Projects make money. The company needs $1.8M and misses break-even by 2031.
GoodWatchProblem
Per project-month
Your challenge: is the AI cost underrated?
You were right in two ways. Code-first stays cheap; every agentic build does not.
12 delays
a project-month, not 1
We under-counted use.
174 tokens
per activity, measured
Schedule rows are heavier than we assumed.
$1,570
your 9 Oct coding day
17,886 model calls at Opus 5.5 list price; tokens measured.
$580 a delay
an agent loading the whole schedule
Code-first: about $2.20. Shaped like your sessions: $17 to $42.
AI and hosting per project-month: $19 to $117 code-first, bands A to D
- $19 / $46 / $69 / $117: code-first, bands A / B / C / D; band C's price is $8,000
- Blind: $63 at typical; the margins use the higher figure
- An agent that loads the whole schedule: about $7,300, more than the price
- Shaped like your sessions: about $570 a month
One project-month on each build, against your coding month
Code-firstMulti-agent, no toolsAgentic, disciplinedAgentic, undisciplinedShaped like your sessionsBlind agent's estimate
Each step on the scale is ten times the last. USD a month at list prices (house code-first on Sonnet, agentic builds on Opus 5.5), middle of the simulated range. estimate your days: estimate on measured tokens
What the product reads now: cost follows delay analyses, not data volume
- The product no longer reads mailboxes: emails and team chats are never pulled in automatically checked: scheduler decision log DL-023
- It reads the schedule, the delay report, documents the reporter attaches and meeting minutes
- Cost follows delay analyses, not data volume.
You were right in two ways
- We under-counted use: plan on about 12 delay analyses a project-month, not 1
- Schedule rows are heavier than we assumed (measured: see the chart)
- The 12 is the test bank's design density (20 delays planted in 7 weeks), not an observed rate
- Your original figure was 1 to 2 big delays; Janika scopes the first product to big delays only (as relayed, not read at source), nearer the Light column. We carry both.
- If big delays run 1 to 2 a month, the checked decisions included at bands C and D (4 and 6) exceed demand: safe on cost, but size the offer after buyer interviews
Measured: a schedule never fits in one read
Code must cut the schedule down in every build. measured: Anthropic token counter, 11 Oct
- Caveat: those rows carry names, tags and link reasons; the packet the product sends may be smaller (not yet measured)
- Code-first assumes 45 tokens an activity; at the measured 86 it adds about $0.04 an analysis
Your coding day is a different machine
- 9 Oct: 17,886 model calls, each re-reading about 226,000 tokens: about $1,570
- 10 Oct: 12,546 calls, about $1,150
- An agent re-sends its whole conversation every turn, and the conversation grows
- About 118 turns a session; 151 sessions and sub-agents that day
- Half is re-reading the conversation; 5% is the model's own words
Two ways to build it: re-read everything, or read one packet
Tokens are the units a model reads and writes; it is paid by the token. estimate
One delay analysis, 9,000 activities
House P50; circles are the blind agent. Shaped like your sessions (226,000 tokens and $0.088 a call): $17 to $42 a delay. estimate
What the rulebook allows
- Agentic builds are forbidden: "the AI has no tools and never acts" checked: SPECS.md, Security Invariants
- Multi-agent without tools breaks today's $5 per analysis cap at its central case checked: SPECS.md, Numeric Constants
- Test set before each release (140 delay reports x 10 runs): about $3,100 code-first, $18,600 multi-agent, $100,000 disciplined agentic estimate
- Next measurement: one code-first analysis on a 9,000-activity schedule, timed and costed, before any figure leaves the founders
AI and hosting per project-month, house P50 / blind central
| Build | Light: 2 delays, 5 people, schedule and typed report only | Typical: 12 delays, 10 people, attached documents and minutes read | Heavy: 30 to 50 delays, 30 people, 50,000 activities |
|---|---|---|---|
| Code-first | $16 / $8 | $46 / $63 | $193 / $440 |
| Multi-agent, no tools | $48 / $18 | $196 / $124 | $1,464 / $693 |
| Agentic, disciplined (lead and sub-agents with tools) | $229 / $32 | $1,396 / $374 | $12,377 / $2,696 |
| Agentic, undisciplined (loads the whole schedule every turn) | $1,221 | $7,343 | $24,910 |
| Shaped like your coding sessions (blind) | about $570 |
- Why they differ: the house agentic build assumes 6 sub-agents of 12 turns, 2 critics and 2 re-runs a delay at high effort; the blind one fewer agents, turns and re-runs and a smaller prompt
- The 8x agentic gap is not reconciled: treat the house agentic figures as an upper bound
- Both agree on the order: code-first cheap, agentic many times dearer
- At code-first the gap ($46 house, $63 blind) is mostly users' questions: about $0.04 a question in the house model (Sonnet, cached) against about $0.10 in the blind (Opus, uncached)
- Not in these figures: retries beyond 5%; OCR and speech-to-text for scans and voice notes; 1.1 times for US-only processing
- Development AI sits in team cost on a flat subscription; at API list price your coding days would be $35,000 to $47,000 a month
AI cost calculator
Pick a build and a use level, then move any dial. The bill is AI plus hosting for one project, a month.
your settings, a month
AI and hosting for one project.
one delay analysis
Against the rulebook's $5 cap an analysis.
of band B's $4,000
AI and hosting as a share of the price.
one more person
At your questions a person, a month.
Your settings
Build
Use level
Documents attached to each delay
On: the AI reads the documents the reporter attaches to each delay. Light: off; Typical and Heavy: on.
Opus 5.5 list prices. Unit costs: the middle of 600 simulated runs; straight lines between schedule sizes.
Presets add up those unit costs, so they differ a little from the table in AI cost.
Where the money goes
The bill against each band's price
Dashed lines are the monthly prices of bands B, C and D.
Users: limited or unlimited?
Both are industry practice, depending on who uses the tool. Where users are unlimited, the AI is usually metered.
7
whole-team tools checked
All offer unlimited users.
Metered
AI, when users are unlimited
Unmetered AI usually sits behind a paid seat.
$2.27
one extra heavy person
Code-first, 60 questions a month; $134 agentic.
75% and 90%
alerts, then a stop
Questions stop; the core delay analysis never does.
Which is industry practice? Both, depending on who uses the tool
- Tools a whole site team uses: unlimited users. Tools one specialist drives: per named user (P6, Foresight, Bluebeam, Fieldwire, SYNCHRO) checked
- A delay-analysis tool sits between the two: a planner runs the analysis; managers, owners and subcontractors read the results judgement
- Paid named seats for specialists are normal: Foresight, the nearest data-centre competitor, £24,000 a named user a year; P6 £244 a user a month, minimum 25 checked
- Cheap or free viewer seats exist: P6 progress reporters £26 a user a month; Intercom's free "Lite" seats (outside construction) checked
- Combining the two in a construction tool is our proposal, not an observed norm judgement
- What would be unusual is limiting who can see the options
Three ways to set it: founders to choose
(a) Unlimited viewers plus a fixed AI allowance per project
- Everyone can view and approve
- One AI pool per project
(b) A limited number of named seats, each with its own allowance
- Only named people use the AI
- Each seat has its own pool
(c) Unlimited viewers plus a few named analysts who can start AI runs
- Everyone views
- A few people spend the allowance
- In (b) and (c) each seat still needs its own allowance
- The allowance: alerts at 75% and 90%; at the limit, questions stop; the core delay analysis never does, so the next-working-day promise holds; the customer chooses auto top-up or stop
- This reverses a deliberate rulebook choice: today the monthly count alerts the team "rather than block the customer" checked: SPECS.md, Numeric Constants
What one more person costs a month
Questions only, 9,000-activity schedule, middle of the simulated range. estimate
How the allowance works
Works with any of the three options, (a), (b) or (c). judgement
What the market does
- Unlimited users is standard for tools a whole project team shares checked: pricing pages, 11 Oct
- Per-seat survives in tools one specialist drives: Primavera P6 £244 a user a month, minimum 25; Foresight £24,000 a named user a year
- Where AI is unmetered, a paid seat usually bounds it (Intercom, Salesforce, Microsoft 365 Copilot); Outbuild includes AI agents with unlimited users and publishes no limit
- Where users are unlimited, the AI is usually metered: Procore says "We don't charge by the seat" and sells credits per task checked
- Flat fees without limits lose money: GitHub Copilot reportedly lost $20 a user a month, up to $80 on heavy users one anonymous source; Microsoft disputed it
What drives our AI bill
- Analyses and questions, not log-ins
- One extra person asking 60 questions a month: $2.27 code-first, $4.54 multi-agent, $134 agentic; 12 times a light user estimate
- Your fear is right for one design: an open AI box for everyone, built agentic
- A seat limit alone does not fix it (GitHub, Cursor); a monthly AI allowance does
Evidence grid: who allows unlimited users, and how AI is limited
| Product | Unlimited users? | How it prices | How AI or usage is limited |
|---|---|---|---|
| Procore | Yes | Company fee by construction volume | AI credits per task: 5 to 15 standard, 50 to 100+ complex; prepaid, no rollover |
| SmartPM | Yes | Per project slot, yearly | No AI limit published |
| ALICE | Yes | By project size and scope | No AI limit published |
| Planera | Yes | Tiers by portfolio size | No numeric limits |
| Sitedrive | Yes | Per active project a month | Spaces per project |
| Outbuild | Yes | Company construction volume | AI agents included; no AI limit published |
| Doxel | Yes | Per square foot | Not stated |
| Primavera P6 | No, minimum 25 | Per named user a month | n/a |
| Foresight | No | Per named user a year | n/a |
| Intercom Copilot | Per agent seat | $29 an agent a month | Unlimited use, behind the paid seat |
| Salesforce Agentforce | Per user add-on | $125 a user | Unmetered for employees, behind the paid seat |
| Microsoft 365 Copilot | Per user | Per-user licence | Agent use included in the licence |
| GitHub Copilot | Per user | Per user plus AI credits | Budget alerts at 75%, 90% and 100% |
What we sell, the price, and why
Our draft prices, and beside them a second case priced against the schedulers the AI replaces.
$12,000
90-day paid pilot
$4,000 a month; Janika checks 2 decisions a month.
$8,000
band C, a month
Contracts of $250M to $1B; 84% margin after the pilot.
$18,333
second price case, a month
Half the cost of two schedulers; 92% margin.
10 to 35x
SmartPM's price, bands B to D
The premium has to be the AI doing the scheduler's work.
In short: what we sell, how it is delivered, the price
- An AI that does the construction scheduler's job, starting on data-centre builds
- Keeps the schedule current, catches slips, gives costed recovery options and the delay statement within a day
- A person at the contractor approves every option; the delay decision is the first job
- 3-month paid pilot: Janika checks the AI's decisions
- After it, the AI runs on its own beside the client's planners and takes over more scheduling work over time; we review only options the client flags
- Seats versus allowance is the founders' call
- $12,000 paid 90-day pilot, then $4,000 / $8,000 / $15,000+ a project-month for contracts of $50M to $250M / $250M to $1B / over $1B
- Band A ($1,500, under $50M) stays as a software-only tier, sold later
- Second price case, beside ours: about $18,300 a project-month, half the cost of the contractor's two schedulers
Two price cases, side by side: price and gross margin a project-month
Founders' draft pricesSecond price caseGross margin after the pilot
- Second price case: a data-centre contractor runs about two schedulers on a job at about $220,000 a year each, loaded; charging half their cost is about $18,333 a project-month estimate
- 1.2 times band D, 2.3 times band C, at a 92% margin once the AI runs on its own
- It fits the end-to-end system, not the delays product alone, and best on projects over $250M
- YC partners argue AI is paid from the wage budget, not the software budget (What we tell YC)
Two units, never mixed
What each band includes
| Pilot (90 days) | Band A, software only (proposed) | Band B | Band C | Band D | |
|---|---|---|---|---|---|
| Contract value | any | under $50M | $50M to $250M | $250M to $1B | over $1B |
| Price | $12,000 ($4,000 a month) | $1,500 a month | $4,000 a month | $8,000 a month | from $15,000 a month |
| People who view and approve | unlimited | unlimited | unlimited | unlimited | unlimited |
| Software analyses included a month | 15 | 5 | 15 | 30 | 60 |
| AI questions included a month (only if a question box is added) | 600 | 150 | 600 | 1,500 | 3,000 |
| Checked decisions included a month | 2 (Janika) | none | none | none | none |
| Our AI and hosting cost, typical / allowance fully used (P50) | $46 / $67 | $19 / $24 | $46 / $67 | $69 / $168 | $117 / $327 |
| AI cost the margin uses (higher of house and blind) | $63 | $26 | $63 | $95 | $161 |
| Gross margin as offered | 52% | 77% | 71% | 84% | 91% |
The band ladder: contract value sets the price
- US data-centre starts, January to July 2026: $618M on average across all 136 starts ($84.1B), $1.06B across those with a published price; median not found checked: ConstructConnect
- Projects are getting bigger fast (701,000 sq ft against 454,400 in 2025), so C and D are where we sell first and D is growing
- Only about 30% of US data-centre projects are over $250M (15% to 50%): the value sits in a few giant campuses shares: estimate
Against competitors, a month
SmartPM includes time impact analysis, with managed services as add-ons. Outbuild is company-wide. checked
Why the price holds, if it does
- Per project, by contract value: bigger jobs mean bigger schedules, more delays, more people, more at stake
- Against one more planner: band D ($180,000 a year) is about one loaded planner; band C about half of one judgement
- Against one day of delay: site overheads about $41,000 to $82,000 a day; a month of band C is a tenth to a fifth of one day speculative
- Extra checked decision: test $2,500 to $10,000; our cost about $612 (4 hours at $153; $1,224 at 8 hours)
- The buyer's in-house cost (2 to 3 planner days): $1,600 to $2,400 at $100 a paid hour, $2,450 to $3,670 on our $153 delivered-hour basis; buyers will cite the lower: the premium must come from days recovered
- The independent strategist suggested $10,000 to $25,000 on value judgement
Band A stays
- Kept (Munim, 11 Oct: "we may eventually be selling to them")
- Software only, sold later, is the research's proposal judgement
- $1,500 a project-month for contracts under $50M; no checked decisions included; unlimited viewers
- An allowance of 5 analyses and 150 questions a month (about $24 of AI and hosting when fully used) estimate
- At 4 analyses and 1 hour of support a month it keeps 77%: AI and hosting about $19 house, $26 blind-scaled, which the margin uses; AI alone about $9
- At most about $14,000 to win a customer (12-month payback): it needs self-serve sign-up, not founder-led selling
- With full support (4 hours a month) it would keep only 26% use: judgement
- At 3.6 times SmartPM with no checked decision, its premium has to rest on data-centre rules and costed options judgement, for the founders
What is out
- Deep (multi-agent) analyses are not in the allowance unless the founders raise the rulebook's $5 and $3 caps: they stay within both in only about 3% of runs estimate
Revenue types
- Subscriptions are recurring
- Pilots and extra checked decisions are services revenue, which investors leave out of ARR
- The checked decisions inside the subscription are also a service; investors will carve them out how the bundle splits: unverified
- A company-wide annual contract comes once a contractor runs 3 or more projects
Unit economics calculator
After the pilot the AI runs beside the client's planners, so the margin is set by the AI, not our hours.
84% (52% pilot)
gross margin, band C
After the pilot; the pilot at a hired planner's rate.
12 months
payback, band C
Pilot and its credit included; $60,000 to win.
7.3
LTV to CAC, band C
2.7 if one project only; 2.0 at the median cost to win.
5.2 hours
a decision, at most
Only if a customer needs our checks every month: band C below 50% margin above it.
Why 84%: price minus what one project costs us, band C after the pilot
Your assumptions
The project's band (price a month)
After the pilot
The plan: the AI runs on its own beside the client's planners; we review only options the client flags. What if: a customer still needs our planner's checked decisions every month. The pilot is always checked by Janika.
Pilot credited to year one
On: the $12,000 counts against year one, so those months are served unbilled.
Gross margin as offered, at your hours and rate
Pilot: Janika checks 2 decisions a month, costed at a hired planner's rate. After the pilot: the AI runs alongside; we review flagged options. estimate
From price to gross profit, a project-month
Lifetime value against cost to win
Who checks the decisions, and what it costs estimate
- Janika checks only during each customer's 3-month pilot (Munim, 14 Oct)
- No extra cash in 2027 and 2028: her founder pay is already in the team cost; at 60 hours a month she covers about 3 pilots at once
- Our delivery hours stay small: about 15 to 33 hours a month across all pilots from 2028 to 2031 estimate
- How a check is costed: 4 hours a checked decision at $153 an hour (a $220,000-a-year senior planner delivering 120 hours a month)
- The 4 hours is a guess: Janika's experience by hand today is "multiple days"; only a pilot measures it
- Investors judge the pilot at a hired planner's cost (52%); the company's margin is set by the months after it: B 71%, C 84%, D 91%
- Will a contractor let our planner see its schedule? Delay analysis is bought from outside firms today
- One public example: Minnesota's transport department (an owner, not a contractor) paid WSB LLC $69,515 for CPM schedule review and time impact analysis on one highway project checked
- SmartPM sells managed services as add-ons to its schedule software checked: smartpm.com/pricing
- Whether a contractor will let a new company do it is unproven, a first question for buyer interviews; if not, the contractor's own planner checks from the start
- Our rulebook also limits our access today checked: SPECS.md, Privacy Rules
Gross margin by delivery and band, $153 a delivered hour
| Case | Our hours a month | B $4,000 | C $8,000 | D $15,000 |
|---|---|---|---|---|
| After the pilot: AI alongside | 7 | 71% | 84% | 91% |
| 2 checked decisions | 12 | 52% | 75% | 86% |
| 4 checked decisions | 20 | 21% | 60% | 78% |
| 4 checked decisions at 8 hours | 36 | minus 40% | 29% | 61% |
| 12 checked decisions | 52 | minus 101% | minus 2% | 45% |
If a customer needs our checks every month
- The margins rest on the AI running on its own after the pilot. If it cannot, the number that decides everything is our hours per checked decision
- At 4 decisions a month, 50% margin needs under 2.1 hours at band B, 5.2 at C, 10.7 at D; margin reaches zero at 5.4 / 11.8 / 23 hours. Nobody has measured it; the first pilot must
- Why $153, not $100: a planner is paid for every hour but delivers about 120 a month. At $100 the 4-decision row is 47% / 73% / 85%
- Band A, software only: 77% at 4 analyses and 1 hour of support; payback 55.9 months at a $60,000 cost to win, so at most about $14,000 to win one
- Band B with our checks every month works only at one decision a month: 67% and 27 months' payback; with four decisions, 21%
- Planner capacity: at 20 hours a project-month, one planner carries about 6 live projects
- At a typical software company's cost to win: LTV to CAC 1.7 to 2.1 with the AI alongside, 1.4 to 1.6 with our checks every month; payback 44 to 51 months. The business needs founder-led, efficient selling
What each number means, and what good looks like
| Metric | In plain terms | Ours, band C after the pilot (with our checks every month) | What good looks like |
|---|---|---|---|
| Gross margin | Of each dollar, what is left after delivering it (AI, hosting, our hours, fees) | 84% (60%) | 65% to 70% over time (Bessemer); AI products 45% to 60%; Procore 80% |
| Our hours a project-month | Our review of flagged options plus support (checked decisions) | 7 (20) | Must fall each month |
| CAC | All-in cost to win one paying customer | $60,000 founder-led; $121,000 top quartile; $221,000 median | Judged by payback |
| CAC payback | Months of gross profit to earn back the cost to win, pilot and its credit included | 12 months (18); 51 at the median cost to win | Under 24 enterprise, under 18 mid-market (Bessemer); median 17 to 22 (Benchmarkit) |
| LTV | Gross profit a customer brings over 5 years | $439,000 ($310,000); one project only about $162,000 | n/a |
| LTV to CAC | Lifetime gross profit per dollar spent winning the customer | 7.3 (5.2); one project only 2.7; at the median cost to win 2.0 | Over 3 (Skok); median 3.6 (Benchmarkit) |
| NRR | Revenue kept from last year's customers, counting new projects they add | 104% assumed | Median 101% to 104%; Procore 106% |
| Operating margin | Profit after the whole team and selling | negative to 2031 | Procore +14% non-GAAP, minus 9% GAAP |
- The $60,000 counts founder time and pilot delivery that also sit in team cost and pilot cost: conservative, counted twice
Gross margin versus profit, five years
Each project makes money; the company loses money to 2031 on this plan. That loss is what we raise for.
lost through 2028
lost to 2031, in all
Cumulative operating loss.
live projects to break even
to $2M
pre-seed at this team plan
YC's standard $500,000 is under a third.
Why the company loses money: live projects against the cost base
Revenue, gross profit, operating profit and the cumulative loss
RevenueGross profitOperating profitCumulative profit or loss
Live projects against break-even
The path, in short
- 3 / 5 / 7 / 9 / 11 new contractor customers a year, each starting with one project and growing at NRR 104% judgement
- Live projects at year end: 3 / 8 / 15 / 25 / 37
- The path assumes each customer books its next project as the last one ends: 52 in 2031 needs NRR about 127%; 110% gives 40
- 2027 is a pilot year billed at $4,000 a month; from 2028 every new customer starts with the 3-month pilot (Janika checks, fee credited to year one), then the AI runs on its own
- Janika delivers up to 60 hours of checks a month in 2027 and 2028 at no extra cash cost; our delivery hours stay at 15 to 33 a month
- Band mix B 30%, C 40%, D 30%; band A is not in the path (sold later)
- Team 3 / 7 / 11 / 15 / 19 people costing $0.45M to $4.6M a year; $60,000 to win each new customer
Two margins, two questions
- Gross margin is per project: price minus the cost of delivering it
- Operating margin is the whole company: gross profit minus the team, selling and overheads
- Break-even needs about 57 live projects at the 2031 cost base, against 37 on the path estimate
- Profit needs faster customer growth, more projects per customer, the second price case, or construction beyond data centres (Market)
- The margins rest on the AI running on its own after the pilot; if a customer still needs our checks every month, its margin falls to the checked-decision rows in Unit economics
What the pre-seed buys
- YC's standard deal: $500,000 ($125,000 for 7% plus $375,000 on an uncapped SAFE) checked: ycombinator.com/deal
- The demo at real schedule size
- 3 paid pilots; 8 live projects
- Measured hours per checked decision
Change the path: type into the boxes
| Year | New customers | Share run by the AI alongside | Team cost a year | Live projects at year end | Revenue | Gross margin | Gross margin, hired planners only | Operating profit | Cumulative |
|---|
- estimate
The business models, as a pyramid
Sell the foundation now; the apex only once the layers below work.
7
layers, foundation to apex
Commissioning readiness is layer 4.
52%
the checked pilot
Janika's checks at a hired planner's rate.
71% to 91%
the AI alongside
After the pilot, bands B to D.
2 or 3
pilots at most
Measure our hours every month.
Click or tab to any layer. judgement: independent strategist merged with the platform path
Why the pilot is checked by us
- It sells with the least product, teaches us the work and earns the trust that lets the AI run on its own afterwards
- Warren (YC): "You have to sell outcomes, not seats or tokens." checked
- Two YC request rounds asked for AI that does the service (Spring and Summer 2026)
The trap
- Our people quietly covering product gaps
- Measure our hours per decision every month
- Cap pilots at two or three. Warren: "cap your first pilot customers to a small handful" checked
What the rulebook needs
- Today the founders may see customer data "only for options flagged as wrong" or for support checked: SPECS.md
- The checked pilot needs a recorded, customer-switched-on "our analyst on this project" role that keeps the shared-library firewall
All nine models the strategist scored, for fit now
| Model | Who pays | Price unit | Gross margin | Fit now | Main risk |
|---|---|---|---|---|---|
| Checked delay decisions | Contractor's project director | In the paid pilot; extras per decision | 52% in the pilot (ours) | People covering product gaps | |
| Pay per delay analysed (self-serve) | Project director or planning lead | Per analysis | Not found | Lumpy revenue; delays are irregular | |
| Per-project software, the AI alongside the client's planners | Project director | Per project a month | 71% to 91% (ours); Procore 80% | Feature fight with SmartPM and ALICE | |
| Owner-mandated programme | Owner or owner's representative | Per campus or portfolio | Not found | Our delay statement argues against the owner | |
| Claims and entitlement support | Contractor's commercial director | Per claim | Diales 25.3% | Legal exposure; needs named independent experts | |
| Company-wide licence | Contractor head office | Annual enterprise licence | About 80% (Procore) | Sales cycle longer than the runway | |
| Benchmarks and risk scores | Contractors, owners, lenders | Subscription | Not found | Needs written permission from each company and client | |
| Delivery assurance or insurance | Owners, lenders, contractors | Premium on contract value | Not found | Carrier, loss history, regulation, capital | |
| Operating system for critical builds | Contractors and owners | Platform plus usage | Procore 80% | Breadth before depth; crowded |
What we tell YC
-
- Problem: a late chiller or switchgear package stops the critical path; finding what can move takes days of planner work in P6
- Janika: a planner "has to manually think about it for multiple days". No buyer has confirmed it yet
-
- Why now: $84.95B a year, up 73% checked: Census; projects get bigger, not more numerous: about 20 to 23 US starts a month, 701,000 sq ft on average in 2026 against 454,400 in 2025 checked: ConstructConnect
-
- Why us: Janika plans these jobs and knows which links can really move; Munim builds it. Blomfield: "70 team, 20 idea, 10 traction"
- First: the scheduler's job on data-centre builds, after a paid pilot Janika checks
-
- Price: $4,000 to $15,000 a project-month by contract size, about half to one planner
- Gain: a delay analysis that takes two or three planners a week, done in hours
- How YC sizes markets: Seibel prefers bottom-up; Graham: be matter-of-fact and disclose the flaws; partners argue AI is paid from the wage budget, about 4% to 10% of a customer's spend against about 1% for software
-
- Market, last: about 450 US data-centre projects (about $31M a year, up to about $80M as projects grow), then every large project in the US and Europe, about $1B a year; $100M a year is 10% of it
- Say plainly that data centres alone are small: that is why they are the start, not the market
- Where it goes: the same system on every large construction project in the US and Europe
$84.95B
US data-centre build, a year
Up 73% (Census, Aug 2026).
$31M
US data centres, a year
At our prices; up to about $80M as projects grow.
$1B
every large project, US plus Europe
$100M a year is 10% of it.
$500,000
YC's standard deal
$125,000 for 7% plus $375,000 on an uncapped SAFE.
"An AI that does the construction scheduler's job, starting on data-centre builds. It keeps the schedule current, catches slips and gives the contractor costed recovery options and the delay statement for its client within a day; a person approves each one." draft for the founders; a design target until the demo times it at real size
The story in five steps
YC's own request, and the YC companies a partner will name
- Spring 2025 "Datacenters" request asked for data-centre "project management"; none of YC's 40 construction companies from 2025 and 2026 has scheduling as its main product checked
- "AI agents that can quote complex work and schedule teams" (Fall 2026, Warren) checked: ycombinator.com/rfs: it is about crews, not planning teams. Do not overstate it
- Friedman calls project-management tools "simple targets": never pitch a dashboard. Warren: "You have to sell outcomes, not seats or tokens."
- Applied Kinetics (F26): agents that "chase suppliers, track equipment approvals and delivery dates"; Marengo (S26): an AI engineering firm for data centres; Karmen (F24): "AI scheduling assistant for construction"
- Ours starts from the reported delay and ends with a checked recovery plan; one line each, before the interview
The two objections
- "ALICE and SmartPM do this": SmartPM sells time impact analysis and managed services; ALICE claims a data-centre recovery "Within hours" checked
- Our answer: link meanings from a planner's rules; every option costed and checked by code; the statement kept separate from the recovery, so entitlement is never given away
- Proof that settles it: a contractor who has seen them and pays us
- "Isn't this a consultancy?": only the pilot is checked by us; after it the AI runs on its own
Avoid, and say plainly
- Avoid: "a $16 trillion construction industry" (industry size, not our market), project-management language, per-seat pricing, and the flagged words ("wedge", "AI-native", "agentic", "AI-powered platform")
- The 28 Oct demo uses the 138-activity schedule; real ones run 9,000 to 50,000. Say so, and name the size the product is built for
Prove before 2 Nov
- Five or more contractor conversations with numbers, before 2 Nov
- One written commitment: a letter of intent, a paid pilot, or a re-analysed past delay
- Before the interview: one checked decision delivered for money, with hours, days recovered and our hours spent
Pushback: what we disagree with, in short
Nine claims we push back on, each with our answer and a picture.
$580 a delay
an agent loading the whole schedule
About $7,300 a typical month; code-first about $2.20 a delay.
$14.2M
a month: the owner's loss
Not the contractor's.
$30M to $80M
US data centres, a year
For this product, not the $85B of building.
5 to 10 yrs
to replace P6
The last step, not the first.
Market, counted bottom-up
Live projects x price a month x 12. Data centres are the start, not the market.
$31M
SAM: US data centres, a year
At our draft prices; $18M to $49M.
$971M
TAM: US plus Europe, a year
End-to-end system, every project over $1M; $1.9B on all live projects.
31 projects
SOM: year 5
$2.3M a year at our prices; $6.9M at the second price case.
$5B
the bar a venture fund uses
Nothing clears it; $100M a year is 10% of the US plus Europe market.
SAM: the market we serve first, US data-centre projects. TAM: the same system on every construction project over $1M. SOM: what we could realistically win.
Why $31M: live projects x price a month x 12
The market ladder: each step adds smaller projects, against the $5B bar
Central case, a year; each step on the scale is ten times the last. Below $50M the delays product uses market prices, since the founders' $1,500 is 6 times the evidence there. estimate
US data centres by year: the count stays flat
Delays product, our pricesEnd-to-end system
Live projects with a critical-path schedule above each year. Prices held at today's size mix. estimate
2031 if projects grow bigger: the growth is in size
- US starts have run at 20 to 23 a month since late 2024 while the dollars rose about 5.5 times checked: ConstructConnect
- About 412 live today, 460 at the 2029 peak, 416 in 2031
- If every project paid the top price: about $80M a year
Try it: US data-centre SAM
Price case
Year
How it adds up
Live projects x (share over $250M x top price + the rest x the $50M to $250M price) x 12. estimate
SAM, TAM and SOM
| Step | What it counts | US | US plus Europe |
|---|---|---|---|
| SAM | Data-centre projects, delays product, our draft prices | $31M ($18M to $49M) | $41M |
| Same, end-to-end system, market prices | $43M | $57M | |
| Same, second price case ($18,333 a month) | $91M ($64M to $116M) | ||
| TAM | End-to-end system, every project over $1M with a critical-path schedule | $477M | $971M |
| Same, every live project over $1M | $923M | $1.9B | |
| SOM | Year 5: 31 US data-centre projects | $2.3M at our prices, $3.2M end-to-end | |
| Year 10: 96 projects | $7M to $10M; $21M at the second price case |
What sets the size
- Big projects are a minority: about 30% of US data-centre projects are over $250M (15% to 50%) estimate
- July 2026 had 23 starts worth $2.7B in total, so at most 4 were over $250M checked: ConstructConnect
- Our draft prices with half the over-$250M projects paying $15,000: $31M; $26M if none does
The second price case
- Fits only the end-to-end system: the delays product leaves the scheduler in place
- Best on projects over $250M: on those alone, $36M in the US
- It assumes $220,000 loaded per scheduler load factor: unverified
- That needs a salary near the top tenth of construction project managers: 90th percentile $160,510, mean $104,890 checked: BLS May 2025
Against the bar
- A fund looks for a market of at least $5B, ideally a path to $15B, and $100M a year of revenue checked: Euclid Ventures
- Nothing clears $5B in the central case
- $100M a year is 10% of the US plus Europe end-to-end market, 21% of the US one
Not counted, and the weakest inputs
- Not counted: small builders' job-management software (a different product, about $147M in the US), home builders ($564M), China
- Weakest input 1: the share of projects over $250M (no project database pull by value band)
- Weakest input 2: the share we win: "about 60 data-centre contractors" is the length of a published top-60 list, not a count
- Weakest input 3: every price is a desk estimate until buyers name one
Buyers and downturn
- Buyers are concentrated: about 15 US general contractors build most of the volume judgement
- Downturn risk: 75 projects worth $130B blocked or delayed in Q1 2026 checked via Bain
Risks investors will cite
Seven questions a partner will ask; each needs an answer before the interview.
3 months
of pilot, then the AI runs alone
Every margin after the pilot assumes it can.
138
activities in the demo
Real schedules run 9,000 to 50,000.
1
model provider
A price or terms change hits every project.
Team
- Full-time status and the CTO's depth
Whether the AI can run on its own after the pilot
- The margins assume it
- If a customer needs our checks every month, band C falls to 60% at 4 hours a decision, below 50% above about 5.2 hours, and to zero near 12 hours
Accuracy at real size
- One impossible option, or dates that do not match P6, ends a pilot; the demo is 138 activities
Liability and who signs
- Our planner checks in the pilot; the contractor's project lead signs
- If our planner signed the statement the contractor submits, we would author a contractual submittal
- Contract terms, professional liability insurance and who signs need counsel before the first paid pilot
Revenue ends with each project
- Fees stop when the project ends, and data-centre starts can stall
One model provider
- Top-tier list prices flat to rising since 2025; a price or terms change hits every project judgement
Access to customer data
- A contractor may not let our planner see its schedule (the founders already expect clients to resist automatic reading of their emails and chats: scheduler decision DL-023); then checked decisions become self-serve
Decisions and unknowns
Five founders' decisions, for after you have both read this; the gaps only buyers and a pilot can fill.
5
founders' decisions
None needs an answer now.
0
buyer interviews yet
Price, signer and delays a month are open.
0
pilots run
Only a pilot shows the AI can run on its own.
Founders' decisions, for after you have both read this
- Settled by Munim on 14 Oct, the framing: an AI that does the scheduler's job, data centres first
- Settled, the delivery: Janika checks in the pilot, then the AI works alongside and replaces scheduling work over time
- Settled: the second price case shown beside the draft prices
- None of the five below needs an answer now.
- Users: unlimited viewers plus an AI allowance, limited named seats each with an allowance, or unlimited viewers plus a few named analysts who start AI runs?Users: both norms exist
- Band A: kept (Munim, 11 Oct)Software only, sold later, is the research's proposal
- Which AI build?Research recommends: code-first by default; multi-agent only if it proves better on the test set and the caps are raised; never an open agentic chat
- Rulebook changes for Janika: an AI stop at the limit that spares the core delay analysis (reverses "alert rather than block")Also: deep-tier caps; the "our analyst" role for the pilot; a question box or not
- Still open: the product name before YC; who signs (counsel; Risks); whether the draft prices move toward the second price case once pilots measure the value
Not established
- No buyer interview yet: willingness to pay, who signs, delays a month, planner hours per recovery, damages rate and cap, AI acceptance by owners
- Whether the AI can run on its own after a 3-month pilot: the delivery model; only pilots show it
- Our hours per checked decision: only a pilot measures it. A planner's delivered hours a month (120 assumed) sets the hourly cost
- Janika's delivery hours a month (60 assumed) set how many early projects cost no extra cash
- Whether contractors will let our planner see their schedules
- Real cost per analysis on a 9,000-activity schedule (row size measured; the product's packet and a full run not yet)
- Real schedule size: no published activity count for a real data-centre schedule; our bands assume 9,000 / 25,000 / 50,000
- Oracle's P6 sizing guide tiers projects at 7,500 / 25,000 / 75,000 activities; realistic schedules carry 2 to 4 links an activity secondary: test-bank research
- Delays a month on a real job: about 13 delay-causing issues per weekly report on one London high-rise (not data-centre), most of them small secondary: test-bank research
- Unknown: real CAC; pilot conversion in construction; projects per customer; how investors value per-project revenue; how the bundle splits between software and service revenue
- Not found: prices for ALICE, nPlan, Buildots, Document Crunch; US delay-consultant rates; the median data-centre project value
- Not found for the market: the share of projects of each size that run a formal critical-path schedule (the input that swings it most); the size mix of US data-centre projects (30% over $250M assumed)
- No product code yet: specification, checks, research and the test bank; demo due 28 Oct