AI strategy

AI strategy and value creation for mid-market and private equity-backed businesses.

Mid-market businesses tend to be well placed to gain from AI, although returns so far have often disappointed. With the Integrated AI Framework, we help leadership teams assess how AI is changing their industry and lead an AI transformation.

Book a complimentary AI Diagnostic Get the AI Playbook

The argument

The mid-market opportunity, and the window to act on it.

The AI Playbook for Mid-Market Leaders, published by Zenith Strategy Associates in November 2026, argues that, relative to their size, mid-market businesses are better placed to gain from AI than small businesses or large enterprises, provided they act early. In most earlier waves of technology, the largest companies gained the most, because systems from mainframes to enterprise software needed large upfront investment, specialist teams and years to put in place. AI tools are mostly bought by subscription or by usage, and many can be put to work within weeks. A mid-market business has a cost base large enough for the savings to matter, and the flexibility to adapt quickly.

The mid-market’s advantage is unlikely to last indefinitely, because AI is changing where profit is earned in many industries and the strongest positions are likely to go to the businesses that adapt first. In most Australian industries adoption is still at an early stage and few businesses have yet built a clear lead, so the next year or two appears to offer a window to act. So far, fewer than four in ten organisations report any effect on earnings from AI, and for most of those the effect is less than 5% of EBIT (McKinsey, The State of AI, November 2025, self-reported).

We bring the Integrated AI Framework and our AI toolkits, and work alongside the business’s own people, with the leadership team in control of the program throughout. In recent engagements, our reviews across functions including customer support, finance, marketing, sales and HR have found significant opportunities to free up 20% to 50% of team hours through automation and AI agents. Zenith Labs, our technology practice, has also built AI agents for core processes across our own strategy advisory work. They now allow us to deliver engagements around 30% faster, and at around 40% lower cost, than our competitors.

The evidence

What the research shows about AI productivity, adoption and returns.

Most businesses that have spent on AI have little to show for it so far, even as adoption in the Australian mid-market rises. The gains appear to go to businesses that build AI into their everyday work.

Around 15%

more enquiries were resolved an hour by customer service agents using AI, in a field study of 5,172 agents.

22%

of Australian businesses with 20 to 199 staff used AI in 2025, up from 3% in the 2022 financial year.

Around 7%

of organisations can show an established return on their AI investment.

2 – 4 years

is the typical payback on AI, against 7 to 12 months for most technology spending.

Sources: Brynjolfsson, Li and Raymond, Quarterly Journal of Economics, 2025. Australian Bureau of Statistics, Characteristics of Australian Business, 2024-25, June 2026. KPMG Global AI Pulse, Q2 2026, 2,145 senior leaders in 20 countries, self-reported. Deloitte, AI ROI: The paradox of rising investment and elusive returns, October 2025, 1,854 executives in 14 European and Middle Eastern countries.

Why the mid-market

Structural advantages that help the mid-market turn AI into results sooner.

In large enterprises, the gap often lies between what AI can do and what the organisation can change (KPMG, May 2026). A mid-market business tends to move faster, for reasons of organisation, technology and people.

Faster Decisions and Change

A mid-market leadership team and board can usually agree a change to a workflow within weeks. In a large corporation the same change crosses departments and several rounds of approval, and tends to take quarters. Speed matters because new technologies rarely pay off straight away. The large productivity gains from electricity came in the 1920s, four decades after the first power stations opened, once factories had been rebuilt around it (David, American Economic Review, 1990).

Simpler Systems and Data

A mid-market business typically runs a handful of core platforms, many already in the cloud, so getting ready for AI is usually a contained project. Large enterprise systems have been built up over decades, with layered platforms and data kept in separate silos, and rebuilding them for AI is often a major program in its own right. Where a mid-market business is not yet ready, the main shortfall is commonly in its data.

Broad Roles That AI Helps Most

Research finds that AI tends to help less skilled staff the most, and helps people perform well outside their specialism. In a study of 758 consultants, those below average improved by 43% with AI, against 17% for those above average (Dell’Acqua and others, Organization Science, 2026). Mid-market roles tend to be broader, so a financial controller who also manages treasury and tax gains some of the specialist support only a large company could once afford.

The record so far

Why many AI investments disappoint.

The most common cause is that businesses buy AI tools but leave the work itself unchanged. Of Australian SMEs using AI, 58% name staff productivity among the largest gains, but only 9% name profitability (NAB, April 2026, self-reported).

AI Tools Without Workflow Redesign

Where AI tools are bought and the work continues as before, usage tends to fade. Organisations use only around half of the software licences they pay for (Zylo 2026 SaaS Management Index). Time saved by AI also tends to disappear unless it is deliberately turned into more output, lower cost, faster delivery or growth.

AI Spend Without Accountability

AI increasingly arrives inside software renewals and usage credits, where it is rarely visible as a project. The cost is spread across many budgets, so no single executive owns the total or can say what it has returned. Usage-based charges can also rise unnoticed, which has led large companies such as Walmart and Cisco to cap staff usage (Financial Times, June 2026).

Ambition Ahead of Evidence

Large flagship AI projects are often approved on conviction, before there is evidence that they will work. Once started, they tend to continue because of the money already spent, and many are cancelled when the full cost becomes clear. A small pilot, measured against a baseline, would often have shown much sooner whether the idea worked.

Successful adopters

The practices that set apart the businesses earning a return on AI.

The published research and our own experience point to the same practices. A tool introduced without a named owner, a measure of benefit or any change to the workflow around it is unlikely to produce a return.

Changing the work as well as the tools

Each workflow is redesigned around what AI makes possible, with each step given to system automation, people, AI or a third party, whichever does it best. Management then decides how the time freed will be used, whether for more output, lower cost or growth.

of companies earn 5% or more of EBIT from AI. They are nearly three times as likely as others to have redesigned their workflows.

Buying before building

Products already on the market come first, starting with the AI inside the platforms the business uses. Building is kept for workflows specific to the business, or where a proprietary solution would give a lasting advantage.

76%

of AI use cases are now bought rather than built in-house.

Giving the program one senior owner

One senior executive is accountable for the return on AI across the business. In the mid-market this is usually the chief executive or a member of the leadership team.

14%

of businesses with clearly defined accountability for AI report established returns, against 4% of those without.

Tracking benefits against the business case

Each initiative starts from a clear business case and measures agreed in advance, and the benefits realised are tracked against the original plan.

Fewer than 1 in 5

organisations track well-defined measures for their AI, the practice most closely linked to bottom-line impact.

Managing adoption as change

Adoption is run as a change program, in which every user has a specific task, training and follow-up from their manager, so that AI becomes part of each person’s regular work.

83%

of staff used AI where their employer encouraged it, against 47% where it did not.

Sources: McKinsey, The State of AI, November 2025, self-reported. Menlo Ventures, The State of Generative AI in the Enterprise, December 2025, US firms. KPMG Global AI Pulse, Q2 2026, self-reported. McKinsey, The State of AI, March 2025, self-reported. Humlum and Vestergaard, NBER Working Paper 33777, 2025, Denmark.

The framework

The Integrated AI Framework, a strategic view combined with a practical plan.

The Integrated AI Framework is Zenith Strategy Associates’ method for AI strategy and value creation in mid-market businesses. The AI Strategy Assessment shows where profit and advantage are shifting in an industry, and the AI Value Creation Plan captures the gains in the business. A layer of AI risk and governance sits beneath both.

Component 1

The AI Strategy Assessment

The AI Strategy Assessment considers how AI is likely to change an industry and a business’s place in it, with the aim of a defensible position and a clear view of new opportunities.

1 Industry Profit Pools

How AI changes the profit an industry earns, how it is divided and where the industry’s boundaries sit, whether or not a business has an AI program of its own.

1a Enhanced Industry Propositions: AI makes what an industry sells faster, better or more consistent, so customer expectations reset and providers that cannot match the new standard tend to lose ground.
1b Industry Transaction Cost Reduction: AI lowers the cost of finding, comparing and buying, which reshapes value chains built on costly information and exposes intermediaries whose margin depends on it.
1c Industry Boundary Shifts: New services become viable and barriers to entry fall, so a business can enter nearby markets more cheaply, and businesses from those markets can enter its own.

2 Sources of Competitive Advantage

What sets a business apart from rivals that can buy the same AI models and tools. Each source of advantage is built up over time.

2a Data and Learning Flywheel: Using AI in the business’s own work creates data that rivals do not have, which improves the AI and its results with each cycle, and in turn creates more data.
2b Customer Integration: A supplier built into its customers’ processes and systems is harder to replace, and likely to be well positioned for the routine orders their AI agents begin to place.
2c Integrated Propositions: AI is combined with hardware, a service network or people on site in a single offer, which is much harder to copy than software alone.

3 Enabling Architecture

Whether a business’s technology and data can support AI agents at scale, since agents can act only on systems they can connect to, using data they can trust.

3a Cloud Platform Foundation: Cloud-based core systems, which many mid-market businesses already run, tend to be much easier for AI agents to connect to, and gain new AI features with each update.
3b Data Architecture: Clean data, defined the same way in every system, gives AI reliable inputs and a single view of each customer. Data is commonly the largest gap in a mid-market business.
3c Integration and Access: Open interfaces, known as APIs, let AI agents read from and write to systems, with controls over what each agent can see and do, and a record of what it did.

Component 2

The AI Value Creation Plan

The AI Value Creation Plan finds where AI can take on work in each function, decides which workflows to change first, and makes sure the time saved shows up as lower cost or new revenue.

4 AI Transition

The program that moves the business to an AI-enabled operating model, taken in sequence from each function’s remit through its workflows to phased delivery.

4a Function Remit: Each function’s remit is reconsidered now that AI is available, including work it has never had the people to do, which a review of its existing work alone would miss.
4b Workflow Review and Redesign: Existing and new workflows are mapped, rated Anchor, Assist or Automate, judged by their benefit, prioritised and redesigned along with the roles around them.
4c Phased Benefit Delivery: Priority workflows are piloted against a baseline, scaled in phases as they prove their value and tracked to the profit and loss, with savings funding the next phase.

5 Organisational Readiness

The capacity to keep finding and capturing value from AI once the transition program ends.

5a Opportunity Identification: Workflow ratings are refreshed as the tools improve and the opportunity map is kept current, so new openings are found as AI changes the economics of the business.
5b Resource Commitment: The program has one senior owner, capable people assigned to it, funding from its own savings, data quality treated as an asset and clear rules for relying on AI output.
5c Structural Adaptation: Structures, decision rights and ways of working continue to change as the technology develops, beyond the first redesign.

Under both components

AI Risk and Governance

Governance lets a business trust AI with more of the work, and without it much of the saving can be lost to manual checks. The controls align with the National AI Centre’s Guidance for AI Adoption (October 2025).

Accountability and Policies

Each use of AI has a named owner and plain-language rules on what it may and may not do, which reflect legal and regulatory obligations and keep customer data under the business’s control.

External Guardrails

Controls that sit outside the AI tools check each action against the agreed rules before it takes effect, since an AI model’s own safeguards can fail and are not relied on alone.

Human Oversight

People check AI output at first, with clear rules for when a case is passed to a person, and autonomy is extended as the evidence builds.

Audit Trail

A permanent record is kept of every AI action and the information behind it, for review after the event.

Workflow ratings

How far each workflow can be handed to AI, from Anchor to Automate.

Each workflow, existing or new, is rated during the workflow review in the AI Value Creation Plan. Workflows tend to move towards Automate as tools improve and confidence builds.

Anchor

Typically high stakes or heavy judgement

The work stays with a person for now, and the reason is recorded so that the rating can be revisited later.

Assist

Where most knowledge work sits today

AI drafts and prepares the work, and a person reviews it and owns the result.

Automate

Typically high volume, clear rules and easy to check

AI does the work, and people check samples and handle the exceptions.

In a survey of large US companies, two thirds of departments used AI agents in some form, but only 17% let an AI agent act without a person approving each step (Wharton Human-AI Research and GBK Collective, October 2026, self-reported).

Rating factors

The factors behind the rating, each assessed as high, medium or low.

Volume

How often the workflow runs, and how much team time it takes.

Clear Rules

Whether a capable new starter could learn the workflow from a short walkthrough.

Easy to Check

How quickly someone can check the output, compared with doing the work themselves.

Low Stakes

How small the cost of an uncaught error is, and how sensitive the data involved is.

Low Judgement

How little the workflow depends on one person’s expertise or relationships.

Key questions

The questions the report puts to mid-market leadership teams.

Each component of the Integrated AI Framework includes key questions for leadership. They test the business’s exposure to AI in its industry, and whether the gains are reaching the profit and loss.

The AI Strategy Assessment

01 Which of the industry’s returns depend on information customers lack, or on the cost of searching, comparing and coordinating, and how far will AI reduce those costs?
02 How will AI change the minimum scale needed to compete, and the barriers to entry, in the business’s own market and in neighbouring ones?
03 Which of the business’s advantages rest on assets AI cannot replicate, such as proprietary data, close integration with customers or physical delivery, and will AI agents that compare suppliers raise or lower customers’ switching costs?
04 Which system or data gaps would block the most valuable AI opportunities, and what would it take to close them?

The AI Value Creation Plan

01 Has each function’s remit been reconsidered, and each workflow rated Anchor, Assist or Automate, with the benefit and how it will be measured agreed in advance?
02 For each priority workflow, has the AI already in the business’s own software been tested before a specialist service or a bespoke AI agent is considered?
03 Are the benefits tracked in named lines of the profit and loss, with the savings used to fund the next phase?
04 Is there one senior owner for AI, and is the opportunity map reviewed regularly as the tools improve?

How we help

Senior-led support from strategy through to delivery.

Zenith Strategy Associates, a Sydney consultancy, works with companies, private equity funds and their portfolio companies in Australia and New Zealand, bringing the Integrated AI Framework and AI toolkits to each stage.

Zenith Labs

AI Strategy Assessment

Position in the industry

An assessment of where AI is moving profit in the industry, what will set the business apart once rivals have the same tools, and whether its systems and data are ready for AI agents.

A scorecard of the business’s position, the strategic response it calls for, and the new opportunities AI brings within reach.

AI Value Creation Plan

Workflows in each function

A review of each function’s workflows, to find where AI can save time or add revenue, rate each one Anchor, Assist or Automate, and decide which to pilot first.

A prioritised program, with early savings funding later phases and benefits tracked against the business case.

Delivery and Zenith Labs

Pilots and bespoke AI agents

Pilots configured and running within weeks, using the AI in the business’s existing software first, and bespoke AI agents built by Zenith Labs where they are justified.

Proven workflows scaled quickly, with governance in place from the first pilot.

AI Readiness for Investors

Potential target investments

For private equity firms, an AI readiness assessment of potential target investments, covering their exposure to AI and the opportunity it presents.

Evidence for the investment case, and the basis for a value creation plan after acquisition.

Zenith Labs is the technology practice of Zenith Strategy Associates. It specialises in AI diagnostic and due diligence services, and in bespoke AI agents and applications for businesses. It also builds the agents we use to deliver our own engagements.

A practical first step

A complimentary AI Diagnostic for mid-market leadership teams.

It starts with a 30 to 45 minute call with the leadership team on Microsoft Teams. A senior team member prepares findings on industry position, opportunities by function, quick wins and critical follow-up questions, and presents them in a short session. There is no cost, and no obligation beyond it.

Book your AI Diagnostic call

The report

The AI Playbook for Mid-Market Leaders, with the detail behind the framework.

Published in November 2026, the report sets out the Integrated AI Framework with the worked examples by function and case examples that this page leaves out. Andrew Hone, founder of Zenith Strategy Associates and formerly of Bain & Company and L.E.K. Consulting, wrote it with Zenith Labs.

01The case for the mid-market, and for acting now
02Why many AI investments disappoint
03The AI Strategy Assessment, with an illustrative scorecard
04The AI Value Creation Plan and the opportunity map
05Worked examples from customer service to technology
06AI risk and governance, from the first pilot
07The AI Diagnostic, with an example of its output

The AI Playbook for Mid-Market Leaders, cover

Common questions

Questions we are asked.

How should a mid-market business approach AI?

The AI Playbook for Mid-Market Leaders (Zenith Strategy Associates, November 2026) argues that AI may work best as a growth program, funded by the efficiency savings it makes possible, rather than as a cost-cutting exercise alone. The leadership team looks at how AI is likely to shift profit and advantage in its industry, and at the same time at the workflows in each function where AI can take on work. One senior executive owns the program, each initiative is measured against an agreed business case, and the savings from each phase fund the next.

Why are mid-market businesses well placed to benefit from AI?

Mid-market businesses tend to combine a cost base big enough for AI savings to matter with the ability to change faster than a large enterprise. The AI Playbook for Mid-Market Leaders (Zenith Strategy Associates, November 2026) argues that, relative to their size, mid-market businesses that move early have more to gain from AI than either small or large businesses. AI tools costing a few hundred dollars a month can now do part of the work that once needed specialist teams. A compact leadership team can also agree a workflow change within weeks.

Why do so many AI investments disappoint, and what do successful adopters do differently?

In most cases AI investments disappoint because the tools are bought but the work itself is left unchanged, so the time saved is absorbed rather than turned into lower cost or more output. KPMG’s Global AI Pulse found that only around 7% of organisations can show an established return on their AI investment (Q2 2026, self-reported). The businesses earning returns change the work as well as the tools, buy before they build, give the program one senior owner, track benefits against the business case and manage adoption as change.

What is the Integrated AI Framework?

The Integrated AI Framework is the approach Zenith Strategy Associates takes to AI strategy and value creation in the mid-market, set out by Andrew Hone in The AI Playbook for Mid-Market Leaders (November 2026). Its AI Strategy Assessment covers Industry Profit Pools, Sources of Competitive Advantage and Enabling Architecture, to judge how AI will alter the business’s competitive position. Its AI Value Creation Plan covers the AI Transition and Organisational Readiness, to find, deliver and sustain the gains inside the business, and AI risk and governance underpins both.

What do Anchor, Assist and Automate mean in an AI workflow review?

Anchor, Assist and Automate are the ratings the Integrated AI Framework gives each workflow, according to how far it can be handed to AI. Anchor keeps the work with a person for the time being, with the reason recorded. Assist has AI draft and prepare the work while a person remains responsible for the result, and most knowledge work falls under Assist today. Automate hands the work to AI, with people checking samples and dealing with exceptions. Each rating rests on the workflow’s volume, clarity of rules, ease of checking, stakes and need for judgement.

Where should a mid-market business start with AI?

A mid-market business can usually start with two or three quick wins, run as measured pilots with AI that is already part of the software it owns, so that the decision to scale or stop rests on evidence. The leadership team also needs an early view of how AI could change its industry. The complimentary AI Diagnostic from Zenith Strategy Associates covers the industry view and the quick wins at a high level.

How should a business measure and track the return on AI?

The return on AI is usually best measured workflow by workflow, against a baseline of volume, cost, turnaround and error rate taken before AI touches the work, and for growth opportunities against conversion or revenue. Each initiative needs a clear business case before it starts, and the benefits are booked to named lines in the profit and loss. McKinsey (The State of AI, March 2025, self-reported) found that fewer than one in five organisations track well-defined measures for their AI, although this is the practice most closely linked to bottom-line impact.

Should a mid-market business build its own AI agents or buy AI software?

In most cases a mid-market business should buy before it builds. The Integrated AI Framework’s order of preference is the AI included in the software the business already owns, then a specialist third-party service. A bespoke AI agent is built only where no viable product exists, or where a proprietary solution would generate significant competitive advantage. Built-in AI can often be switched on within days or weeks, as it already has the business’s data. Where building is justified, a bespoke agent is far quicker and cheaper to build than traditional software, and becomes the business’s own intellectual property.

What does the complimentary AI Diagnostic involve?

The AI Diagnostic applies the Integrated AI Framework to a single business at a high level, free of charge and with no commitment to further work. A member of the senior team at Zenith Strategy Associates prepares it, drawing on a 30 to 45 minute Microsoft Teams call with the leadership team. It covers the business’s industry position, its largest opportunities by function, two or three quick wins suited to a measured pilot, and the critical follow-up questions. The findings are then discussed with the leadership team at a short second meeting.

Starting a conversation

Schedule a call.

Most of our AI engagements begin with a short conversation with a principal about what AI is likely to change in your industry, and which workflows in your business to pilot first.

Get in touch