↑↓ to moveEnter to openAll results →
    Polaris AI

    AI advisors and agents, governed like your people.

    Polaris reads the whole programme every cycle, checks the evidence behind every claim, tells leadership what needs a decision, and prepares the changes. It works with each person's permissions, goes through the same approvals they do, and leaves the same record.

    Polaris activityIllustrative
    • ExecuterWaiting for your approval

      Move $240M from paused sandbox tracks to the cloud and security programmes. 3 changes.

    • Dependency HunterFound a hidden link

      P16 must finish before P45 starts. If it stays unlinked, the committed date moves 23 working days.

    • Update extractionRead P14-field-report.pdf

      4 changes proposed: progress, a finish date, a new risk and the status note.

    • Insight minerFlagged

      4 projects report green with no new evidence in 5 weeks.

    • Chief of StaffBrief ready

      Needs attention, 55 out of 100. Three decisions owed this week.

    • Email status agentReplies received

      6 owners replied. Their updates are waiting for review.

    Advises

    Reads and recommends. Writes nothing to the plan.

    Proposes

    Drafts changes. A person reviews and applies them.

    Acts on approval

    Prepares the change, shows its effect, applies it when you approve.

    Watches and drafts

    Runs on a schedule or on request and writes drafts and findings.

    Seven advisors and a Chief of Staff, every cycle.

    Each advisor reads the programme from its own seat. The Chief of Staff reconciles them into the one brief leadership acts on.

    Polaris advisory board: the programme read, the top three moves and each advisor's verdict, score and moves
    Advisory board. The programme read and every advisor's verdict on one page.

    Each advisor writes its own brief.

    The financial advisor reads budgets, commitments and earned value; the risk advisor reads the register and the findings; each one ranks and prices its moves.

    • An executive summary with a score
    • Sections that state their own conclusion
    • Every move with an owner and a due window
    Financial advisor brief with its executive summary and score
    Financial advisor. The brief, with its score and summary.

    Agents find what the plan does not hold.

    The Dependency Hunter reads tasks, documents and risks, finds dependencies nobody declared, and prices each one in days of end-date impact.

    • Hidden links across projects
    • Impact computed by the engine
    • Confirm a link and it joins the plan
    Dependency Hunter: a hidden link between two projects and its impact on the committed date
    Hidden links. A dependency nobody declared, priced in days.

    Ask it a question, or hand it the work.

    Polaris answers from the live model, with the records behind every figure. Give it a field report and it proposes the plan updates; give it your template and it builds the board pack. Nothing changes until you approve it.

    Demo thread on sample portfolio data
    PolarisAI programme advisor

    How is the $2.3B portfolio split, and where is the risk concentrated?

    Polaris

    41 initiatives, $2.3B committed across six pillars. Two pillars hold half the money and most of the trouble.

    Committed value by pillar
    Pillar 1, Digital core: $620M, 27%Pillar 2, Infrastructure: $520M, 23%Pillar 3, Services: $410M, 18%Pillar 4, Capability: $330M, 14%Pillar 5, Regulatory: $250M, 11%Pillar 6, Enabling: $170M, 7%$2.3B
    Pillar 1, Digital core$620M27%
    Pillar 2, Infrastructure$520M23%
    Pillar 3, Services$410M18%
    Pillar 4, Capability$330M14%
    Pillar 5, Regulatory$250M11%
    Pillar 6, Enabling$170M7%
    Polaris, Chief Financial

    At ledger level, four initiatives carry the whole variance. P14 and P25 alone account for $87M of the $112M forecast breach.

    InitiativeForecastVarianceStatus
    P14 National Foundation Model$412M+$56M121 days late
    P25 National SOC$268M+$31M90 days late
    P44 Sovereign Cryptography$150M+$12M46 days late
    P90 National AI Standards$96M+$9M28 days late
    P07 Data Exchange Layer$74M-$4MOn plan
    PDFP14-field-report-week-45.pdf2 pages

    Update P14's status from this field report.

    Polaris

    I read the report. These are the changes I would make to P14. Nothing is saved until you approve them.

    P14 National Foundation ModelNowProposed
    Progress41%48%
    Model training, finish30 Sep14 Oct +14 days
    New riskNoneGPU delivery delayed, high
    Status noteLast week'sCluster commissioned; training starts late
    Polaris

    Applied to P14 and recorded in the audit log with you as the approver. The finish date moved 14 days, so I flagged P25, which depends on it.

    PBoard-template-2026.pptx

    Now build Thursday's board pack from this thread, in our template.

    Polaris

    Nine slides in your board template, built from this thread and the P14 update. The charts stay bound to their records, so the pack refreshes every cycle.

    Illustrative. A scripted thread on sample data. In your workspace Polaris works with your permissions, and every change it proposes waits for a person to approve it.

    Every agent, what it reads and what it produces.

    Grouped by what each one is allowed to do. Each can be switched off for a workspace.

    Advises

    Reads and recommends. Writes nothing to the plan.

    Seven chief advisors

    Advises

    Financial, strategy and planning, people, risk, performance, commercial and digital: each reads the programme from its own seat every cycle.

    Reads
    Schedule, budget, KPIs, risks, people
    Produces
    A brief with ranked, priced moves

    Chief of Staff

    Advises

    Reads all seven briefs, reconciles where they disagree, and writes the one page leadership acts on.

    Reads
    The seven chief briefs
    Produces
    Verdict, decisions owed, top moves

    PMO and department briefs

    Advises

    The same advisory read, scoped to one PMO or one department head.

    Reads
    That scope's projects and KPIs
    Produces
    A scoped brief
    Acts on approval

    Prepares the change, shows its effect, applies it when you approve.

    Executer

    Acts on approval

    Turns an advisor's recommendation into the actual changes, previews the knock-on effects, and applies them once you approve.

    Reads
    A recommendation and the live plan
    Produces
    Applied changes, recorded in the audit log

    Polaris chat

    Acts on approval

    Answers questions about the programme and, when you ask it to do something, plans the change, shows it and applies it on approval.

    Reads
    What you can see, and documents you attach
    Produces
    Answers, charts and approved changes
    Proposes

    Drafts changes. A person reviews and applies them.

    Scenario planner

    Proposes

    Builds alternative plans for a goal you set and compares them against the current plan.

    Reads
    The plan, a goal and constraints
    Produces
    Scenarios to compare and adopt

    Update extraction

    Proposes

    Reads a field report or weekly update and turns it into progress, dates, risks and resource gaps on the plan.

    Reads
    PDF, Word, Excel, PowerPoint or text
    Produces
    Proposed plan changes for review

    Email status agent

    Proposes

    Asks project owners for their update by email and turns the replies into proposed changes.

    Reads
    Email replies from owners
    Produces
    Proposed updates for approval

    Dependency Hunter

    Proposes

    Finds dependencies nobody declared across tasks, documents and risks, and prices each one in days.

    Reads
    Tasks, documents, risks
    Produces
    Links to confirm, with impact

    Demand proposer

    Proposes

    Proposes the roles and effort a project will need from its plan.

    Reads
    The project plan
    Produces
    Role demand to approve

    Minutes importer

    Proposes

    Reads meeting minutes and extracts the decisions and actions, linked to the plan.

    Reads
    Minutes documents
    Produces
    Decisions and actions for review

    Import analysers

    Proposes

    Maps Microsoft Project, Primavera and Excel files onto the model when you bring a plan in.

    Reads
    Schedule and spreadsheet files
    Produces
    A mapped plan to confirm

    Workflow builder

    Proposes

    Drafts automation workflows from a description, for an administrator to apply.

    Reads
    A description of the workflow
    Produces
    A workflow to review
    Watches and drafts

    Runs on a schedule or on request and writes drafts and findings.

    Insight miners

    Watches and drafts

    Run on a schedule looking for patterns: stalled work, burn ahead of progress, greens without evidence.

    Reads
    The whole programme
    Produces
    Findings in the AI findings list

    Cascade detector

    Watches and drafts

    Watches the programme history for slips that are spreading from one project to the next.

    Reads
    Change history
    Produces
    Cascade warnings

    Evidence assessment

    Watches and drafts

    Checks whether the evidence attached to an approval supports the change, for the approver.

    Reads
    The request and its evidence
    Produces
    An assessment for the approver

    Report Studio agent

    Watches and drafts

    Outlines and reviews custom reports and builds board packs from a conversation or a comparison.

    Reads
    The live model, your template
    Produces
    Reports and board packs

    Key takeaways

    Watches and drafts

    Drafts the key takeaways for an initiative or pillar for its owner to edit.

    Reads
    The initiative or pillar
    Produces
    A draft for the owner

    Meeting minutes drafter

    Watches and drafts

    Drafts the minutes of a forum meeting for the secretary to review.

    Reads
    The meeting record
    Produces
    Draft minutes

    Translation

    Watches and drafts

    Translates stored briefs and reports so leadership can read them in Arabic.

    Reads
    Briefs and reports
    Produces
    Arabic versions, cached

    How a change happens.

    The same path for every agent that touches the plan, and the same checks a person's own edit goes through.

    1RecommendAn advisor or agent proposes a specific move.
    2SimulateThe engine works out the knock-on effects on a copy.
    3ApproveA person with the right permission approves or rejects it.
    4Apply and recordThe change is applied and kept in a write-once decision log.
    5MeasureWeeks later the engine checks whether the move worked.
    Your permissions, no moreAn agent sees and changes only what the person using it may see and change.
    Nothing applied unseenEvery change is shown with its effect before anyone approves it.
    A record of every moveWho proposed it, who approved it, and what it changed.
    Switch any agent offEach agent can be turned off for a workspace.

    Questions

    Can Polaris change our data on its own?

    No. Agents that act prepare the change and show its effect; it is applied only when a person with the right permission approves it.

    Does Polaris see everything?

    It sees what the person using it may see. The advisors that brief leadership read at the scope that leader is allowed.

    Which AI model does it use?

    Polaris works with the model your organisation chooses. Your administrator configures the provider.

    Can we use it from Claude or Microsoft Copilot?

    Yes. Lodestar has an MCP server, so an assistant that supports MCP can ask it about the portfolio, within each person's permissions.

    Can we turn agents off?

    Yes. Each agent can be switched off for a workspace by an administrator.

    See it on your own programme.

    Send us a slice of your portfolio and we load it before the session, from strategy down to task.