AI Audit

Find where your team's hours actually go.

We interview your team, map every repetitive task and hand back a ranked plan for what to build first.

$2,500 flat · 7 to 14 days · credited in full to your build if you go ahead within 30 days of your report

Book an audit
A pocket compass, printed in teal and black

Before and after

From a list of ideas to a ranked plan.

  1. 01

    You know where the hours go

    Nine interviews turn a list of guesses into opportunities ranked by the hours each one gives back.

    Show You know where the hours go

    12 ideasand not one with an owner or a number

    37 hoursa week back, ranked across four opportunities

    Microsoft Teams

    PPartnersChatFiles

    SO

    Sam Ortiz9:05 am

    We should do something with AI

    PN

    Priya Nair9:20 am

    Agreed. Where do we even start?

    JW

    Jo Walsh9:31 am

    I started a list from the offsite

    AI ideas.docx

    Type a message

    AI ideas.docx

    FileHomeInsertLayoutReviewView

    AI ideas (offsite)

    Chatbot on the website?

    Something for proposals

    Automate the inbox somehow

    AI for client reports

    Lee: onboarding?

    Plus 7 more from the whiteboard

    portal.thehourglass.ai/northstar

    Your AI audit

    Northstar Advisory · 9 interviews · report ready 11 Nov

    RankOpportunityHours back a week
    1Proposal drafting14
    2Inbox triage11
    3Weekly client report8
    4New starter onboarding4
    Proposal drafting · from the interviews

    Lee Chen5 hrs

    Jo Walsh5 hrs

    Sam Ortiz4 hrs

    Total14 hrs a week

    Every number traced to an interview

  2. 02

    You get a plan you can act on

    Instead of sitting through vendor demos, you choose the first builds from a roadmap with a range on each.

    Show You get a plan you can act on

    9 demosbooked, and no way to tell which one fits

    3 buildsin order, each with a range, chosen on the roadmap call

    Inbox · Sam Ortiz · Outlook
    • AI agent vendor8:40 am

      Your personalised demo is confirmed

      See how agents can transform your whole firm.

    • Chatbot vendor8:12 am

      Following up: 30-day free trial

      Most accounting firms are live in a day.

    • Automation webinarYesterday

      Live today: AI for professional services

      Save your seat for 1 pm.

    • AI writing toolYesterday

      Your team licences are about to expire

      Upgrade now to keep 20% off.

    Microsoft Teams

    PNPriya NairChatFiles

    PN

    Priya Nair2:10 pm

    Which of these demos do we actually need to sit through?

    Type a message

    Northstar AI roadmap.docx

    FileHomeInsertLayoutReviewView

    Northstar AI roadmap

    1. Proposal drafting, skills platform: $8,000 to $15,000

    2. Inbox triage, AI Inbox: $7,000 to $12,000

    3. Weekly client report, company brain: $15,000 to $25,000

    Each build takes about 2 weeks per build.

    The $2,500 audit fee comes off build 1 if we go ahead within 30 days of the report.

    Roadmap call Thursday: Sam and Priya choose build 1.

Figure: a staged example. Northstar Advisory, its people, clients and numbers are made up.

What you get

  1. 01

    Interviews

    Up to 10 interviews, 45 minutes each on Zoom, across every team that touches the work. Everything recorded and transcribed.

  2. 02

    Your AI report

    Every opportunity scoped by input, process, output and what it could become (IPO STAR), ranked by impact and build effort, with build pricing as a range. An ROI summary covers hours saved, annual savings and build costs. A bonus cybersecurity audit reviews data governance, prompt injection exposure and vendor risk.

  3. 03

    A roadmap call

    We walk through the Opportunity Matrix and phased roadmap together, confirm scope and sequence, and choose what to build first.

Four steps from kickoff to roadmap.

  1. Step 1

    Scope

    Agree who we interview with the person who will decide what gets built.

  2. Step 2

    Interview

    Up to 10 interviews, 45 minutes each on Zoom. We ask about handoffs, repetition, judgement calls and frustrations.

  3. Step 3

    Report

    Pain points and opportunities, mapped and ranked, with a range for each build. Your roadmap lands in 7 to 14 days.

  4. Step 4

    Roadmap

    Choose the first builds. Each build takes about 2 weeks per build.

Results across our clients

  • 8 hrs

    A week back from repetitive weekly reporting, reported in the client feedback form

    Integr8

  • 5 days to 2

    From credit application to approval, for a new customer

    Heaps Normal

Would have in the old days, pre-Hourglass, pre-AI, spent 1 to 2 weeks. Today we're spending half a day.

Phil Morle, Partner, Main Sequence

Data sovereignty

Your data never leaves your business.

Everything we build runs on your own servers and your own accounts. We never store your data, you own all of the IP, and if we disappeared tomorrow it would all keep running. It is model-agnostic too: Claude today, whatever wins tomorrow.

  • Runs on your servers and accounts
  • You own all of the IP
  • We never store your data
  • Point any AI at it: never locked in

The questions every discovery call asks.

What do I actually get?

We interview your team (45 minutes each), then deliver a ranked AI roadmap in 7 to 14 days: every opportunity we found, what it saves, what it costs to build, and the order to build it in.

How much does it cost?

The audit is a flat $2,500 for up to 10 interviews. Go ahead with the build within 30 days of your report and the full $2,500 comes off it. Builds, the Company AI Brain included, are $5,000 to $25,000 each, quoted as a range and confirmed after a scoping call. Maintenance is $500 to $1,500 per month per build. No surprises: you will know the cost of every step before you commit to it.

Where does our data live? Is it secure?

On your servers and in your accounts, full stop. We adopt your existing privacy settings, we never store your data, and nothing leaves your organisation.

Are we locked into Claude (or you)?

No. Everything we build is model-agnostic: point Claude, OpenAI or Gemini at it and swap any time without a rebuild. You own the IP outright.

Cannot my team just do this with ChatGPT?

That is where most teams start, and where most teams stall: individual workflows that break, cannot be shared, and depend on one person. The audit finds what is worth building properly, on a foundation the whole company can use.

Will it work with our existing tools?

Yes. We build on top of whatever you already run: Xero, SharePoint, HubSpot, Pipedrive, your ERP. No rip-and-replace, no data migration project.

Who owns it, and what if you disappear?

You own everything. The brain and every agent live on your infrastructure and keep running without us.

What does maintenance actually cover?

It never breaks: fixes, model upgrades as AI improves, onboarding new staff, and new capabilities we roll out across clients.

Is our business the right size?

Best fit is 10 to 100 people with no internal engineering team. If that is you, the audit pays for itself in the first build.

Will my staff worry about their jobs?

We design for the opposite: AI pre-drafts and prepares, your people review and decide. The question we ask every person we interview is how to make their job more enjoyable, not how to replace it.

Are AI usage costs included?

You own your own AI accounts, so usage costs go to you directly, and flat team subscriptions keep them predictable. Our fees are fixed per build.