Four weeks of chasing. Or Monday morning.
A battery lifecycle and servicing business rebuilt its commercial-opportunity workflow on MorpheusOS. Moving a single opportunity from a signal somewhere in the market to something a salesperson could act on took two to six people, ten separate sources and two to four weeks. That journey now runs in minutes inside one guided workflow: fifty priority accounts ranked on published weights, a change board showing what moved, proposals drafted per project, scope and customer, and a territory summary in the team’s inbox before Monday starts. A named account manager reviews every proposal before it leaves.
The numbers, with their baselines.
Every figure below is an operating fact, read off the running workflow against the business’s own prior process. The account stays confidential; the work does not.
| Before | After | |
|---|---|---|
| People per opportunity | 2–6 | One reviewer |
| Sources consulted | 10 separate places | One guided workflow |
| Intake to actionable | 2–4 weeks | Minutes |
| Accounts ranked each cycle | As many as one person could read | 50, on published weights |
| Battery capacity covered | — | 11.2 GW |
| In near-term commercial windows | Found by hand, if at all | 18 |
| New or reactivated, week on week | Not tracked | +9% |
| Proposal preparation | Assembled manually, per opportunity | Drafted from the run — project, scope and customer specific |
| Where the knowledge lives | One strategist’s judgement | Published weights and logic |
| Cost of more activity | More coordination | The same run |
The expertise did not change. The number of people did not change. What changed is that the process no longer depends on someone remembering how to run it.
The expertise was never the problem. Finding the pieces was.
The business already knew which assets mattered, which service triggers were commercially real, and which operators were worth a call. What it did not have was one place where any of that met the market as it changed.
Ten sources, none of them talking
Email, spreadsheets, shared folders, internal records, PDFs, CRM and project-management systems, public project sources, OEM and EPC announcements, interconnection and permitting references, and internal chat.
The time was not the execution
It went on finding the right information, confirming it was current, checking for missing documents, asking someone for clarification, copying between systems, and deciding whether the opportunity was commercially relevant at all.
Then the output, then the handoff
Producing something usable, and making sure the next action was recorded or communicated to whoever owned it — because if it was not written down, it depended on someone remembering.
One opportunity: two to six people, two to four weeks, and four costs that never appeared on an invoice — cost, delay, inconsistency and handoff risk.
Why more people would not have fixed it.
The process depended on people knowing where to look, what to check, who to ask, and how to turn scattered information into a decision. That is a knowledge problem wearing a capacity problem’s clothes.
More activity meant more coordination
Every additional opportunity added handoffs rather than absorbing into a system. Growth in the pipeline arrived as growth in the meeting load.
The same case ran differently twice
Two people working the same opportunity checked different sources in a different order and weighted them differently. Neither was wrong. Neither was repeatable.
The method left when the person did
The judgement that made a ranking good — which trigger actually predicts a service need — sat with the people who had learned it, and nowhere else.
An opportunity identified in week four is a different commercial proposition to the same opportunity identified in week one. In a market where interconnection queues, ownership changes and warranty windows all move, the delay is the loss.
What a good analyst actually does, written down.
The build did not start with software. It started with the question of what the best version of this process looks like when the person who is best at it is not in the room.
Required fields
What must be present before an opportunity is assessed at all — so nothing is scored on a guess.
Source hierarchy
Which source wins when two disagree, and how recency is weighed against authority.
Service-trigger logic
Which asset events, ownership changes and operating conditions actually indicate a service need — the judgement that used to live in one head.
Documentation standard
What evidence has to sit behind a ranking for a salesperson to be able to defend it in the room.
Output structure
The shape of the radar, the change board and the proposal — so every cycle is comparable with the last.
Escalation rules
What stops the run and goes to a person, rather than being decided by a score.
Those six things are the asset. The workflow is what executes them.
One cycle, end to end.
The run executes unattended and stops before anything reaches a customer.
01 · Sweep
The approved source set is read — project data, permits, ownership changes, OEM and EPC announcements, safety incidents, interconnection references.
02 · Resolve
Signals are mapped to real assets, projects and accounts, so the same operator under three different names is one account.
03 · Score
Published weights applied against the service-trigger logic. What could not be reached is recorded as missing, not skipped.
04 · Compare
This cycle against the last: what is new, improving, declining, or now needs action.
05 · Prepare
Account plays and next actions per territory, and a proposal drafted per opportunity — specific to that project, that scope, that customer.
06 · Park
The summary lands before Monday starts. Nothing leaves the business until a named account manager releases it.
A priority radar of fifty accounts on published weights; a change board of new, improving, declining and needs-action; every project above 10 MW mapped; a sourced safety and recall view; a territory summary at 06:00 Monday; and a one-page brief per opportunity.
The strategy improves because the corrections are kept.
The difference between a report and an operating layer is what happens to a disagreement. Here, a strategist overruling a score is not a workaround — it is the input.
Which recommendations are taken
Accepted and rejected rankings both carry information about the weights.
Which accounts convert
Outreach outcomes logged against the trigger that surfaced them.
Which sources hold up
Reliability tracked per source, so authority is earned rather than assumed.
Which signals mislead
False positives are recorded, and the trigger that produced them loses weight.
Where an expert overrides
The correction and its reasoning are retained, so the next cycle starts from the corrected position.
Which proposals move
What progressed to a real conversation, and what stalled and where.
Institutional knowledge stops being something the business hopes to retain and becomes something the workflow accumulates.
Not people. Four things with a right answer.
Every one of these was work a capable person did well and would rather not have been doing. None of them is where the commercial judgement lived.
Finding
Locating the signal across ten sources that do not know about each other, and knowing which ones to check for this kind of opportunity.
Confirming
Establishing whether what was found is still true — the ownership that changed, the permit that lapsed, the project that moved.
Carrying
Moving information between systems so the next person could use it, and chasing the person who already knew the answer.
Shaping
Turning the result into the same structure every time, so a ranking this week is comparable with a ranking last week.
Deciding whether an opportunity is worth pursuing is judgement, and it stayed. Assembling the evidence needed to make that decision is retrieval, and it did not.
The run prepares. A person still decides.
The workflow sits at Prepare on the five-stage autonomy ladder: it assembles, ranks and drafts, and it stops before anything reaches a customer.
Commercial decisions
Whether to pursue an account, at what price, on what terms. The radar ranks; it does not commit.
Technical commitments
What the business will undertake on an asset, and to what standard. A score is not a scope.
Anything customer-facing
A named account manager releases every proposal. There is no path through the run that puts a document in front of a customer on its own.
The scoring weights and the reasoning behind them; the source hierarchy that decides which record wins when two disagree; the service-trigger logic that says which asset events actually indicate a commercial need; the documentation standard a ranking has to meet to be defensible in the room; and every correction an expert has made since. That set used to live in one strategist’s judgement. It is now the logic the next cycle runs on — inspectable, arguable, and owned by the business rather than by whoever happens to hold it.
What people ask about this build.
Answered against the run record, not the intention.
How long did this workflow take before MorpheusOS?
Two to four weeks from intake to completion, involving two to six people and ten separate sources — email, spreadsheets, shared folders, internal records, PDFs, CRM and project-management systems, public project sources, OEM and EPC announcements, interconnection and permitting references, and internal chat. The run now completes in minutes inside one guided workflow.
Does the workflow send proposals to customers?
No. It drafts them. Every proposal is project-, scope- and customer-specific and is reviewed by a named account manager before it leaves the business. On the five-stage autonomy ladder this workflow sits at Prepare: it prepares the artefact and stops before any external action.
What does the scoring actually rank on?
Published weights. The ranking is not a black box — the weights, the source hierarchy and the service-trigger logic are all inspectable. That is what lets a strategist disagree with a score and have the disagreement change the next cycle rather than get lost in a thread.
What happens when a source is stale or missing?
It is recorded as missing rather than quietly absent. Confirming whether data was current was one of the hidden costs in the manual process; every ranked account now carries its source and its recency.
Does this reduce the size of the sales team?
No. It changes which work they do. The coordination load — finding, confirming, copying between systems, chasing the person who knew — is carried by the workflow. Commercial decisions, technical commitments and anything customer-facing stay with people. More activity is now absorbed by the same run instead of by more coordination.
How does the workflow improve after it is live?
It learns from which recommendations the team accepts or rejects, which accounts convert, which sources prove reliable, which signals produce false positives, which service triggers lead to real opportunities, and where an internal expert overrides an assumption. Those corrections become the logic the next cycle runs on.