The Six-Week AI Transformation Sprint: A Replicable Engagement Methodology

The three-month transformation engagement is a legacy of a world where the diagnostic phase took a month of interviews, the documentation layer took another month, and the recommendation deck took a month to produce. With 2026-vintage tooling, those three months have collapsed into six weeks, and the compression is not achieved by working faster. It is achieved by working on different activities.

The argument for six weeks rather than three is structural. A program shorter than four weeks skips either the proper ESSII exploration with the client team or the tool arbitrage phase, and the deliverable shows it. A program longer than eight weeks, in our experience, loses stakeholder momentum: the operations director gets pulled into the quarterly close, the CFO reallocates budget, the original sponsor takes a new role. The six-week window is wide enough to do the real work and narrow enough to hold the room.

The sprint is designed for an SMB client in the 50 to 500 person range, covering three to seven processes. It assumes one lead consultant (you), optional analyst support, and a client team of a process sponsor plus three to five process owners. Consultants who try to scale it above eight processes or below three tend to break the rhythm. The scope envelope matters.

Below is the week-by-week breakdown. Each week has a deliverable, a meeting cadence, and a pitfall that tends to derail consultants running the sprint for the first time. Follow the structure. The weeks are load-bearing and should not be reshuffled, even when the client asks.

Week one is not kickoff theater. It is the week that determines whether the engagement ships on time or slips into month three. The work here is scoping: which processes are in, which are out, who owns each, what 'done' looks like for the sponsor, and what the hard constraints are.

Deliverables

• Signed scope document: three to seven named processes, with owners and expected outcomes
• Stakeholder map: sponsor, process owners, veto holders (typically CFO, CIO), observers
• Document inventory request: SOPs, meeting transcripts, tool list, ERP/CRM access if needed
• Working-cadence agreement: weekly sponsor check-in, twice-weekly consultant-to-lead-contact sync

Meetings

• Kickoff with sponsor and top contact (90 minutes)
• Individual 30-minute calls with each process owner
• One meeting with the CFO to align on budget and expected financial impact
• One meeting with the CIO or equivalent on tool-stack constraints

Pitfall

The classic week-one mistake is accepting the sponsor's first list of processes without challenging it. Sponsors pick processes that feel visible, not processes where the transformation actually has the highest return. Push back. Ask what the CFO would say matters most. Ask which three processes, if transformed, would make the sponsor feel the engagement was worth it. The list you end up with is usually different from the one you start with.

Week two is the week AI tooling earns its place. The goal is a validated current-state BPMN for every in-scope process, with KPIs on tasks and a preliminary bottleneck register. A pre-2026 consultant would have burned this week and the next on stakeholder interviews alone. Now the interviews validate the tooling's output rather than generate it.

Deliverables

• Current-state BPMN for each in-scope process, validated with the process owner
• KPI layer: duration per task, cost per task, frequency per task, labor role assignment
• Bottleneck register: top three cost drivers and top three duration drivers per process
• Data-quality note: where the numbers are reliable, where they are estimates, where they are guesses

Meetings

• One 60-minute validation session per process owner to confirm the BPMN and the KPI layer
• Mid-week sponsor sync (30 minutes) to flag any discoveries that change the scope
• One cross-process workshop (90 minutes) to surface dependencies between processes

Pitfall

The tooling will produce a BPMN that looks right but is not the process that actually runs in the building. Every engagement has at least one process where the written SOP has not matched reality for three years. If you ship the SOP-based BPMN, your target state is wrong from week three onward. The validation meeting is the checkpoint. Ask the process owner to walk you through the last real execution, step by step. Note the deviations. Update the map.

Week three is the heart of the sprint. ESSII (Eliminate, Simplify, Standardize, Integrate, Intelligize) is applied task by task, in sequence, across every in-scope process. This is the work that does not compress well. It requires a person in the room who knows the business and a consultant who has done this twenty times before.

The format we recommend is a two-hour working session per process, with the process owner and one operational expert from their team. The consultant facilitates, the AI tooling surfaces candidate optimizations, the team votes. The output is a task-by-task classification: kept as is, simplified, standardized, integrated with an existing system, or automated with AI. The 'kept as is' column is usually larger than clients expect, and naming it explicitly is what keeps the recommendation honest.

Deliverables

• Task-level ESSII classification for every task in every in-scope process
• Shortlist of AI-candidate tasks, typically 10-25% of total tasks across the portfolio
• Preliminary cost-reduction estimate per process based on classifications
• List of tasks flagged for change management (not tooling, but people)

Meetings

• One two-hour ESSII session per in-scope process
• One wrap-up session with all process owners (90 minutes) to cross-check assumptions
• Sponsor check-in (30 minutes) to flag politically sensitive recommendations

Pitfall

Consultants new to the sprint tend to rush through Eliminate, Simplify, and Standardize, and land most tasks in Intelligize. This produces a recommendation that is 80 percent AI, which is wrong and expensive. Honest ESSII work lands maybe 15 to 25 percent of tasks in Intelligize. The rest should be eliminated, simplified, standardized, or integrated with existing systems. If your week-three output has more than a third of tasks headed for AI, redo the session.

Week four turns classifications into tool recommendations and cost numbers the CFO will actually read. The goal is a per-process cost model with three scenarios (conservative, expected, aggressive), and a tool shortlist that includes a recommended option, one alternative, and at least one dismissed option with explicit reasoning.

Deliverables

• Tool shortlist per AI-candidate task: recommended, alternative, dismissed (with reasons)
• Three-scenario cost model per process: tool licenses, implementation effort, expected savings
• Aggregate portfolio model: total investment, year-one return, year-two return
• Risk register: integration risk, change management risk, vendor risk

Meetings

• Tool review session with the CIO and process sponsor (90 minutes)
• Cost model review with the CFO (60 minutes)
• Optional vendor calls if the client requests them (usually 30 minutes each)

Pitfall

The arbitrage document is where consultants lose client trust. If every recommended tool is from the same vendor family, the client spots the laziness. If the dismissed options are vague, the recommendation looks like a sales pitch. The fix is to force yourself to name at least two tools you actively dismissed for each AI task, with a one-sentence reason each. The arbitrage is the proof the recommendation is honest.

Week five is the build-out. Target-state BPMNs for every in-scope process, a phased roadmap that sequences the work in waves, and a resourcing plan that tells the client who needs to do what, when, in their organization.

Deliverables

• Target-state BPMN per process, aligned with the ESSII classifications
• Roadmap in three waves: quick wins (weeks 1-6 post-engagement), core automation (months 2-4), advanced agents (months 4-9)
• Resourcing plan: internal owner, external support required, tool-vendor involvement
• Change-management plan: communications, training, success metrics per wave

Meetings

• Target-state walkthrough per process (60 minutes each) with the process owner
• Roadmap workshop with sponsor and top management (90 minutes)
• Optional operations all-hands briefing (30 minutes) to signal the program exists

Pitfall

The roadmap is the place clients over-commit. They see the target state, get excited, and agree to a nine-month plan they will not execute. The consultant's job in week five is to compress the roadmap aggressively: what can realistically be shipped in the next six weeks, by whom, with what support. A smaller wave one that actually ships is worth more than a comprehensive wave one that stalls.

Week six is the narrative week. Everything produced in the first five weeks gets folded into a single deliverable the sponsor can carry into a board meeting and defend without you in the room. This is the week most consultants underestimate, and the underestimation is what turns good engagements into forgettable ones.

Deliverables

• Board deck (8-12 slides) with narrative and one-chart-per-page discipline
• Executive summary (2-3 pages) the CEO reads in seven minutes
• Detailed appendix: BPMN maps, cost models, tool shortlists, roadmap, change plan
• Handoff pack for the internal program owner: action items, templates, meeting cadence

Meetings

• Pre-board dry run with sponsor (90 minutes): rehearse the questions the board will ask
• Board presentation (60 minutes): the consultant presents, sponsor co-signs the recommendations
• Handoff session with internal program owner (90 minutes)

Pitfall

The week-six deliverable is not a summary of what you did. It is a decision document for the board. If the deck reads as a trip report (what we did in week one, what we did in week two), it will not pass the sponsor's internal review. Restructure the deck around the decision: what the board is approving, what the expected return is, what the risks are, and what the next six weeks of execution look like. The weeks one through five exist to earn this deliverable, not to populate its slides.

The six-week sprint ends with a decision and a plan. Execution starts in week seven and runs for six to nine months, depending on the wave structure. The consultant's role shifts from full-time facilitator to part-time coach. The internal program owner takes primary responsibility. This is the handoff most consultants fumble.

Our recommendation is a structured phase-2 engagement: biweekly two-hour reviews with the program owner, monthly sponsor check-ins, and one full-week consultant engagement per wave transition. Total phase-2 time is usually 20-30 percent of phase-1 effort, priced separately. Consultants who skip this layer watch their transformation plans die in committee by month three. Consultants who over-engineer it end up doing the execution work themselves, which is not what the client signed for.

Phase 2 pricing is typically 25 to 40 percent of phase 1 pricing, for a six-to-nine-month engagement. It is monthly retainer work, not daily rate, and it should be booked before phase 1 closes. The sponsor is most bought in during the board presentation. That is when the phase-2 conversation lands.

A two-week variant of the sprint exists, and it covers a narrower scope: one or two processes, one focused diagnostic, one clear recommendation, no full roadmap. It is appropriate when a client wants to pilot the consulting relationship before committing to a larger program, or when the scope is genuinely small.

We usually run it as week two compressed to three days and week six compressed to four days, with a one-day buffer. Fee is in the $12k-20k range for an independent consultant. It does not replace the six-week sprint. It serves as a door-opener: a scoped pilot that demonstrates the tooling, the methodology, and the working style, after which the client signs for the full program.

The 2026 tool landscape has introduced a category that did not exist at meaningful SMB price points two years ago: agentic AI systems that can own a task queue, make decisions within a defined boundary, and hand off to a human only when an exception is triggered. This changes the sprint in two specific places, and ignores it in the rest.

The two places it changes things are week four and week five. In week four, the tool arbitrage now has a third tier to evaluate: point automation tools (RPA-style, one task), workflow orchestration tools (multi-step, rule-based), and agentic systems (multi-step, judgment-based). The evaluation criteria are different for each tier, and conflating them is the most common mistake consultants make when presenting tool recommendations to a CFO who has been reading about AI agents in the trade press.

What agentic tools require that standard automation does not

• A defined decision boundary: the agent must know what it can decide alone and what requires human sign-off
• An audit trail requirement: most SMB compliance frameworks now require a log of agent decisions, not just outputs
• A fallback protocol: what happens when the agent hits an edge case it was not trained on
• A retraining cadence: agentic systems drift when the underlying process changes, and the roadmap must include a review cycle

In week five, the target-state BPMN for any process that includes an agentic component needs a new lane: the agent lane. This is not a cosmetic change. The agent lane carries its own SLA, its own escalation path, and its own success metric. Processes that mix human tasks and agent tasks without a clear lane boundary tend to produce the worst post-engagement outcomes, because accountability becomes ambiguous the moment something goes wrong.

The practical implication for sprint pricing is modest: week four adds roughly four hours of additional evaluation work when agentic tools are in scope, and week five adds a half-day for the agent-lane BPMN work. Neither change breaks the six-week structure. What does break the structure is treating agentic tools as a drop-in replacement for simpler automation without adjusting the governance deliverables. Clients who deploy agents without a decision boundary and an audit trail tend to call their consultant back at month four with a problem that is more expensive to fix than it would have been to prevent.

How to position this with the client in week one

The week-one scoping conversation now needs one additional question: has the client already purchased or trialed any agentic AI tools? In 2026, a meaningful share of SMB clients arrive at the sprint with a vendor contract already signed, often by the CTO or a department head acting independently. If that contract exists, it is a constraint, not a recommendation, and the tool arbitrage in week four must treat it as such. Discovering a pre-committed vendor contract in week four rather than week one is a scope disruption that costs time and trust.