Walkthrough — Cache Valley Electric · white-glove
A large sub run end-to-end through the human-guided funnel. Same engine as self-serve — the difference is the last mile: the agents do all the prep off-stage, and a person carries the value. The point of this page: show how much work the agents do, and how armed the AE/SDR is the moment they reach out.
Perch · CRO cockpit · Jun 25 2026 · mock-ups schematic · siblings: Squires (self-serve) · journey-map
Cache Valley Electric (CVE)Large sub · enterprise
- Type
- Electrical subcontractor · Logan, UT · ~2,300 emp · $700M–$1.5B (est)
- Does
- Design-build electrical, mission-critical / data centers, network integration, low-voltage
- Buying map
- Champion: VP Preconstruction/Estimating · EB: COO/President · sophisticated in-house estimating + VDC
Two actors, one engine. 🤖 The agents work the whole account off-stage — dossier, a pre-run takeoff on CVE's real pursuit, drafted outreach, a pre-filled MEDDPICC. 👤 The AE/SDR is the only human CVE meets — and thanks to the agents they show up knowing the company cold. White-glove is just self-serve where a person carries the value instead of the agent closing in-flow.
🤖 agent off-stage prep
👤 human AE/SDR on-stage
surface real app surface
built
to build
customize
🤖 PHASE A · Agents — back office, before any human contact
Signal → swarm fires, deep🤖 agentto build
⚙ working account: cve.com
⌖ Account dossier
→ 2,300 emp · design-build
⌖ Buying committee
→ champion · EB · blockers
⌖ Active pursuits
→ hyperscale DC pkg
⌖ Current stack
→ Accubid + in-house VDC
⌖ Displacement angle
→ read-the-set wedge
⌖ Sarah · pre-takeoff
→ queued on real pkg
Same swarm as the self-serve flow, but it runs deeper for an enterprise account — not just "who are they," but the buying committee, their live pursuits, what they use now, and the angle to displace it.
Customization here The enterprise swarm is greenfield. Off the critical path (no one's waiting), so it can go as deep as the deal warrants. Output isn't a workspace — it's an AE battle pack.
The battle pack — the AE's brief🤖 agentto build
perch.internal/accounts/cve/brief
Who
CVE — UT's oldest/largest electrical; design-build; mission-critical heavy.
Buying committee
Champion: VP Precon · EB: COO · Blocker: IT/Security, in-house VDC lead.
Live pursuit
Mountain West hyperscale data center — electrical pkg bidding now.
Pain / wedge
Switchgear 40+ wk lead times; spec-buried gear; estimator hours on page-flipping.
Objection to expect
"We have a strong team + Accubid." → wedge is read-the-set, not replace.
The agents hand the AE a one-screen brief: who CVE is, the people who decide, their live job, the pain, and the objection they'll hit — before the AE has spent a minute of research.
Customization here This is the whole "how much work the agents do" story. Everything a rep used to spend days assembling, assembled automatically per account.
Pre-run the value — Sarah on the real pursuitJobs · Sarahbuilt
perch.internal/cve/dc-package · pre-run
Pre-takeoff · MW data-center electrical · 812 pages read
Switchgear / distributionflagged
Feeders + branch (devices)14,200 ea
Lighting fixtures3,180 ea
⚠ Spec section 26 calls for an arc-flash-rated gear option not shown on E-sheets — 40-wk lead, missed on base bid.
Before the AE ever calls, Sarah has already run CVE's actual live package and found something real — a gear item buried in the spec with a 40-week lead. The demo won't be hypothetical; it's their own job.
Customization here Sarah exists. What's new: running her pre-emptively on a prospect's real pursuit to manufacture a concrete, undeniable hook.
Draft the outreach + pre-fill MEDDPICC🤖 agentcustomize
perch.internal/cve/meddpicc
MMetrics: bid capacity, gear-risk avoided
EEcon buyer: COO
DDecision criteria: fits VDC/Accubid
IIdentify pain: gear lead-time misses
CChampion: VP Precon
CCompetition: Accubid + status quo
The agents draft the personalized outbound (next step) and pre-fill the MEDDPICC scorecard from the research — green = known, amber = the AE confirms live. The rep starts the deal already half-qualified.
Customization here Reuses the MEDDPICC scorecard we already built for the Zwick deal — now seeded by research instead of filled by hand.
👤 PHASE B · Human — the AE/SDR, armed
SDR reaches out — and it lands👤 humancustomize
SDR → VP Precon, CVE
From Taylor · Perch · To VP Preconstruction, CVE
Your Mountain West DC bid — a switchgear risk
Hi — saw Cache Valley's chasing the MW hyperscale electrical package. On jobs that size, switchgear's running 40+ weeks and the spec usually buries a gear item the E-sheets miss. We built Perch to read the full set and surface those before bid day. I already ran your package — it flagged 3 worth a look. Worth 15 min?
CVE replies
…how do you know what we're bidding? Send a couple times.
The SDR sends a message that could only be written about this company — their real pursuit, their real pain, a real finding. That's why it gets a reply instead of the trash. The agents wrote the first draft; the SDR adds the human touch.
Customization here The outbound is generated from the battle pack. The SDR's job shifts from "research + write" to "review + send + book." 100× more personalized than a sequence.
AE call — insight, not discovery👤 humancustomize
discovery call · CVE
AE (Perch)
"I'll skip discovery — I know Cache Valley: ~2,300 people, design-build, big on mission-critical, mature team on Accubid. So here's the only thing worth your 15 minutes…" [shares the 3 flags on their real package]
VP Precon
"We've got a strong estimating team and we run Accubid. What's this replace?"
AE
"Nothing. Your team's why you win — Perch just reads the 800-page set first so they spend hours on judgment, not page-flipping. It caught that 40-wk gear item in 90 seconds. That's the wedge."
The AE walks in already knowing CVE cold, so the first call is insight, not discovery. And because the battle pack flagged the objection, the AE has the rebuttal ready: Perch arms the estimators, doesn't replace them.
Customization here Talk track + objection handling pre-loaded per account. The AE confirms the amber MEDDPICC cells (EB, champion) live on this call.
The demo — pre-loaded with their dataJobs · Analyticsto build
app.perch.pro/demo/cve
AE shares screen — already CVE's world:
MW data-center pkg · scoped3 gear flagsdevice counts
Estimating-at-scale view
6 estimatorsstandardized bidsCompany Brain · senior knowledge
The AE doesn't demo a generic sandbox — they open Perch already populated with CVE's real pursuit (the step-3 pre-takeoff) plus the enterprise frame: how a 6-estimator team standardizes, and how Company Brain keeps senior knowledge.
Customization here The demo env is seeded from the pre-run, not built live. Same surfaces a self-serve user sees — just driven by a human on a screen-share.
▼ PHASE C · Converge — same product surfaces, seated by the FDE
Seat the team + migrateTeam · Importcustomize
app.perch.pro/team · CVE
FDE seats the estimating dept
VP Precon (champion)Admin
6 estimatorsEstimator
VDC leadMember
Import historical jobs
Instead of one self-serve user, an FDE seats the whole estimating dept, assigns roles, and Perch imports CVE's historical jobs so the team lands in a populated, familiar system.
Customization here Role-aware seating at enterprise scale + historical-job migration. Same Team surface as self-serve — operated by a human, for many seats.
Close — AE + MEDDPICCPlan · Contractbuilt
perch.internal/cve/meddpicc · closing
MMetrics agreed: bids/mo + gear-risk avoided
EEcon buyer confirmed: COO, in the room
PPaper process: annual enterprise, security review
CChampion confirmed: VP Precon
Enterprise annual · signed
A pre-qualified, multi-threaded close: the AE works the (now-green) MEDDPICC to an annual enterprise contract. Not a card-on-file — a signed deal, with Perch migrating their data.
Customization here The close is human + MEDDPICC vs self-serve's in-flow card. Billing exists; enterprise contracting + security review is the added path.
Self-serve vs white-glove — same engine, different last mile
| | ⚡ Self-serve (Squires) | 🤝 White-glove (CVE) |
|---|
| Research swarm | Arms the agent | Arms the human — deeper, full dossier |
| First human touch | None — agent only | SDR, fully briefed by the battle pack |
| The "wow" | Agent runs it live, in-flow | AE shows it — pre-run on their real pursuit |
| Qualification | Implicit (they self-select) | MEDDPICC, pre-filled by research |
| Close | Trial → card, one session | AE + annual enterprise contract |
| Speed | Minutes | Days — but first touch is insight, not discovery |
| Product surfaces | Identical — Jobs · Chat · Sarah · Company Brain · Team · Plan. White-glove just seats more of them, via a human. |
The matrix is forming: cold/warm (how much we know) × self-serve/white-glove (who carries it) = the four routes the decision tree has to handle. Next: turn the three walkthroughs into that tree.