The living enablement bible and the org-build blueprint. Disciples of MEDDPICC. We don't sell a faster horse — we sell the day the work does itself. v0.1 · scaffold · Jun 2026 · pre-launch.
North Star
We are not competing with Procore. We're selling the future they can't build.
Perch isn't software a crew operates — it's a team of AI agents that does the precon and construction busywork and pulls a human in only for judgment, context, or a signature. Every conversation sells that day: the estimator shows up, reviews five things, approves three, provides context on two, and goes to the jobsite. The legacy stack is the faster horse. We're the car.
01 The Thesis — the future we sell Grounded
Before a rep says a word about features, they sell the category shift. This module is the worldview every other module hangs off.
◆ The fundamental inversion
Most "construction tech" is a tool humans operate — a faster way to do the same clicking. Perch inverts it: agents are the primary workers; humans are the specialized resource the agents call on when judgment, context, or physical action is needed. You don't manage Perch. Perch hands you a short, prioritized list and you approve, correct, or provide.
◆ The Company Brain
Every customer gets a living Company Brain that starts learning at onboarding (their trades, sectors, the way they work) and gets smarter with every job, doc, email, and approval. Agents ground every decision in that specific company's knowledge — not generic AI. This is the moat: the longer they use it, the more it knows, the harder it is to leave.
◆ The Henry Ford reframe (your #4)
"If I'd asked people what they wanted, they'd have said a faster horse." Procore, Bluebeam, BuildingConnected are faster horses — better versions of work the human still does. If we let the prospect compare us feature-for-feature against their stack, we've already lost the frame. We don't win the "better PM tool" argument; we make it irrelevant. The pitch is the destination, not the dashboard.
🔨 Build together
Category name. What do we call this so it's ownable and not "AI for construction"? Candidates: Agentic Preconstruction · The Construction Company Brain · Autonomous Precon. Pick / coin one.
The one-sentence pitch a superintendent repeats to his buddy at lunch. Needs your voice.
Three "from → to" shifts (e.g. "from chasing emails → to approving five things"). Let's draft together.
02 Personas — GCs & subs, by role Draft
Persona-first: before any play, name whose driver's seat we're in. Two segments (GC, sub), and within each the roles think very differently.
General Contractor (GC)
Owner / Principal — economic buyer at SMB; cares about margin, risk, winning more work without more headcount.
VP Preconstruction / Chief Estimator — #1 champion target; lives in takeoff, scope, leveling. Proud of precon as a strength.
Estimator — daily user; buried in plans/specs, addenda, sub bids. Time-starved.
Office manager — the human glue; chasing paper, payroll, compliance.
What they all distrust (your #5)
Blue-collar operators have been burned by software that promised the world, took months to set up, and made their day harder. They distrust "tech bros," jargon, and anything that smells like it was built by someone who's never been on a jobsite. Respect the trade. Speak plainly. Prove it on their real job.
🔨 Build together
Rank the personas for day-one targeting per segment. (Hypothesis: Sub Owner/Estimator and GC VP-Precon are the wedges.)
For each top persona: their one daily moment of pain we lead with, in their words.
Map champion vs economic buyer per segment (Built taught us the sub story; Zwick is teaching the GC story).
Anything from real Built / Zwick conversations that should overwrite my assumptions?
03 The status-quo stack Grounded
Your #7: understand what they run today — the layers, how they're stitched, what they cost — so we can show them the leap. We don't pitch feature-vs-feature; we map the seams that bleed money.
The Software Kill Chart — what it costs (from our pricing page)
$15k
Procore /yr
$6.0k
BuildingConnected
$5.4k
iSqFt / ConstructConnect
$4.2k
Bluebeam
$3.6k
PlanSwift
~$34k
Total /yr — that don't talk
The killer insight isn't price — it's that six tools that don't talk to each other force humans to be the integration layer. Perch is one brain, value-priced against the whole stack it replaces (see Module 6). The cost we attack is the human glue + labor, not the license fees.
🔨 Build together
Validate the named tools per segment — what do Built and Zwick actually run? (Real stacks > my generic list.)
Where are the worst seams — the handoffs where work falls through (estimate→PM, GC→sub, field→office)? Those are our entry wounds.
Should we keep a living "stack teardown" worksheet per prospect (feeds the research/ folder)?
04 The product — what Perch does Grounded
Your #2. The rep must know the surface cold — but always translate to outcome (Module 5), never recite features.
Precon engine
Upload plans/specs → AI reads them → scope of work generated
Built-in MEDDPICC sales pipeline — we run the playbook our customers run
Why it's structurally different (not just "better")
1) It does the work (agents, not buttons). 2) It learns the company (brain, not config). 3) It collapses cross-org communication (GC↔sub on one fabric, not email tag). No incumbent can bolt this on — it's a different architecture, not a feature.
🔨 Build together
Which capabilities are demo-ready today vs roadmap? (Critical: Zwick GC use case readiness is an open risk.)
What's the single "holy sh*t" moment in a live demo — the thing that makes a skeptic lean in?
Confirm the field-ops story for the whole-brain pitch — is it in-scope for launch messaging or "coming"?
05 Value translation — capability → outcome Draft
Your #3. Reps never sell a feature; they sell the outcome in the customer's numbers. This is the bridge from Module 4 to MEDDPICC Metrics.
What Perch does
What the customer gets
The number it moves
Agents read plans → scope
Estimators bid more, faster, without hiring
Pursuits/mo · win rate
Scope-gap analyst
Fewer margin-killing change orders missed at bid
Margin protected /job
Company Brain
The best estimator's judgment, scaled & retained
Throughput · key-person risk
Collapsed GC↔sub comms
Less email tag, faster RFIs/submittals
Hours/week · cycle time
One platform replaces stack
Kill ~$34k/yr of tools that don't talk
Software spend · IT glue
The built-in ROI model (from the product)
35%
Estimator productivity gain
60%
Software cost savings
+8%
Win-rate improvement
6%
Avg project margin (lever)
$85k
Avg estimator salary base
These are the defaults in the live ROI calculator — reps should pressure-test them against each prospect's real numbers and let the deal's own math sell.
🔨 Build together
Which one outcome leads per segment? (Subs: "win more of the right work"? GCs: "scale precon without hiring"?)
Are the ROI defaults credible/defensible — can we source them? Blue-collar buyers smell inflated claims instantly.
Do we want a shareable one-page ROI leave-behind generated per deal?
06 Pricing & packaging Working model · pre-launch
Our working model: six tiers, segmented by company size, value-priced against the stack + labor each tier replaces — not commodity per-seat pricing. Deliberately premium: we're selling the future, not a cheaper tool, and the price has to signal it. GCs sit above subs at every tier — they're the head of the beast (Module 11). The sticker is still a fraction of what it kills.
General Contractors — the head of the beast, priced accordingly
Premium is the message. A cheap price says "tool." This price says "this changes your business." Don't apologize for it — anchor it.
Anchor on what it replaces, never the sticker. Lead with their stacked software + labor cost, then Perch as a fraction of it.
Tiers unlock Brain + autonomy, not just seats. Core/Essentials = doc-level brain + manual approval · Pro/Growth = full Company Brain + config autonomy · Enterprise = full autonomy, SSO, multi-division, dedicated CSM, SLAs.
Size-segmented = self-qualifying. Employee count + revenue route the prospect to a tier — which feeds MEDDPICC Metrics + Economic Buyer instantly.
This is the SLG ceiling. Enterprise ($12.5K–$50K/mo, annual) is where sales-led + MEDDPICC discipline earns its keep; lower tiers ride PLG.
🔨 Build together / validate
Willingness-to-pay. These are premium, judgment-set numbers — pressure-test the ROI story against Built & Zwick before locking external pricing.
Reconcile the marketing site. Live pricing (Sub $299–$999, GC $499–$1,499) is an order of magnitude below this — a real code change when you're ready.
Entry-tier PLG friction. At $30K+/yr, is the motion truly self-serve, or "free trial → sales-assist converts"? Decide it on purpose (pairs with Module 9).
07 Messaging & language — the blue-collar voice Needs your input
Your #5, and the highest-leverage module for trust. This needs your voice heavily — I can scaffold the rules, you make it ring true.
Do
Talk like the jobsite — plain, direct, confident
Lead with their problem, in their words
Prove it on their real plan set
Respect that they're the expert; Perch is the apprentice that does grunt work
Show, don't explain the tech
Don't
Say "AI/ML/LLM/agentic architecture" to a superintendent
Lead with features or the dashboard
Over-promise or use startup hype words
Talk down to them or imply they're behind
Compare feature-vs-feature with Procore
Draft lines to react to — tap copy, then tell me what's wrong
It reads the plans so your estimator doesn't have to. You just check its work.
Bid more jobs without hiring more people.
It catches the scope you'd miss — before it eats your margin.
One system that actually talks to itself. No more six tools and a spreadsheet.
The trust-building moves
Trust isn't claimed, it's structured. Start conservative (agents propose, human approves), show citations ("it flagged this because of that note"), and widen autonomy as they gain confidence. The guardrail dial is the trust message: "You stay in control until you decide you don't need to be."
🔨 Build together
The tagline and the 30-second elevator pitch — in your voice.
Your banned-words list (what makes you cringe when a vendor says it).
A short story / proof anecdote (Built Exteriors?) we can tell instead of claims.
How hard do we lean on "American-made / for builders, by builders" type framing?
08 MEDDPICC applied to a Perch deal Grounded
The spine. Same living-scorecard discipline as the Zwick game plan — generalized into the reusable trainer. 🔴 unknown · 🟡 hypothesis · 🟢 confirmed by customer.
M — Metrics
What it isThe economic value of solving the pain, in their numbers — never "saves time."
Perch ammoPursuits/mo × precon hrs × loaded estimator cost; bids passed on for lack of capacity; $/job lost to scope misses; ~$34k/yr stacked software. Tie to the live ROI model (35% productivity, +8% win rate, 6% margin).
Discovery
How many bids/pursuits a month? How many estimators?
Win rate — and when you lose, is it price or scope?
What does a missed scope item cost you?
How many jobs do you pass on for lack of precon capacity?
E — Economic Buyer
What it isThe one who can release the money over objections.
Perch readSMB sub/GC → usually the Owner/Principal. Larger GC → exec (VP Precon sponsor + CFO gate). PLG twist: a user may start a trial, but the EB signs the expansion. Get to them before late stage, via the champion.
Discovery
Who signs off on software like this?
What dollar threshold needs the owner?
Bought tools like this before — how'd it go?
D — Decision Criteria
What it isThe formal + informal yardsticks. Shape them toward our edge (agentic scope from their real specs, the Brain) before anyone anchors on "takeoff clicks."
Discovery
What would you need to see to trust this on a live job?
What's in your estimating/VDC stack today?
Whose buy-in matters — estimators, field, IT?
D — Decision Process
What it isThe actual steps/people/timeline from interested → signed. PLG: trial → activation → expansion → procurement.
Discovery
Walk me through how you'd evaluate and roll this out.
Who else gets a say?
Pilot on one live pursuit first? Timeline?
P — Paper Process
What it isProcurement/legal/security mechanics after the yes.
Perch angleLight SaaS contract at SMB; but we handle owner plans/specs (sometimes NDA'd) → expect data-handling/security questions. SOC 2 Type II, encryption, RBAC are in the deck.
I — Identify Pain
What it isThe compelling, owned, quantified pain. Identify and implicate — make the cost of doing nothing vivid.
Perch nuanceIf they're proud of precon (like Zwick), frame as leverage, not remediation: scale the edge without hiring, protect margin, chase more, de-risk the single-estimator point of failure.
Discovery
Where does precon stop you chasing more work?
Tell me about a scope miss that bit you.
What happens to throughput if your best estimator leaves?
C — Champion
What it isInsider with power + influence + a personal win + access to the EB, who sells when you're not in the room.
Test, don't assumeA friendly "coach" gives info; a champion spends political capital. Test: "Can you get us 30 min with [owner]?"
C — Competition
What it isEvery alternative — named tools, status quo, "do nothing," and internal hire-vs-buy.
Perch readBiggest rival = inertia + "just hire another estimator." Named: Bluebeam/PlanSwift/STACK; maybe Procore/BuildingConnected. Never get dragged into feature-vs-feature (Module 1).
🔨 Build together
Do we want a blank reusable scorecard artifact (like Zwick's) reps clone per deal into deals/?
Per-segment qualification thresholds — what "green enough to advance" means for PLG vs enterprise.
Any house MEDDPICC conventions/scoring from Brock we should standardize on?
09 The sales motion — PLG → SLG Draft
Your call: PLG first, sales-led second. The product already encodes it — 21-day free trial (no card), self-serve tiers, "Talk to Sales" only at Enterprise.
Today's reality: pre-launch, GC product + pricing still being built → first deals (Zwick) run as design-partner / pilot, founder-led, to earn the proof PLG will later scale on.
The demo arc (cardinal rule: discovery, not demo)
Open with pain + buying map, not the product
Run their real bid package through Perch — tailored proof beats a canned tour
Land one "holy sh*t" moment, then shut up and let it breathe
Close on a next step that fills a MEDDPICC cell, not a feature recap
🔨 Build together
Define the activation metric (the "aha" event) and the PQL (when does PLG hand to sales?).
What triggers a sales-assist touch — usage threshold, team size, Enterprise interest?
Self-serve vs sales-led price/segment cutlines — where exactly does PLG stop?
10 Objections & battle cards Draft
Every objection below is really the same objection: "I'm measuring this against the world I already know." We're showing them something they've never seen — so the job isn't to win a feature fight, it's to expand the imagination. Move them from comparing tools to picturing their business in an agent-first world: the mechanical work runs itself, and the humans are there to observe, guide, and decide.
◆ The reframe move (use it on every objection)
1. Honor the frame. "Totally fair — based on every tool you've ever been sold, that's the right question."
2. Break the frame. "But those were all built for a world where people do the work and software just remembers it. We're built for the world where the work does itself."
3. Paint their business in it. "Picture your shop where the plans are read, the scope is built, the bids are leveled, the RFIs are drafted — overnight — and your people just review the calls that matter. What would you do with that?"
You're not selling a tool against their tools. You're selling the day their job gets easier — and letting them talk themselves into it.
"We already have Procore / Bluebeam / our stack."
The real objection"I already pay for software that does this."
Reframe to the future"Keep it — that's the filing cabinet for the work. The question I'd ask isn't 'is Perch a better cabinet.' It's: who's doing the work that fills it? Today that's your people, by hand. In an agent-first shop, the work shows up done and your stack just stores the result. We're not replacing the cabinet — we're replacing the late nights." Different layer entirely (Module 1).
"This is too different — I can't picture it."
The real objectionThe honest one — you're showing them something with no reference point. Don't over-explain. Demonstrate one slice.Reframe to the future"Nobody could picture email until they had it either. So let's not picture the whole thing — let's do one piece. Hand me your next real bid package. Tomorrow morning it's scoped, gap-checked, and on your desk for review. Start there. The future shows up one autonomous task at a time."
"I don't trust AI to run my jobs."
The real objection"I'll lose control / it'll make an expensive mistake."
Reframe to the future"You shouldn't hand it the keys on day one — and it doesn't ask you to. Treat it like a sharp new hire: it does the grunt work, shows you why it did what it did, and waits for your yes. You hand it more rope as it earns it. The real question isn't 'do you trust AI' — it's 'how much of the mechanical work do you want to stop doing yourself?'" Control is the trust mechanism (Module 7).
"My guys won't use it — they're not techy."
The real objection"Change is hard and my field people hate software."
Reframe to the future"Good — because in an agent-first shop nobody operates software. There's nothing to learn. The work arrives done and your foreman taps approve, the same way he reads a text. The guys who hate clunky tools are exactly the ones who'll love a world with less screen time, not more."
"We're too small / too busy to switch."
The real objection"I don't have bandwidth for a rip-and-replace."
Reframe to the future"You're too busy because you're still doing the mechanical work by hand — that's the thing we take off your plate. There's nothing to rip out; it learns your business in the trial and runs alongside what you've got. The busier you are, the more there is to hand off."
"Show me proof it works."
The real objection"This is new and unproven — why should I be the guinea pig?"
Reframe to the future"Fair — it is new. So we don't ask you to believe a deck. We run your real bid package through it and you judge the output yourself. Early partners shape the roadmap and lock in design-partner pricing — you're not a guinea pig, you're getting in before everyone else does." (Honest pre-launch posture.)
🔨 Build together
What objections did Built / Zwick actually raise? Real beats hypothetical.
The "just hire another estimator" objection — sharpen the agent-first counter (this is the true competitor: human headcount, not other software).
Which future-state picture lands hardest with blue-collar owners — pin the one or two we lead with (pairs with Module 7 voice).
The data-security card for handling owner plans/specs.
11 The org build → $100M Draft + math
Your #6 — now GC-first and enterprise-weighted. We go for the head of the beast: our PLG wedge is the plans, scope, and precon docs, and every bid a GC sends out comes back through our stack. Land one enterprise GC and you don't win one logo — you drag that GC's entire sub base onto Perch through the bid funnel. The dual-audience payoff: this is what you hand a future VP Sales to show the path.
Annual contract value (ACV) anchors — from pricing (Module 6)
~$42k
GC Core /yr
~$120k
GC Pro /yr
$300–600k
GC Enterprise /yr
~$84k
Sub Growth /yr
~$150k
Sub Enterprise /yr
Premium, enterprise-weighted, GC-first pricing means $100M needs only ~570 customers — and roughly half the revenue comes from ~120 enterprise GCs. Concentrated, not sprayed.
One illustrative path to $100M ARR — GC-first, enterprise-weighted
Segment
Paying orgs
ACV
ARR
GC Enterprise · the head of the beast
120
~$400k
$48M
GC Pro · mid-market
250
~$120k
$30M
Subs · pulled through GC bid funnels
200
~$110k
$22M
Total
~570 orgs
~$100M
Go for the head, not the tail. Nearly half the revenue here comes from ~120 enterprise GCs. We don't acquire subs one at a time — once a GC runs its bids through Perch, its subs have to engage through the stack, and they convert themselves. The GC's pipeline becomes our distribution. With ~700k+ US firms, ~570 customers is a rounding error on the TAM — the game is concentrated execution against the biggest GCs, not spray-and-pray. (Illustrative — swap in your real assumptions.)
The team that gets there (staged)
Land motion:GC-first, top-down — win the GC, inherit their subs through the bid funnel.
Now: founder-led + design partners (Brock leading deals; Jeremy learning the motion).
Enterprise field motion (the engine): AEs + Sales Engineers landing big GCs on MEDDPICC discipline — this is where the revenue concentrates.
PLG wedge: plans/scope/precon docs — low-friction entry that routes every bid back through the stack and seeds the sub base.
Pull-through + expansion: growth/lifecycle converts the subs the GCs deliver; RevOps + data from day one.
Hiring profile: construction-native credibility and modern sales rigor — MEDDPICC-trained, jobsite-fluent.
🔨 Build together
Your real targets: timeline, ACV assumptions, mix between subs/GCs/enterprise.
Hiring sequence & comp — when does the first non-founder rep land, and the comp model?
PLG efficiency assumptions (CAC, conversion, expansion) that make the math hold.
What does the MEDDPICC training program for new hires look like (since the whole team is raised on it)?
Perch Sales Playbook · v0.1 scaffold · built to be filled in together. Notes & progress save to this browser.