research Review Approval
Running 19/25
Waiting 22/15
Done 15
1 Needs Attention 1 Running Now 4 Active 1 Failed
Schedule Scope Trigger Next Run Last Result
Agents
Browse governed agents, their operating volume, and the approval patterns around them.
Slack Anthropic
Connected
MCP Server Low Risk
2 agents 2m ago
Google Drive Google
Connected
Connector Medium Risk
1 agent 18m ago
GitHub Anthropic
Connected
MCP Server Medium Risk
1 agent 1h ago
Salesforce Salesforce
Connected
Connector High Risk
1 agent 3d ago
Browser Use Softworker.ai
Connected
Browser Use High Risk
2 agents 34m ago
PostgreSQL Internal
Connected
Internal Service High Risk
1 agent 5m ago
Figma Anthropic
Connected
MCP Server Low Risk
1 agent 12m ago
Gmail Google
Connected
Connector Medium Risk
2 agents 45m ago
Stripe Stripe
Connected
Connector High Risk
1 agent 2h ago
All Knowledge 50 Sources 6 Folders 40 Files 5 Links 5 Connectors
30 Verified
2 Syncing
18 Review
Organization 9 Sources
4
4
1
My Desk 2 Sources
2
Workspaces 29 Sources
12
7
4
Approvals
Authorize sensitive, consequential, external, or policy-gated agent actions.
Share the renewal recommendation with the procurement owner Medium Risk Pending
Contract notice date, pricing schedule, utilization trend, and draft recommendation.
Authorize vendor disbursement of $75,000 to Snowflake Services High Risk Pending
Statement of work, manager approval, and payment threshold policy.
Waive receipt requirement for executive travel lodging Medium Risk Pending
Booking confirmation, statement of charge, and policy exception record.
Audit
Search, inspect, and preserve business-readable records for every run and approval.
Selected event
Approval requested
Workspace policy required approval before the renewal recommendation could be posted back to the vendor evaluation thread.
AI Spend Over Time
Budget Status
All Scopes Budget $23 / $500
Default Model $23 / $100
Workspace Global $9 / $100
Workspace Global $7 / $100
Cost Breakdown
Name Cost Share Runs Input Tokens Output Tokens Tool Calls Retries Failed Avg / Run
Cost figures are estimates based on token usage and provider pricing. Actual billed amounts may vary. View audit log →
Governance
Configure Members, Groups, Roles, Workspace Policy, Model Providers, Model Profiles, Model Policy, Provider Health, Knowledge Access, Approval Rules, Identity, and Audit behavior for the Finance Workspace.
Members
Manage member lifecycle, role assignments, and external collaboration boundaries.
Member Status Organization Role Workspace Role Groups Actions
Pending Invites
No Pending Invites
External Collaborators
Northstar Vendor
To workstream-vendor-evaluation, thread-vendor-renewal, output-vendor-brief ActivePersonal and Shared AI Inbox
SoftworkerAI centralizes business requests from Slack, Microsoft Teams, email, forms, webhooks, integrations, and direct workspace instructions. Every request arrives in one place, helping teams capture important work, assign the right AI agent or human owner, and avoid missed follow-ups.
Workspace
Inbox
Collaborative AI Workflow Execution
SoftworkerAI gives every piece of work a clear execution path: understand the request, route it, run the right workflow, review the result, and approve sensitive actions. This makes AI-powered work visible, controlled, and ready for real business operations.
Intent Cards and AI Triage
SoftworkerAI analyzes each request or instruction and routes it to the right Workstream, Agent, and Skill.
Collaborative Work Threads
Every task becomes a shared thread for AI agents, team members, reviewers, and approvers.
Run Trace, Evidence, and Audit Trail
SoftworkerAI shows the steps, evidence, artifacts, blockers, retries, and failures behind every AI agent run.
Human Review and Approval
Sensitive, risky, legal, financial, or external actions can pause for human approval before they are completed.
Workspace
Northstar BI Renewal Owner Brief
Marcus Lee
Northstar BI sent a 12% renewal uplift. Keep Finance, Procurement, and Legal in one thread.
Renewal Research Agent
Owner brief is ready. Approval is required before vendor outreach.
Medium risk 7 events logged
AI Kanban Task Boards
Teams can view AI-assisted work in simple stages such as Running, Waiting, and Done. Leaders can quickly see what AI agents and people are working on across departments, workstreams, and business processes.
Scheduled AI Tasks
SoftworkerAI can run recurring work such as reports, reviews, checks, follow-ups, reconciliations, and alerts. Routine business processes keep moving without constant reminders, manual chasing, or repeated prompting.
Governed AI Agents
Teams can create AI agents with clear roles, tools, permissions, memory, Skills, and approval rules. This helps enterprises scale AI adoption while maintaining governance, security, and operational control.
Extensions and Connected Tools
SoftworkerAI connects with the tools your teams already use so governed agents can retrieve context, update systems, and complete work inside approved boundaries. AI agents can work inside your existing stack while teams keep every connection governed, observable, and risk-scored.
Existing tool connections
Work can move through the systems teams already rely on instead of forcing another migration.
Governed access
Admins can keep plugin access scoped by workspace, agent role, risk level, and review policy.
Extensions
Knowledge Hub for Business Context
SoftworkerAI gives AI agents and teams controlled access to company documents, policies, SOPs, process notes, and operational knowledge. Agents can work with the business context they need while respecting access controls and approved knowledge sources.
Bring Your Own LLM Provider
SoftworkerAI lets teams use their preferred large language model provider and apply model policies across workflows. Enterprises keep flexibility, cost control, vendor choice, and governance over how AI models are used.
Enterprise AI Governance and Audit
SoftworkerAI supports role-based access control, permissions, model policies, approval rules, audit logs, and output history. This makes AI adoption more suitable for enterprise standards, compliance needs, and internal risk management.
AI ROI and Impact Dashboard
SoftworkerAI tracks run volume, cost, turnaround time, approval rate, rework, and business impact. Leaders can see where AI agents are creating value, which workflows are improving, and where additional governance or optimization is needed.
Glossary
Key terms in governed AI work
Plain definitions for the core concepts behind SoftworkerAI.
- Governed AI agent
- An AI agent that operates under a defined identity, scoped access, human approval gates, live supervision, policy, and audit, so it can execute real work within boundaries you control.
- AI work execution
- Moving beyond prompts and suggestions to having AI agents complete approved work across your systems, with governance built into the workflow rather than added afterward.
- Approval gate
- A mandatory human checkpoint that pauses an agent before a sensitive or irreversible action, such as sending a contract or triggering a payment, until a person approves, rejects, or redirects it.
- AI audit trail
- A complete record of an agent’s goal, plan, approvals, actions, and outcomes, captured as the work runs so every action is explainable and reviewable.