A precon director sat across from an owner's capital committee with one line item left to defend: a $90K allocation for the precon AI stack, buried inside the $42M GMP. He led with the slide every vendor had taught him to lead with, a bold "30% productivity gain on takeoff," and the owner's CFO did exactly what a CFO does with an unsourced percentage. She asked where the number came from. He did not have an answer she could check, so she struck the line, the GMP came in $90K lighter on paper, and the GC ate the tooling cost out of its own fee on a job it had already won. The expensive part was not the $90K. It was that the firm now had no template for the next owner, no repeatable way to fund the tools that its estimators had already reorganized their workday around. This lesson treats the AI ROI memo not as a one-off project document but as a firm-strategist instrument: a standard, reusable artifact that a precon leader can hand to any owner and have it survive scrutiny, because it is built from substantiated hours, real rates, and named outcomes rather than a percentage no one can verify. By the end you will be able to author the AI ROI memo that gets an owner to fund a $90K precon AI stack inside a $42M GMP, and to do it the same way every time.

Why the Owner Funds What It Can Verify, Not What You Claim

When the owner pays for the preconstruction phase, every dollar the GC spends in precon is the owner's dollar, and the AI tool spend is no exception: Togal.AI on takeoff, the estimate sandbox, ALICE on schedule scenarios, the licenses and seats, all of it sits inside the GMP as a cost the owner is being asked to fund. That changes the request. It is not the GC deciding how to spend its own overhead; it is the GC asking the owner to approve an expenditure, which means the owner gets to ask the question any prudent payer asks: what do I get for this, and how do I know. The honest answer to "how do I know" is the entire game, because an owner does not fund a claim, an owner funds a verifiable accounting.

This is why the productivity percentage fails so reliably. A figure like "30% faster" or "40% productivity gain" is a conclusion with no audit trail: the owner cannot trace it to hours, cannot tie it to a rate, cannot test it against the project in front of them, so it reads as marketing, and a sophisticated owner discounts marketing to zero. Worse, leading with the percentage signals that the GC either does not have the underlying numbers or does not want to show them, which is the inference that gets a line item struck. The owner's instinct is correct: a number you cannot verify is a number you should not pay for, and the GC that leads with the unverifiable percentage has handed the owner the reason to say no.

The firm-strategist reframing is to build the memo backward from the owner's verification, not forward from the GC's enthusiasm. Start from the question the owner will ask, where does this number come from, and make every line answer it: this many estimator hours, at this rate, saved on this task, netted against this license cost, paying back in this window. When every figure can be traced to a source the owner can check, the owner can do the one thing that turns a request into an approval, which is verify it. The owner funds what it can verify, so the memo's job is to make the payback verifiable.

The Memo as a Firm Standard, Not a One-Off

The L3 work introduced the honest ROI memo at the level of a single project: produce it for one owner workshop, justify one stack on one GMP. The firm-strategist move is to recognize that this memo recurs, because the firm will face this same conversation on every owner-funded precon it pursues, and a precon leader who reinvents the justification each time is slower, less consistent, and more exposed to the one memo that overclaims and damages credibility with a repeat owner. So the memo becomes a standard template: a fixed structure the firm fills in per project, with the same sections, the same substantiation discipline, and the same disclosure of limits every time.

A template is worth more than a document for three reasons. First, consistency: when every precon leader uses the same memo structure, the firm speaks to owners with one voice, and an owner who funded the stack on the last job recognizes the same rigor on the next, which compounds trust. Second, defensibility: a structure that has survived owner scrutiny before is one the firm knows how to defend, so the precon leader is not improvising under questioning but working from a format whose every line has already been stress-tested. Third, speed: the firm-strategist does not have hours to rebuild the argument from scratch on each pursuit, and a template lets the leader populate the project-specific hours, rates, and stack while the reasoning and structure carry over, the same acceleration logic the program applies to the work itself, applied here to the firm's own selling instrument.

Treating the memo as a standard also disciplines the firm's claims over time. A one-off memo can overclaim once and be forgotten; a template that overclaims teaches every precon leader to overclaim and trains every owner to distrust the firm, so the template's substantiation discipline is a governance control, not just a writing convention. The firm that standardizes the honest memo standardizes the trust; the firm that standardizes a percentage-driven memo standardizes the skepticism. The memo as a firm standard is the difference between a justification you happen to win and a justification mechanism the firm owns.

An owner does not fund a claim, an owner funds a verifiable accounting, so the firm-strategist's ROI memo is not a one-off pitch but a standard instrument built backward from the owner's verification: every figure traced to hours, rates, and outcomes the owner can check, every limit disclosed, and never a vague percentage that hands the owner the reason to say no.

Substantiating the Payback: Hours, Rates, and Outcomes

The substance of the memo is the payback math, and the discipline is that every figure must be traceable. Take the $90K stack on the $42M GMP. The memo names the stack and its cost line by line: Togal.AI takeoff licenses and seats, the estimate sandbox subscription, ALICE schedule-scenario access, and any integration or onboarding cost, summing to the $90K the owner is being asked to fund, so the cost side is itemized rather than a lump. The cost side is the easy half; the value side is where the substantiation discipline earns the approval.

The value side is built from estimator hours saved at the real labor rate, not from a percentage. The memo states the manual baseline for each task, for instance the estimator hours a manual takeoff of the structural and envelope packages would consume, then the hours the AI-accelerated workflow actually took on this project, and the difference is the hours saved, costed at the estimator's fully-burdened rate. It does the same for the schedule scenarios ALICE generated against the hours a planner would have spent building those scenarios by hand, and for the estimate sandbox against the manual repricing it replaced. Each line is hours times rate, a number the owner can check against the firm's actual labor cost, which is precisely why it is fundable. Beyond the labor hours, the memo quantifies the outcomes the speed enabled that a manual process could not reach: the additional design options priced against the target within the precon window, the tighter cost feedback that kept the design on the $42M number, the schedule the faster precon protected, each stated as a specific outcome rather than a productivity claim.

The 4-to-9-month payback figure the firm anchors to is then a result of the math, not an assertion: the stack cost divided by the monthly value of the saved hours and enabled outcomes yields a payback window, and the firm shows the window falls in the 4-to-9-month range by showing the arithmetic, the saved hours and rates and outcomes that produce it. The owner sees the payback because the owner can rebuild it. A payback the owner cannot rebuild is a percentage in disguise; one the owner can rebuild from the hours and rates is a number the owner can fund.

Disclosing the Limits Is How the Memo Earns Trust

The instinct under a funding request is to present only the upside, but the firm-strategist does the opposite, because the disclosed limit is what makes the rest of the memo credible. A memo that claims the AI saves hours everywhere and costs nowhere reads as a sales pitch, and an owner discounts a sales pitch. A memo that says where the tooling saves hours and where it does not, where the estimator still spends the time, and what the stack cannot do, reads as a clear-eyed accounting, and an owner trusts that. The limit is not a weakness in the memo; it is the evidence that the memo is honest, and the honesty is what funds the approved lines.

Concretely, the memo discloses that the AI accelerates the takeoff but the estimator still owns and verifies the quantities, so the labor is reduced, not eliminated, and the memo's saved-hours figure is net of the verification time the estimator still spends. It discloses that ALICE generates schedule scenarios but the planner still owns the selection and the float reasoning, that the sandbox prices options fast but the estimator still owns the numbers and the variance explanations, the same ownership disciplines the program has insisted on throughout, now disclosed to the owner as the reason the saved-hours figure is conservative rather than inflated. The memo says plainly that the tooling does not replace judgment; it accelerates the work the judgment then verifies, which is exactly why the payback is real and bounded rather than fantastical.

This disclosure is the same discipline as the variance explanations the L3 TVD lesson insisted on, turned on the GC's own claim. Just as the estimator must not let the AI fabricate a tidy narrative for why a cost moved, the precon leader must not let the memo fabricate one for why the tooling pays back, and the defense in both cases is to ground the account in what is real and disclose what is not. The owner who reads the disclosed limits knows the firm is not hiding the downside, which is what lets the owner believe the upside, so the limits are not a concession that weakens the request; they are the substantiation that strengthens it.

The Owner-Approval Mechanics: Structuring the Memo to Get a Yes

A memo that is substantively right can still fail on structure, so the firm-strategist designs the memo's mechanics around how an owner actually decides. The memo leads with the request and the bottom line, the $90K stack, the 4-to-9-month payback, stated plainly up front, because an owner reads the conclusion first and then decides how hard to scrutinize the support. It then lays out the itemized cost, the substantiated value with the hours and rates shown, and the disclosed limits, in that order, so the owner can move from the request to the verification to the honest caveats without hunting for the math. The structure mirrors the owner's reading: tell me what you want, show me the number, let me check it, tell me what it does not do.

The mechanics also account for who in the owner's organization reviews the memo. The capital committee or CFO wants the payback and the verifiability; the owner's project executive wants the schedule and design outcomes the tooling protects; the owner's counsel may want the data-handling and contract terms for the tools inside the GMP. A memo built as a firm standard anticipates these readers and includes the lines each one needs, so it does not bounce back for a follow-up that delays the approval. The firm-strategist treats the approval as a process with multiple gatekeepers and structures the memo to satisfy each, part of why the standardized template beats the improvised one: it has already learned which questions the owner's organization asks.

Finally, the mechanics tie the request to the owner's interest, not the GC's. The framing is not "fund our tools" but "this spend protects your budget and schedule," because the saved estimator hours let the precon explore more options against the $42M target, the tighter feedback keeps the design on the owner's number, and the faster precon protects the owner's schedule, so the $90K is an investment in the owner's outcomes that happens to run through the GC's tooling. An owner funds what serves the owner, so the memo's mechanics make the owner's benefit the headline and the GC's tooling the means, the framing that turns a cost the owner might strike into an investment the owner approves.

The Owner Relationship Is the Real Long-Term Asset

The firm-strategist sees past the single approval to the relationship the memo builds or burns. An owner who funds a $90K stack on a memo whose payback turns out to be real, traceable, and conservative learns that this GC tells the truth about its tooling, and that lesson is worth far more than the $90K, because it lowers friction on every future request and differentiates the firm from competitors who lead with percentages. The honest memo is a deposit in the relationship; the overclaimed memo is a withdrawal, and the firm that understands this treats every memo as relationship capital, not just a project expense to recover.

This is why the substantiation discipline matters beyond the single project. A firm that wins one approval with an inflated payback and then under-delivers has taught the owner to distrust the next memo, so the next stack is harder to fund and the relationship more expensive to maintain, which can cost far more than the one stack it over-recovered. Conversely, a firm whose memos consistently under-promise and deliver builds an owner relationship in which the AI investment becomes a settled, expected part of the precon, no longer fought line by line, which is the position the firm-strategist is playing for. The memo is the instrument; the trust it accumulates is the asset.

The Applied Problem: The AI ROI Memo for a $90K Precon Stack on a $42M GMP

Here is the exercise, from the playbook. The owner is funding the precon on a $42M GMP project, and the GC must justify a $90K precon AI tool stack inside that GMP. Produce the AI ROI memo for owner approval, built as a firm standard: a reusable structure populated with this project's specifics, designed to survive the owner's scrutiny because every figure is traceable to hours, rates, and outcomes the owner can verify.

Produce the memo with these parts. First, the request and bottom line up front: the $90K stack and the 4-to-9-month payback stated plainly. Second, the itemized cost: Togal.AI takeoff, the estimate sandbox, ALICE schedule scenarios, and onboarding, summing to $90K. Third, the substantiated value: the estimator and planner hours saved on each task, the manual baseline against the AI-accelerated actuals, costed at the real fully-burdened rates, plus the enabled outcomes (the additional options priced against the $42M target, the design kept on the number, the schedule protected), with the arithmetic shown so the 4-to-9-month payback is a result the owner can rebuild, never a percentage. Fourth, the disclosed limits: where the estimator and planner still own and verify, so the saved-hours figure is net and conservative, and what the stack does not do. Fifth, the owner-interest framing: how the spend protects the owner's budget and schedule, with the lines each owner-side reviewer needs.

The lasting product is not just this one memo but the firm template it instantiates: a standard AI ROI memo the precon leader can hand to any owner on any owner-funded precon, with the same substantiation discipline and disclosure of limits every time. The professional who masters this funds the firm's AI investment on owner money, repeatedly, because the memo is built backward from the owner's verification rather than forward from the GC's enthusiasm, names the stack and rates rather than a percentage, discloses the limits rather than hiding them, and accumulates the owner trust that makes the next stack easier to fund. The AI accelerates the precon work; the memo substantiates the investment; the owner funds what it can verify, and the firm-strategist makes that verification a standard the firm owns.

Key Takeaways

• When the owner pays for the precon, the AI tool spend (Togal.AI on takeoff, ALICE on schedule, the estimate sandbox) sits inside the GMP as the owner's dollar, so the owner gets to ask "how do I know," and the honest answer to that question is the entire game: an owner does not fund a claim, an owner funds a verifiable accounting.
• The productivity percentage fails reliably because it is a conclusion with no audit trail: the owner cannot trace it to hours, tie it to a rate, or test it against the project, so it reads as marketing, and leading with it signals the GC lacks or is hiding the underlying numbers, which is the inference that gets the line item struck.
• The firm-strategist builds the memo backward from the owner's verification, not forward from the GC's enthusiasm: start from "where does this number come from" and make every line answer it, so the payback is something the owner can rebuild from the hours and rates rather than a figure the owner must take on faith.
• The memo is a firm standard, not a one-off: a fixed template with the same structure, substantiation discipline, and disclosure of limits every time, which buys consistency (one voice to owners), defensibility (a format already stress-tested), and speed (populate the specifics, carry over the reasoning), and disciplines the firm's claims as a governance control.
• The payback is substantiated from estimator and planner hours saved at the real fully-burdened rate, manual baseline against AI-accelerated actuals, plus the enabled outcomes (additional options priced against the $42M target, the design kept on the number, the schedule protected), with the cost itemized to $90K, so the 4-to-9-month payback is a result of the arithmetic the owner can rebuild, not an asserted percentage.
• Disclosing the limits is how the memo earns trust: the estimator still owns and verifies the takeoff, the planner still owns the schedule selection, the estimator still owns the sandbox numbers and variances, so the saved-hours figure is net and conservative, and the disclosed limit is the evidence the memo is honest, which is what lets the owner believe the upside.
• The owner-approval mechanics structure the memo to how owners decide: request and bottom line up front, then itemized cost, substantiated value with the math shown, disclosed limits, and an owner-interest framing ("this protects your budget and schedule") that makes the owner's benefit the headline and the GC's tooling the means, with the lines each owner-side reviewer (CFO, project executive, counsel) needs.
• The owner relationship is the real long-term asset: an honest, conservative, traceable memo is a deposit that lowers friction on every future request and differentiates the firm from percentage-leading competitors, while an overclaimed memo that under-delivers is a withdrawal that makes the next stack harder to fund, so the memo's substantiation discipline is relationship capital, and the artifact is a standard AI ROI memo the firm owns and reuses on every owner-funded precon.