The AI strategy that a SaaS company needs is built around product features, model integration, and developer tooling.

The AI strategy that a $15M distribution company needs is built around quoting time, customer notification quality, and the Monday morning operations briefing.

These are different strategies for different businesses.

The fact that most AI strategy content describes the first does not mean the second does not exist or does not produce returns. It means the content has not been written for you yet.

This article is.

This article answers the question directly: yes, a non-tech company needs an AI strategy.

Then it describes what that strategy is, specifically calibrated to the operational reality of a $5M to $25M non-tech company, what it produces, and where to start.

What “AI strategy” means for a non-tech company — specifically

What it is not

Training a custom AI model on proprietary data: requires ML engineering, significant data infrastructure, and ongoing model maintenance. A $15M non-tech company does not have or need these.
Building AI features into a product: the non-tech company’s product is a service, a physical good, or a professional capability. Not software.
Hiring an AI engineer or data scientist: the operational AI workflows that produce the most return for a non-tech company do not require these skills.
“Digital transformation”: a buzzword that describes something large and complex and usually fails for the same reasons it always fails: too broad, too slow, too disconnected from operational reality.

What it is

An AI strategy for a non-tech company is: the plan for deploying commercially available AI tools on the company’s recurring operational workflows, in a way that produces measurable time recovery and quality improvement.

Sustained by a maintained Foundation and an ongoing improvement loop.

It is the answer to five operational questions:

Which five workflows in this company consume the most staff time and are most amenable to AI assistance?
What is the context pack that makes AI produce outputs that reflect how this company communicates, what industry it operates in, and what quality standard it applies?
How do we train the team so that AI use becomes a habit rather than a compliance requirement?
Who maintains the Foundation so the system compounds rather than stagnates?
What automations make sense after the Foundation is stable and the team is fluent?

These questions are answerable by a non-tech founder with no AI background. They require operational knowledge — knowing what the business’s workflows are, what its communication standards are, what its quality requirements are. They do not require technical knowledge.

Why non-tech companies need AI strategy more urgently than many tech companies

The operational leverage argument

Tech companies have already automated significant portions of their operational work through software: product management systems, developer tools, deployment pipelines, customer success platforms.

The proportion of their staff time spent on unstructured document production, manual communication drafting, and data compilation is lower than at non-tech companies.

The $15M manufacturing company, the $20M distribution company, and the $22M healthcare group have not automated these workflows, because the software tools that address them have historically required technical implementation capability the non-tech company does not have.

AI changes this. The large language model that drafts the payer appeal letter or the customer delay notification does not require the technical implementation that an ERP integration requires. The AI strategy for a non-tech company addresses operational automation opportunities that have been inaccessible until now.

The competitive gap argument

Non-tech companies compete with other non-tech companies.

The $15M specialty manufacturer competes with other $10M to $25M specialty manufacturers, not with $500M automotive tier-one suppliers. In this competitive set, the AI adoption gap between early and late movers is significant because:

The competitive set is starting from a relatively low baseline of operational automation, so AI adoption produces proportionally larger relative advantages
The early mover does not need to be dramatically better. They need to be faster to quote, more consistent in communication, and more productive in proposal development than the specific competitors they are actually competing against.
The switching cost for the customer or client is low enough that 10 to 15 percentage point improvements in quote speed or proposal quality affect purchasing decisions

The talent retention argument

Non-tech companies consistently struggle to retain administrative and operations staff who are spending a significant portion of their time on repetitive desk work.

The HVAC parts distributor whose customer service team processes 200 manual notifications per week has a staff retention problem that the competitor who has automated that process to 30 minutes of batch review does not.

The non-tech company that deploys AI on its most repetitive operations work is offering a meaningfully different job — more relationship work, more judgment work, less formatting and drafting — a distinction the screen-room vs. judgment-room framework maps precisely — that is more attractive to the staff it wants to retain.

The five-component non-tech AI strategy

Component 1: The primary task mix

The five to eight recurring operational tasks where AI produces the most time savings. Identified by asking function leaders: “What is the most time-consuming recurring task in your function that involves writing, drafting, or compiling information?”

Sector	Primary task mix
Distribution ($15M)	Back-order notifications, RFQ responses, account health summaries, supplier communications, weekly briefing
Specialty manufacturing ($20M)	RFQ response drafts, NCR documentation, production scheduling communications, customer delay notifications, management reporting
Healthcare group ($15M)	Payer appeal letters, funder reports, staff communications, referral source communications, operations briefings

Component 2: The Foundation

The context pack: the set of company-specific documents (voice guides, communication standards, vocabulary guides, workflow specifications) that tell AI how this company communicates, what industry it operates in, and what quality standard its work requires. See what an AI Foundation contains and how it is built for the full component breakdown.

State	What AI produces
Without the Foundation	Generic outputs requiring significant editing
With the Foundation	Company-specific outputs requiring review but not rewriting

The Foundation build takes 8 to 15 hours of structured sessions with function leaders, producing 5 to 8 documents of 200 to 400 words each. Not a technology project: an operational documentation project.

Component 3: The training programme

Individual anchor workflow sessions of 25 to 35 minutes per team member, using real current work, ending with a usable output. Followed by day-seven follow-up sessions. Running for 30 days with the full team.

Not a group session where everyone learns how AI works. A session where each person uses AI on their specific most time-consuming task and gets something useful before the session ends.

The difference between AI training and AI adoption is exactly this: whether the team member leaves with compliance or with a genuine habit anchored to real work.

Component 4: The improvement loop

The AI system owner maintains the Foundation: reviews the week’s AI-assisted outputs, identifies quality gaps, and updates the context documents and custom instructions.

State	Outcome
Without the improvement loop	System stagnates at initial build quality
With the improvement loop	System produces measurably better outputs at month six than at month two

Component 5: Phase 3 automations

After the Foundation is stable and the team is fluent (typically months three to six): the first automations that run without manual initiation.

Examples:

The Monday morning operations briefing generated automatically from the ERP export
The customer back-order notifications triggered from the back-order exception report
The denial triage produced automatically when the billing system exports the daily denial batch

Phase 3 is where AI stops being a tool the team uses and starts being infrastructure the company operates on. The full progression from Foundation build through automation is described in the four-phase mid-market AI strategy.

Where to start — specifically, this week

The one-task start

The starting point is not a strategy document. It is not a vendor evaluation. It is not a technology assessment.

It is the founder identifying the single most time-consuming recurring task in their own work and using AI on it this week.

The criteria for the first task:

The founder does it themselves (so they can evaluate the output personally)
It is genuinely recurring (so the habit can form)
It is primarily a writing or drafting task (the highest-return category for first AI use)
It involves information the founder has (so they can provide the context AI needs)

Common first tasks for non-tech founders:

The weekly management report or operations briefing
The investor update letter or board memo
The key client or customer communication about a specific relationship issue
The key supplier negotiation or performance communication

The two-week experiment

Week one: use AI on one instance of this task. Use the blank Claude or ChatGPT session (no context loaded, no special configuration). Evaluate the output: how much editing was required? What was missing?

Week two: add the company context before running the same task. Brief description of what the company does, the tone it uses, and the vocabulary conventions. Evaluate: is the output closer to what was needed? How much editing was required this time?

The difference between week one and week two outputs is the Foundation gap — the missing context that explains why generic AI is inadequate and specifically contextualised AI is useful. This is the founder’s direct experience of why the Foundation matters.

The planning conversation after week two

The founder who has completed the two-week experiment has direct personal evidence of what AI does and does not produce.

From this position, the planning conversation about team-wide deployment, Foundation build, and training programme is grounded in specific personal experience rather than vendor claims or theoretical frameworks.

The planning conversation is more productive after the two-week experiment than before it.

Common questions on AI strategy for non-tech companies

”What if our industry has specific regulations about AI use that we don’t understand yet?”

Identify the specific regulation first. Most mid-market non-tech industries have no specific prohibition on using AI for operational writing and drafting, with appropriate human review.

The regulated industries that have specific requirements (healthcare: HIPAA and BAA requirements, legal: professional conduct rules, financial services: specific data protection obligations) all have documented compliance frameworks for AI use.

The practical guidance: engage your compliance officer or legal counsel with a specific question about your primary workflow types. “Can we use a BAA-covered AI tool to draft payer appeal letters for human review before submission?” is a specific question that gets a specific answer. “Is AI compliant in our industry?” is not specific enough to produce useful guidance.

”Is AI strategy the same for a $5M company as a $25M company?”

The five-component framework is the same. The scale is different.

A $5M company typically deploys one to two function Projects with three to five workflows. A $25M company typically deploys four to six function Projects with eight to fifteen workflows.

The Foundation build for the $5M company takes two to three days. For the $25M company: one to two weeks. The training programme scales similarly.

The starting point is identical: the founder using AI on their own most time-consuming recurring task this week.

Ready to move from the two-week experiment to the full five-component strategy?

A non-tech company needs an AI strategy, but not the same AI strategy a tech company needs. This strategy is calibrated to operational workflow automation using commercially available tools.

The starting point is the founder using AI on their own most time-consuming task this week. Everything else — the Foundation build, the team training programme, the Phase 3 automations — builds from that personal experience.

Path one: start the two-week experiment Monday. Take the one task you identified as most time-consuming. Open Claude or ChatGPT. Paste the relevant inputs. Evaluate the output. Add your company context in week two. Compare the two outputs. That comparison is the evidence that determines whether and how to proceed with the team-wide deployment.

Path two: bring in a partner. Phos AI Labs builds the five-component AI strategy for non-tech mid-market companies: Foundation build, team training programme, and Phase 3 automation roadmap. Learn what Phos AI Labs is and how we work before booking. Thirty minutes, no deck. Start here.

Non-Tech Companies Need an AI Strategy Too