Key takeaways
- Strategy-only AI work can produce clarity and still leave operating metrics unchanged.
- Workflow automation starts from the job-to-be-done, then chooses the AI pattern.
- Governance and human-in-the-loop must ship with the feature, not as a later appendix.
- Judge AI consulting by cycle time, turnaround, and adoption—not pilot theaters alone.
Demand for AI consulting in Chicago is high across manufacturing, insurance, healthcare, financial services, construction, life sciences, and CPG. Budgets exist. Pilots multiply. Yet many organizations can point to activity—roadmaps, vendor bake-offs, a chatbot in a sandbox—while quote cycles, diligence queues, and exception piles look roughly the same.
The gap is often not model quality. It is engagement design: strategy-only versus work that puts AI into the workflow people already run.
We wrote about that pattern in detail in Putting AI into the workflow, not beside it. This Perspective applies the same theme to how Chicago buyers should evaluate AI consulting offers—without repeating that article wholesale.
What strategy-only consulting is good for
Strategy has a legitimate job. Leadership needs a shared language for opportunity, risk, data readiness, and sequencing. A clear operating model for where AI may assist decisions—and where it must not—can prevent expensive false starts. In regulated environments, naming accountability before tooling is wise.
Strategy-only becomes a problem when the deliverable is the destination: a maturity score, a heat map, a multi-year roadmap that never specifies which step in which workflow gets shorter, safer, or clearer. The organization feels aligned. The queue does not move.
If your AI consulting partner’s primary artifact is a deck—and there is no path into product, integration, and change enablement—you should treat the engagement as education, not transformation. Education can be valuable. It is not the same as automation that operators must use to finish the day.
What “beside the workflow” looks like in consulting form
In our Insights piece on embedding AI, we described AI beside the workflow: a capable assistant on a separate tab; an automation that never writes back to the system of record; informal use with informal oversight. Consulting engagements recreate that pattern when they optimize for demos and pilot metrics—accuracy on a sample set, thumbs-up rates, session counts—while cycle time and turnaround stay off the board.
Beside-the-workflow consulting also delays governance. If use is optional, audit trails, escalation, and ownership stay optional. That may be fine for brainstorming. It is a poor fit for underwriting support, clinical operations, diligence review, or manufacturing quote paths where an incorrect suggestion has cost and compliance consequences.
Workflow automation as the consulting brief
Our AI and intelligent automation practice is built around a different brief: where does this capability sit in the sequence of work someone already owns? Not only “what can the model do?” but “what step changes—and who remains accountable when it is wrong?”
That usually means:
- Mapping the job-to-be-done (intake, triage, draft, review, approve, hand off, audit)
- Choosing patterns that fit those steps—knowledge assistants, intelligent process automation, human-in-the-loop agentic workflows
- Integrating with ERP, CRM, case systems, document repositories, and identity so the assistant is not a side channel
- Designing human-in-the-loop as UX and process—not a disclaimer slide
- Shipping oversight with the feature: access controls, audit trails, retention, model/prompt change management
- Treating change enablement as part of delivery, not a training burst after launch
This is slower to demo than a chat window. It is faster to matter when the measure is how the work runs.
Strategy and workflow are not enemies—sequence them
The useful buy is rarely “never do strategy.” It is “do not stop at strategy.” A short opportunity assessment tied to a measurable outcome can precede build. What it should produce is decision-grade: which workflow, which metric, which constraints, which first slice—not a generic AI maturity narrative.
From our Chicago location, engagements start with a short discovery sprint—weeks, not quarters—that ends in a working plan and an honest estimate. AI work follows the same posture. Cross-functional pods (product manager, architect, engineers, QA, DevOps) embed with your team or deliver end to end. We do not sell AI as staff augmentation seats.
How to tell the difference in a vendor conversation
Ask:
- Will success be judged by operating metrics (cycle time, turnaround, adoption) or by pilot theater alone?
- Does the plan put AI on the critical path of a named workflow—with write-back—or beside it?
- How are human confirmation, escalation, and audit designed into the product from the first cut?
- What systems must integrate for the assistant to stop being optional?
- What does discovery produce that we can estimate and decide from?
- Which published outcomes are examples of past work—and which statements are guarantees?
On that last point, we reuse only company-published examples, framed as such: a manufacturer’s quote-to-order cycle cut by 40%; due-diligence turnaround reduced by more than half at a global law firm, including an AI-powered risk-assessment platform used by 4+ Fortune 100 companies. Those illustrate what “in the workflow” can look like when AI is tied to an operating number. They are not a promise that your program will match them.
What Chicago buyers should optimize for
Proximity helps when workshops and steering sessions need to happen in person in the West Loop or at your site—and when Midwest headquarters want shared time zones. Proximity does not replace delivery discipline. Optimize for partners who will connect strategy to embedded software, data, integration, and adoption.
Industry shape still matters. The same AI pattern fails or succeeds based on how field data, claims files, or plant quoting actually move. Respect the floor of the work.
Closing
If your AI consulting program looks impressive in isolation, ask the question we closed our Insights article with in different words: does anyone have to leave their real workflow to use what you bought—and did the consulting engagement ever intend to change that?
Strategy-only can leave you with language. Workflow-embedded automation aims to leave you with a shorter queue, clearer accountability, and a metric that moved. That is the standard we hold for AI consulting from Chicago: put AI into the workflow, not beside it—and judge the work by how the day runs.
