Reality ◯ · Diagnose before prescribing
Solar365 already knows how to design and deliver solar. The uncertain part is not expertise. It is how much expert attention each enquiry consumes before a decision-ready proposal can leave the business.
The honest map
Public proof and named tools exist. Lead qualification, proposal framing, site heuristics, follow-up, and delivery-readiness judgment remain human-held. That is a diagnosis to test, not a verdict to automate around.
The readings
Discovery does not begin with an AI recommendation. It reads the current system so Mike and Austin can decide whether the proposal loop deserves the first intervention.
Each problem names its current leak, its opportunity cost, and the signal that would prove or disprove it. None is a verdict — each is a hypothesis the discovery sprint tests.
Current leak: Every qualified commercial or school lead requires expert triage before a decision brief is possible. Mike and Austin are the bottleneck.
Opportunity cost: High-value leads wait while expert attention is rationed. Response speed — the key competitive factor in agent-mediated procurement — is structurally capped.
Proof signal: Tribal audit: 6 of 7 areas are HIGH tribal risk. Discovery sprint: time 5 proposal paths to reveal where expert hours are consumed.
Current leak: The public form captures contact, location, build type, and property type. It does not capture bill size, decision date, budget, urgency, or roof constraints.
Opportunity cost: Every qualified lead requires a manual discovery call before the proposal process can begin — multiplying expert time at the earliest stage.
Proof signal: Data audit §LOW: bill size, urgency, budget, and decision date all confirmed missing from intake.
Current leak: 3,000 installs, 16% average return, and 6-year payback are homepage claims. The method, segment sample, and performance data are not structured or machine-readable.
Opportunity cost: When AI buying agents compare suppliers, Solar365's strongest asset — its install record — is invisible to the agent. Reputation becomes a handicap instead of a moat.
Proof signal: Northstar § Protect Trust: the two-year position requires machine-verifiable performance proof. The data trail starts in Stage 1.
Current leak: Proposal outcomes, follow-up decay, and lost reasons are not captured in any system. Each proposal cycle starts from scratch.
Opportunity cost: Every proposal cycle loses learning that should compound — ICP targeting, proposal language, pricing signals, and delivery-risk flags.
Proof signal: Cockpit instrument: win/loss outcome field is required from cycle 1. Without it, Stage 2 ICP targeting has no data to build on.
Current leak: No system checks subcontractor readiness before a commercial promise is made. Faster proposals can create install promises the business cannot keep.
Opportunity cost: The trust built across 3,000 installs can be damaged faster than it was built if sales speed outpaces delivery capacity.
Proof signal: Kill signal in transformation-plan.md: 'Delivery stress — faster sales creates unowned subcontractor or install handoff risk.'
Every claim below is tagged by confidence. HIGH is publicly verifiable or confirmed in the operating brief. MEDIUM is inferred and must be measured. LOW is confirmed missing — the discovery sprint's shopping list.
Exists — high confidence
Likely but unread — medium confidence
Confirmed missing — low confidence
Tribal knowledge is expertise that lives only in expert heads. High tribal risk with low documentation means the business cannot delegate, systematise, or prove that judgment — and AI cannot read a signal that was never written down.
The diagnosis is not a deficit story. These assets are why the constraint is worth fixing — they are what a better proposal loop would leverage.