What both terms mean, the frameworks that make an assessment repeatable, and why the assessment is the most sellable AI engagement an MSP has.
Process Intelligence is the practice of documenting how a business actually works before automating any part of it. For an MSP it is a paid assessment engagement: interview the executives, inventory and score the processes, map the workflows into written procedures, and produce a roadmap the client will fund.
AI Readiness is the scored output. Five domains, five-point scale each, producing domain scores and one aggregate AI Readiness Score on a maturity continuum.
Written by Tony Ferrigno, Founder and Managing Partner, AiT Advisory Group. Author of AI for MSPs: Process Intelligence & AI Readiness.
Most MSPs trying to enter AI reach for the hardest thing first. They try to sell an automation build to a client who has never bought anything like it, at a price the client has no way to evaluate.
The assessment inverts that. It is small, fixed-fee, short, and low-risk. It answers a question the client is already asking themselves, which is some version of "everyone says we should be doing AI, are we even ready." It produces a document the client owns. And it puts you in a room with the CEO and the COO for several hours, which is worth more than the fee.
It also does the qualification work you would otherwise do for free. By the end you know whether this client has processes worth automating, data worth using, and a sponsor willing to fund anything. Clients who fail that test have paid you to find out.
Worth stating plainly, because the term collides. In the enterprise software market, "process intelligence" means log-based process mining: software that reconstructs workflows from system event data, priced for large organizations with a team to run it. That is a different thing solving a different problem.
| Dimension | Enterprise process mining | Process Intelligence for MSPs |
|---|---|---|
| Method | Reconstructs processes from system event logs | Structured executive and functional-leader interviews |
| Assumes | Processes already run inside instrumented systems | The processes that matter live in people's heads and spreadsheets |
| Client size | Large enterprise with a dedicated team | 25 to 300 users, no analyst on staff |
| Software purchase | Required, and substantial | None |
| Duration | Months | Two to five weeks |
| Output | A dashboard someone has to maintain | Written documentation and a funded roadmap the client owns |
Most "AI readiness assessments" in the channel today are vendor lead magnets. They are free, they take an hour, and they conclude that the client should buy Copilot licenses. That is a licensing checklist wearing an assessment's clothes. A real assessment can conclude that the client is not ready, and sometimes should.
Five components, run in order. Components one and two qualify. Three and four create the lasting value. Five is what gets funded.
Structured interviews with the CEO, the COO, and functional leaders. Not a survey. Not an email questionnaire.
Every process worth naming gets scored on frequency, effort, error rate, and AI suitability.
Turn the priority processes into documented workflows and written standard operating procedures.
Information architecture plus a governance model, so the documentation stays current instead of decaying.
Synthesize everything into a sequenced plan with scores, priorities, and dependencies.
Each domain scored on a five-point scale. The domain scores matter more than the aggregate, because they tell the client exactly where the work is.
| Domain | What it measures | Common finding |
|---|---|---|
| Technology Infrastructure | Whether the systems can support AI workloads and integrate with each other | Integration gaps nobody knew existed until someone asked |
| Data Quality and Availability | Whether the data an agent would need is accessible, current, and trustworthy | The data exists but lives in three places that disagree |
| Business Process Maturity | Whether processes are consistent and documented enough to automate | Process debt, usually more than anyone expected |
| Workforce Readiness | Whether people will adopt, resist, or quietly route around new tools | Enthusiasm at the top, unspoken concern below it |
| Governance and Compliance | Whether policy, approval, and audit structures exist to control AI use | No AI policy at all, and often no plan to write one |
The fifth domain is the one that turns an assessment into recurring revenue. Almost no SMB has an AI governance policy. Finding that gap is not a problem you point at, it is a service you propose.
Business impact on one axis, automation potential on the other. Four quadrants, four different recommendations.
Immediate focus. These become the agent candidates that carry the business case.
Process improvement and data quality work before any AI. Saying so builds more trust than promising automation you cannot deliver.
Automate for efficiency. Valuable early, because momentum is worth more than magnitude in the first ninety days.
Operational maintenance only. Revisit annually. Saying “leave this alone” is a finding too.
Process debt is the accumulated cost of informal workarounds, undocumented exceptions, and legacy procedures that nobody owns. It does not show up on a balance sheet. It shows up when someone leaves.
Naming it is what creates urgency, and quantifying it is what validates the fee. The framing that lands in a boardroom is specific and countable:
"You have 23 undocumented processes that represent significant knowledge-loss risk if your key people leave."
That sentence does more work than any slide about AI. It reframes the engagement from a technology project the client can defer into an operational risk they already have. Note that it also stands entirely on its own merits. Even a client who never buys a single agent is better off having the list.
Three packages. Price against the risk the assessment removes, not against the hours it takes.
Client size moves those bands considerably. A small client of 1 to 25 users sits at the bottom of each range. An enterprise client above 300 users can reach $55,000 to $100,000 or more for a full program.
Bill 50 percent at engagement start and 50 percent at report delivery. Put the annual governance retainer, $800 to $3,000 per month, in the original proposal. Never hold it back as a post-engagement upsell, because a client who has already received the report has less reason to buy the thing that keeps it current.
The assessment prevents failed AI projects costing $50,000 to $500,000. It identifies automation opportunities worth $5,000 to $50,000 per month. It creates governance structures that reduce regulatory exposure. Against that, a $10,000 to $25,000 engagement is not a cost decision. It is insurance with a roadmap attached.
Five failure modes. Four of them are visible in the first week if you know what to watch for.
The sponsor books the kickoff then delegates every interview. The report lands with nobody senior invested in it. Catch this by making the CEO interview a condition of starting, not a scheduling preference.
The client has already decided what the assessment should say, usually that they need a tool someone already sold them on. Your findings become an obstacle. Surface it early by asking what they expect the report to conclude.
Staff attend interviews and say nothing useful, because they think the exercise is about cutting headcount. Nobody describes their workarounds to someone they think is auditing them. Address it before the first interview, not after.
The report is delivered, everyone agrees it is good, and nothing is funded. Almost always a sequencing failure: the roadmap named ten things instead of the one thing to do next quarter.
The assessment surfaces something the sponsor did not want documented. Handle it privately and ahead of the written report. How you manage this determines whether you ever work with that client again.
The roadmap is the bridge. High-impact, high-automation processes are agent candidates, and agent candidates are where Managed Intelligence begins. The governance gaps from the fifth domain justify the retainer. The documented SOPs from component three are the input those agents need.
That sequence is the point. An assessment sold on its own is a good engagement. An assessment sold as the front door to a recurring practice is a business.