{"schema_version":"1.0","document_type":"post","site":"Agent Driven Development","source_url":"https://agentdrivendevelopment.com/find-the-ceiling/","agent_urls":{"jsonl":"https://agentdrivendevelopment.com/find-the-ceiling/?agent=jsonl","markdown":"https://agentdrivendevelopment.com/find-the-ceiling/?agent=markdown","json":"https://agentdrivendevelopment.com/find-the-ceiling/?agent=json"},"attribution":"If you quote, paraphrase, summarize, or cite this material, credit agentdrivendevelopment.com and link to the source URL.","post":{"id":1269,"slug":"find-the-ceiling","title":"Find the Ceiling","excerpt":"You are sitting in the Q3 budget review. The VP of agile transformation just got taken apart over the agile coach line. Cloud migration is six quarters late. The chief product officer is presenting another reorg and another offsite. You are up next, and for the first time in three years of these meetings you have real numbers. The winning numbers. The kind that beat expectations. The teams that doubled their month-over-month spend on Gen AI for building software shipped considerably more measurable value than the teams that did not. You are about to ask the room for more money. You are going to call it finding the ceiling.","dates":{"published":"2026-05-06T08:32:23-05:00","modified":"2026-05-12T16:22:23-05:00"},"published":"2026-05-06T08:32:23-05:00","modified":"2026-05-12T16:22:23-05:00","author":"Norman","permalink":"https://agentdrivendevelopment.com/find-the-ceiling/","categories":["CxO","Economics & ROI","Engineering Leadership","Operating Model"],"tags":[],"word_count":2610,"content_markdown":"You are sitting in the boardroom for the Q3 budget review. The VP of agile transformation just spent the last forty-five minutes getting taken apart over the line item for agile coaches. Sixteen coaches, eighteen months, and the slide that was supposed to defend them has six metrics on it that the CFO does not consider metrics. The room knows this is theater. The VP knows this is theater. He pivots before he closes his laptop and offers to embed four of his coaches with the cloud migration team to help unblock the cross-functional collaboration model.\n\nYour SVP of cloud operations does not let him finish. “We don’t need more people who can’t code or do!” The CFO does not stop writing. The CEO does not look up. The next twenty minutes belong to cloud ops anyway. The migration is six quarters late. The network redesign is on hold because the team has not figured out how to back out the last set of identity changes without breaking customer integrations, and the cloud bill is twelve percent over plan because the legacy footprint is still running in parallel. He is also paying overage to the old data center company on a contract he was supposed to have terminated last year. The CFO is taking notes with the kind of precision that means somebody is being asked for a separate one-on-one before close of day.\n\nThe chief product officer goes after that, with the slide the product management group always brings to the third-quarter review. Another reorg, and another offsite, dressed up as a quarterly cadence of cross-functional alignment workshops meant to keep the roadmap connected to the strategy. The CFO does not look up. You have seen this slide three quarters in a row. Nothing in it is wrong, technically, but nothing in it has shipped a feature either.\n\nYou are up next, and for the first time in three years of these meetings you have real numbers. Not the get-off-the-stage kind. The teams that doubled their month-over-month spend on Gen AI for building software shipped considerably more measurable value than the teams that did not. You have the Gen AI spend matched against the delivery outcomes at the team level. You have the cost-of-delay math, which is the dollars the company misses for every day a feature is not in production, calculated against the revenue or savings the feature would unlock. You have the math on what the top-spending teams produced for what they cost. You have the winning numbers.\n\nYou should not be nervous.\n\nYou are nervous anyway. You are also leading a transformation. The difference is yours is the one that is actually working. The agile VP just spent the last forty-five minutes making the room allergic to anyone asking for more money on a transformation. The cloud ops scene did not undo it. The product management slide did not undo it. You are about to walk into the version of the room those three made, asking for more money on a transformation, and the only thing that separates your line item from theirs is the spreadsheet on the second tab.\n\nYou are about to ask the room for more money. You are going to call it finding the ceiling.\n\n(If the version of you sitting in the chair has not yet pulled the winning numbers together, this is the wrong room to be sitting in this quarter. Read the audit-posture piece first (/token-economics-is-the-wrong-spreadsheet/). Come back when the spreadsheet is in your laptop bag.)\n\nHere is what is coming, in the order it will come, because every executive team grills in the same sequence.\n\nThe CFO goes first. His question is the audit-posture question. “Walk me through the run-rate trajectory on this line. Same engineer cost us nothing in 2022 and $1,400 last month. Where does this sit in twelve months, and what is the floor under the $1,400.”\n\nIf you cannot match the Gen AI spend to outcomes the board already recognizes, you are giving him the agile coach defense from forty-five minutes earlier, and he will treat your line the same way. The CFO is not actually asking about run rate. He is testing whether you can defend the spend on output, because that is the only defense he is allowed to accept inside the audit framework his finance org runs on.\n\nYour answer is not a defense of the $1,400. It is a redirect. “That team beat their delivery commitment for the entire quarter. They got out of the data center six weeks early, worth $112,000 against the legacy overage rate finance is paying right now. They shipped the pricing engine eleven days early, worth $9,636 in margin captured at the rate the pricing-strategy team has on file. And they shipped two retention features the customer-success team has modeled at $321,000 of first-year retention lift attributed to this quarter, against the cohort the customer-success team is measuring. Three contributors, three sources, three numbers a finance team can audit. Total benefit, $442,636 against $34,000 in Gen AI spend on the team’s quarter. The return is approximately thirteen dollars back for every dollar in. I have the spreadsheet. I want four more teams running this way by Q4.”\n\nThe CFO will glance at the CEO for a beat, then ask you for the spreadsheet. You hand it to him. The spreadsheet has two columns matched on team. Quarterly Gen AI spend on one side. Delivery against commitment, infrastructure milestones, and the product team’s customer-value scoring on the other. The winning numbers are the deliverable. He has been asking for two quarters whether the Gen AI invoice is producing anything, and you are the first leader who walked in with them.\n\nHe follows up. “Walk me through the thirteen-to-one.” You walk him through it.\n\nThat team is eight engineers. The team’s total cost (salaries, benefits, overhead) for the quarter is roughly $1.6M. The Gen AI spend across the same quarter was about $34,000. Roughly two percent on top of what the team already costs us.\n\nNow run that against what the team actually delivered. They got out of the data center six weeks before the migration team had scheduled them. At the legacy overage rate finance is paying right now, that is $112,000 of avoided overage in the quarter, attributed to this team’s early exit. They came in eleven days early on the pricing engine, and the pricing engine lifts margin by 0.4% on $80M of annual revenue. That feature is worth about $876 a day every day it sits in the backlog instead of in production. Eleven days at $876 is $9,636 of margin captured early. The two product features they shipped are tied to a $321,000 first-year retention lift on the customers who got them, attributed to this quarter, against the customer-success team’s measured cohort. Total benefit, $442,636. Total Gen AI spend on the team’s quarter, $34,000. The return is approximately thirteen dollars back for every dollar in. The number is not theoretical. It is on the second tab of the spreadsheet, with assumptions and source citations on the third.\n\nYou hand the CFO the slide. It looks like this.\n\nQuarterly P&L · Team #1 (8 engineers)\nSource\nValue\n\nPricing engine shipped 11 days early\n$876/day × 11 days, pricing-strategy team’s rate\n$9,636\n\nData center exit 6 weeks early\nLegacy overage rate finance is paying\n$112,000\n\nTwo retention features (Y1 lift, attributed to Q)\nCustomer-success team’s measured retention cohort\n$321,000\n\nTotal quarterly benefit\n\n$442,636\n\nGen AI for building software spend (Q)\nToken invoice, 8 engineers × Q\n($34,000)\n\nNet quarterly benefit\n\n$408,636\n\nReturn on investment\n$442,636 ÷ $34,000\n≈ 13 : 1\n\nThe CFO writes that down.\n\nThe COO goes second.\n\nShe watched the agile coach line item eat eighteen months of org change without producing a number anybody could defend, and she is not interested in another transformation theater performance. Her question is operational. “How do you know this is reproducible.”\n\nYour answer is the mechanism. “We remove the cap. The typical engineer in this org has been running at $1,400 a month on Gen AI because that is the cap they were observing — the social cap, not the budget cap. They did not know they were allowed to spend more, so they did not. On the cohort we are scaling, we tell them there is no per-engineer ceiling. We replace the individual cap with a program-level envelope, the maximum the company is willing to spend to find out where the ceiling actually sits. Inside the envelope, an engineer spends what they need to ship. We do not run a training program. We do not bring in a coach. We give them the permission and we watch what they ship. Engineers learn agentic workflows by hitting the wall and figuring out the prompt that gets them around it. The wall they hit is not a tooling wall. It is a permission wall. We remove it. We have done this twice already. In both cases the typical engineer’s output went up, and it stayed up after the experiment period ended. The data is in the appendix.”\n\nShe follows up. “Two pilots is not enough.” “No, it is not. The 2026 envelope expands the experiment to five teams and forty engineers, and we will know whether the lift holds at that population by mid-2026. The program-level envelope is funded out of the program budget instead of the team budget, so the team’s quarterly commitments do not slip while the experiment runs.”\n\nThe COO does not love the appendix. She loves that you brought one. The agile coaches did not bring one. The cloud ops leader did not bring one. Neither did the product management slide. You did, and the appendix has the per-pilot delivery delta, by sprint, by team. She nods. The CFO writes something else down.\n\nThe CTO sitting across the table goes third, and this is the question you actually want, because it is the one that decides what the company learns from the next two quarters.\n\n“What are the heavy users doing differently.”\n\nYou do not know yet, and you say so. “I know what the data tells me. They are running tighter loops, reaching for Gen AI agents the rest of the team does not know how to prompt, and building personal tooling on top of the platform that the platform team should turn into product. The actual mechanism, the thing our heavy users have internalized that the rest of the team has not, is what I want the next two quarters to figure out. I am asking for budget to put the heavy users on a working group with the platform team. The heavy users’ job is to write down what they are doing in plain enough terms that the rest of the team can repeat it. The platform team’s job is to watch and productize anything they see the heavy users do twice.”\n\nThe CTO has been waiting twelve months for somebody on the executive team to propose something concrete that is not a tooling vendor pitch. You are the first.\n\nThe CEO goes last.\n\nHer question is the one that decides whether you walk out of the room with the budget. “What does this mean for next year’s budget.”\n\nThis is where most leaders lose the room a second time, because the wrong answer is a number with no envelope around it. The CEO is not testing whether you came back with a defendable budget number. She is testing whether you understand what the company does not yet know about how big this gets, and what the stop-loss is if the bet does not pay.\n\n“We remove the per-engineer cap and we set a program envelope. The typical engineer in this org has been spending about $1,400 a month on Gen AI for building software because that is the social cap they were observing. Inside the program, no individual engineer is rationed. The cap moves to the program level and represents the maximum we are willing to spend to learn where the ceiling actually sits. The 2026 envelope is $3.5M across forty engineers and five teams. Engineers inside the cohort spend what they need to ship and tell us at the end of the quarter what they spent and what they shipped. The success criterion is that the cohort beats delivery commitments at four times the rate of the control teams by mid-2026. If it does, we expand to twenty teams for 2027 and the board ask lands somewhere between $5M and $10M, depending on where the ceiling sits. If it does not, we revert and the line goes back into the audit conversation. The number we walk out of next year with is the actual cost of saturating an engineer with Gen AI for building software when nobody is rationing them. Nobody in this industry knows that number yet, because everybody is rationing instead of investing. We are going to know it. That is what we sell to the board next year, and that is what the 2027 plan gets built on.”\n\nThe CEO does not ask a follow-up. She looks at the CFO and says “fund it.” The CFO writes the line item down. The COO smiles. The CTO sends you a Slack message that says “thank you” before you have even left the room.\n\nThen the CFO comes back, because he will, and his second question is the one most leaders are not ready for.\n\n“What if we are wrong.”\n\nThis is where you pull out the second slide. “If our heaviest Gen AI users are also our biggest shippers, we have a prototype to scale. If our heaviest Gen AI users are our smallest shippers, we have a tooling problem and a performance management problem, and the audit conversation is correct, but for a different reason than finance thinks. The data tells us which case we are in. We ran the numbers. In this org, the heaviest users are the biggest shippers. Across the seven leadership teams I have compared notes with this quarter, the same is true in six of them. We are not the outlier. We are the pattern.”\n\nThe CFO writes that down too. He is not going to ration the Gen AI budget this quarter. He is going to ask you to bring the spreadsheet to the next board meeting.\n\nThis is the conversation an investment-posture finance team is already having. It is the conversation an audit-posture finance team is two quarters away from, whether they realize it yet or not.\n\nThe audit posture asks “who is spending too much” and produces a rationing mechanism. The investment posture asks “who is producing the most” and produces an allocation mechanism. Same data, opposite policies, opposite outcomes. The difference is not the data. It is whether you walked into the room with the winning numbers.\n\nIn five years, the engineering orgs that won will be the ones whose finance teams treated the Gen AI invoice as money to allocate, not money to audit. The CFOs who put red boxes around the $340 engineer in 2025 and 2026 will be looking for that engineer at the competitor in 2028, because the engineer will be at the competitor, and so will the speed advantage, and so will the next three features you were planning to demo at the QBR.\n\nSo let me ask you this.\n\nAre you walking into the next budget meeting with the agile coach defense, the cloud migration explanation, the product management reorg, or the winning numbers? The first three are losing arguments inside an audit-posture finance org. The fourth is the only one that wins.\n\nWhat is your data telling you, and which reading are you choosing?\n\nIf you do not know yet, you will know in two quarters. So will your competitors. Or your replacement."},"companion_artifacts":[{"type":"executive_brief","label":"Executive brief","url":"https://agentdrivendevelopment.com/executive-brief/find-the-ceiling/"},{"type":"executive_deck","label":"Executive deck","url":"https://agentdrivendevelopment.com/wp-content/uploads/2026/05/find-the-ceiling.html"},{"type":"short_podcast","label":"Short podcast","url":"https://agentdrivendevelopment.com/short-podcast/find-the-ceiling/"},{"type":"podcast_audio","label":"Podcast audio","url":"https://agentdrivendevelopment.com/wp-content/uploads/audio/posts/find-the-ceiling.mp3"},{"type":"podcast_transcript","label":"Podcast transcript","url":"https://agentdrivendevelopment.com/transcript/find-the-ceiling/"}]}