The work AI can take on.

Most companies run on the same back office: invoices, inboxes, documents, schedules, reports and follow-up. Here is each kind of work AI can take on, the tasks inside it, and what a person keeps. Where it already runs in one of our own companies, we say so. Work you already outsource moves first: the budget line exists and you already buy an outcome.

Volume
The same work, many times a day.
Rules
A checklist or a policy decides most cases.
Digital inputs
It arrives as email, PDFs or rows in a system.
Checkable
The answer can be checked against a source.
A gate
A mistake can be caught before it does harm.

Level one

AI does it, your people check

Work that repeats all day, follows clear rules, and can be checked against a source. AI does it. Your people handle anything unusual.

  • Paying bills (accounts payable)

    • Bills read as they arrive
    • Each bill matched to its purchase order and delivery
    • Vendor onboarding
    • Payments lined up for approval

    A person stillApproving payments and settling vendor disputes.

    In our companiesNot yet running in our own companies.
  • Billing and collecting (receivables)

    • Invoices built from hours or usage
    • Payment reminders
    • Collections follow-up
    • Payments matched to the right invoice
    • Claims coded and submitted to payers

    A person stillWrite-offs and customer escalations.

    In our companiesBuildingReceivables and agreement billing, each line behind a gate verdict.
  • Bookkeeping and month-end close

    • Every transaction put in the right account
    • Bank and card reconciliations
    • Month-end checklist
    • Month-end adjusting entries drafted

    A person stillSigning off the close.

    In our companiesBuildingAn accounting platform: US GAAP close, a reconciliation queue, period locks.
  • Paperwork and data entry

    • Documents classified and read
    • Fields keyed into the ERP, CRM or loan system
    • Files renamed and filed

    A person stillAnything the agent is not sure of.

    In our companiesRunningNew leads become applications in the property system. AI that reads and files documents is in pilot.
  • Inbox and customer questions

    • Requests sorted and routed
    • Replies drafted
    • Status updates sent
    • Common questions answered

    A person stillAnything off script or sensitive.

    In our companiesNot yet running in our own companies.
  • Scheduling and dispatch

    • Bookings and reminders
    • Rescheduling
    • Crews and drivers assigned
    • Customers updated as work moves

    A person stillConflicts and priorities.

    In our companiesBuildingAssigning cleaners to jobs, the first part of a back office moving onto AI.
  • Weekly and monthly reports

    • Weekly and monthly packs
    • A plain summary of what moved and why
    • Client reports

    A person stillWhat the numbers mean for a decision.

    In our companiesNot yet running in our own companies.
  • Research and keeping watch

    • Market, regulatory and competitor watch
    • Contact and company records filled in
    • Daily briefs

    A person stillChoosing what to act on.

    In our companiesRunningFilings and news read daily, scored, and published as a morning report.

Level two

AI does it, a person signs off

AI does the whole job and stops before anything goes out. A person reads it and signs off.

  • Finding and following up with customers

    • Likely customers found and ranked
    • Outreach drafted per account
    • Inbound answered in minutes
    • Customer records kept up to date
    • Renewals flagged

    A person stillEvery send passes an approval queue. People run the closing conversation.

    In our companiesRunningTargeting engines: who to call, why now, who decides.
  • Marketing

    • Articles drafted, every claim backed by a source
    • Pages that help people find you on Google
    • Campaign reports
    • Social posts scheduled

    A person stillBrand voice and the final approval.

    In our companiesRunningA content engine whose gate blocks unsourced claims before anything publishes.
  • Recruiting and onboarding

    • Candidates sourced
    • Applicants screened against the role
    • Interviews scheduled
    • Onboarding paperwork run

    A person stillThe hiring decision.

    In our companiesNot yet running in our own companies.
  • Compliance paperwork

    • Audit evidence collected on a schedule
    • Policies checked
    • Identity and documents verified
    • Licences and renewals tracked
    • Sales and income tax filings prepared across states

    A person stillAttestation and every judgment call.

    In our companiesNot yet running in our own companies.
  • Buying from suppliers

    • Quote requests sent to suppliers
    • Quotes compared
    • Purchase orders raised
    • Invoices checked against negotiated terms to recover overpayments

    A person stillNegotiation and signature.

    In our companiesBuildingVendor discovery: listings found, validated and enriched before outreach.
  • IT support and software

    • Updates installed and systems watched
    • User accounts provisioned and removed
    • Access requests
    • Security alerts reviewed first
    • Update tickets closed
    • Software changes built, tested and sent for review

    A person stillCode review and incident command.

    In our companiesRunningA request comes in, the change is built and tested against the live system, and a person reviews it.

Level three

Your people lead, AI helps

Decisions that need judgment, trust or accountability stay with people. AI gets the information ready. We say this out loud because it is the work people want to keep, and the work a company is trusted for.
  • Strategy and pricing
  • Negotiation
  • Relationship selling
  • Hiring and letting people go
  • Clinical and legal judgment
  • Crisis response

“In our companies” reuses the audited statuses on the work board, Q3 2026. Running: in use today. Pilot: proven in a first run. Building: not yet relied on.

Pick the function that costs you most.

Tell us which of these your team spends the most hours on. The assessment measures it as it runs today, then shows what handing it to AI would save.

info@16x9.ai