Skip to main content

AI operating model

For mid-market companies and the PE firms that back them

From a few weeks

We map how your company runs, process by process, and redesign the workflows where AI saves the most time or brings in more revenue. Each redesign comes with its EBITDA impact, worked out from your own volumes and costs, so you know what to build first.

How we build your AI operating model

Leadership sets the direction, your people show us how they operate, and we find and quantify the AI opportunities for you.

  1. Understand

    1. Align on priorities

      Who takes part: Leadership

      What happens: One-to-one interviews with your leadership team, each prepared with a short questionnaire.

      OutcomeYour strategic priorities for AI, the guardrails, and the scorecard we use to judge every opportunity.

    2. Hear your teams

      Who takes part: Your teams

      What happens: A questionnaire for every team, then interviews with department heads and the people behind each process.

      OutcomeWhere time goes, what slows people down, and their ambitions, in their words.

  2. Redesign

    1. Map how you operate

      Who takes part: Our team, checked with your teams

      What happens: We map who does what, in which system, and where work is manual, waiting or done twice.

      OutcomeYour processes today, step by step, and where the time is lost.

    2. Redesign and find the opportunities

      Who takes part: Our team

      What happens: We redesign each process around AI and value every opportunity it opens, in hours, costs and revenue.

      OutcomeYour processes redesigned, and a scored list of opportunities, each with its EBITDA impact and the assumptions behind it.

  3. Decide

    1. Prioritise with leadership

      Who takes part: Leadership

      What happens: A working session to choose what comes first: quick wins and more strategic bets.

      OutcomeA shortlist your leadership agrees on.

    2. Design the architecture and the roadmap

      Who takes part: Leadership and our team

      What happens: We design the target architecture on the systems you already run, and sequence the shortlist by dependencies, with an owner for each item.

      OutcomeA target architecture and a roadmap to act on.

What you leave with

  • Your processes mapped and redesigned around AI

    Each process today and redesigned, step by step: removed, automated, or done by a person with AI.

  • Every opportunity valued, with its EBITDA impact

    Built up from your own volumes and costs, with every assumption shown.

    Top prioritiesImpactFeasibility
  • A target architecture and a roadmap to act on

    What to build first, on which systems, and who owns it.

    NowNextLaterQuick winsReporting agentBuild connectors to data sourcesScale across teamsContinuous coaching

What clients say

An example of what steps 3 and 4 produce for one process.

Client onboarding redesigned: from about 6 months to about 3 weeks

We redesigned client onboarding for a company's B2B programme. Before, each new client had to be qualified, send its data, have it prepared and checked, then receive a first report. Every step was manual, so the programme grew only as fast as the team could keep up. Now a client is onboarded in about 3 weeks instead of about 6 months.

Step 1 / 4

Map the existing process

  • 5key steps
  • 15tasks
  • 4stakeholders
  • 4tools
  • 3handovers
  1. 1QualifyOutputClient signed up.
    • Explain the programme
    • Check eligibility by hand
    • Sign up in the CRM
    SalesEmail, CRM
  2. 2Collect dataOutputData files received.
    • Request the data
    • Receive files by email
    • Forward to the analystHandover
    Sales, clientEmail
  3. 3Prepare dataOutputClean data.
    • Clean and format files
    • List what is missing
    • Send the list to salesHandover
    AnalystExcel
  4. 4Follow upOutputA complete data set.
    • Chase missing data
    • Track status in a spreadsheet
    • Pass data to the expertHandover
    SalesEmail, Excel
  5. 5ReportOutputFirst report delivered.
    • Analyse the data
    • Write the report from scratch
    • Present it to the client
    ExpertPowerPoint

An example based on a workflow we designed and built end to end with a client.

Built around the systems you already run

We map each process on the tools it runs on today (ERP, CRM, helpdesk, document stores) and plan each change around them.

How we handle security
  • SharePoint
  • Excel
  • PowerPoint
  • Outlook
  • Microsoft Teams
  • Microsoft 365
  • Claude
  • ChatGPT
  • Google Drive
  • Salesforce
  • HubSpot
  • Attio
  • SAP
  • Dynamics 365
  • NetSuite
  • Odoo
  • Workday
  • ServiceNow
  • Jira
  • Docusign
  • Slack
  • Azure
  • AWS

Practical details

Who it is for
The leadership of a mid-market company, and the PE firm that backs it. Department heads and process owners join the interviews.
Length
From a few weeks, depending on scope.
Your time
About an hour of interview for each leader, 45-60 minutes for each person interviewed in the teams, a questionnaire for everyone, and regular syncs with leadership to review the findings.

And after?

We build the priority workflows with you: an AI build sprint for use cases on Copilot, Claude or ChatGPT, forward deployed engineering for custom AI agents embedded in your processes.

Which workflow would you redesign first?

Talk to an expert

30 minutes with one of our founders.

In these 30 minutes

  • You tell us how your firm works and where you are with AI.
  • We share what has worked at firms like yours.

“Voila AI brings a rare combination of great technical skills, a structured approach (interviews, weekly check-ins, stakeholder alignment) and business acumen.”

David Kurmann, Partner, EGS Beteiligungen

Talk to an expert

30 minutes with one of our founders.

In these 30 minutes

  • You tell us how your firm works and where you are with AI.
  • We share what has worked at firms like yours.

“They were excellent sparring partners, working with autonomy, flexibility, empathy, and low ego.”

Jordi Campos, Co-Founder & CTO, Optiml
Loading the calendar