Customer zero · Measured operating results · Drafts reviewed before release

A reviewed research draft from one governed run.

Q: Does this workflow send email to clients?

No. The workflow creates a Gmail draft and notifies the team. Sending is a manual human action. There is no path in the workflow that releases a client email automatically.

HtAG Analytics uses a decade of property-research methodology to prepare briefs from client requests. Required inputs and evidence checks guide each run, and a researcher reviews every draft before release. The case records how the service’s price, delivery volume, revenue and capacity changed from their earlier baselines.

$750 → $350: price per research brief
3–5 → 50: briefs delivered per week
50 per day: measured capacity, with review on release

Chapter 01 / 06

Measured results

The commercial result is shown with the prior price, throughput, revenue and capacity baselines.

Workflow context and documented result

HtAG Analytics rebuilt its property-research service on MorpheusOS. A decade of methodology — roughly 150 indicators, the filter cascade, the argued recommendation — became governed operating memory, and the run now executes from a client’s email to a reviewed draft. The price of a brief was cut from $750 to $350, delivery moved from three-to-five briefs a week to fifty, and annual research revenue moved from roughly $150,000 to a run rate approaching $1,000,000 — with measured capacity for fifty briefs a day, no added headcount, and a person on every release.

The numbers, with their baselines.

Every figure below is an operating fact — measured on the running instance against the service’s own baseline. This is the deployment we publish at full depth, because the business running it is ours.

Briefs delivered

Before: 3–5 a week
After: 50 a week

Price per brief

Before: $750
After: $350

Time per brief

Before: Several days
After: Minutes, unattended

Annual research revenue

Before: ~$150,000
After: ~$1,000,000 run rate

Measured capacity

Before: Capped by hours
After: 50 briefs a day

Headcount added

Before: —
After: None

Where the team’s hours go

Before: Producing reports
After: Coaching clients

The binding constraint

Before: Research capacity
After: Customer acquisition

End of chapter 1 · Measured results

Chapter 02 / 06

The research

The valuable deliverable is a reasoned, reviewable recommendation rather than an isolated ranking.

The deliverable is the reviewed argument.

A client submits a brief — a budget, a horizon, a risk tolerance, an objective. HtAG returns the market that best fits it, and the evidence for why. That sentence hides several days of work.

~150 indicators per market

Supply scarcity over four time horizons, socio-economic banding, affordability as years-to-own, cycle position, growth deviation, rental depth, risk composites, transaction confidence.

Meaning lives in the cross-reference

Not one of those numbers means anything alone. A market at twenty-five days on market that has been rising from fifteen tells the opposite story to one at twenty-five falling from fifty.

Then the document

Why this market, for this brief, over every alternative that also qualified. Which indicator carried the decision, at what weight, and why. What the second-best market lacked. What could go wrong.

A research team working at full stretch produced three to five of these a week.

Make the methodology reusable.

It was a knowledge problem — the exact gap an operating-memory layer exists to close.

The method lived in one head

Built over more than a decade, most of the research was performed by the founder — the method lived in the person who had run it thousands of times, and around him in coaching recordings, checklists and corrections. Hiring did not add capacity for months, because a new analyst cannot be told the method; they have to absorb the judgement inside it.

The quieter cost

The people who understood the method best were the people best placed to explain it to a client — and they were the people spending their weeks assembling documents. And a business whose core process depends on its founder carries a structural discount on its own value: key-person risk is the first thing a buyer prices in.

Demand exposed the delivery constraint

Demand arrived faster than research could be produced, so marketing spend converted into a backlog rather than revenue. At $750 a brief, five a week put the absolute ceiling at $195,000 a year; the service actually ran at roughly $150,000 — and the cap was structural.

End of chapter 2 · The research

Chapter 03 / 06

The method

The archive exposes contradictions and turns the research method into governed operating memory.

Morpheus read the archive — and found where the method contradicted itself.

HtAG gave Morpheus the whole archive: methodology documents, research rules, coaching transcripts, and sixty-seven client engagements’ worth of completed work — including the corrections, rebuttals and version histories that show how conclusions were argued down and rebuilt.

THE RECONCILIATION

Across the archive, the same metric carried different thresholds in different documents. One reference file treated a socio-economic band as monotonic — higher is always better. The coaching material held the actual research finding: the relationship inverts above a certain price point and holds only for certain horizons. Both were true statements about the method. Only one was the method.

WHY THIS IS THE PRIMITIVE

A decade of institutional judgement is not a document. It is a set of rules, exceptions to those rules, and a sense of when the exception applies. Accepting and pruning beliefs — keeping what survives review, retiring what does not — is what turned it into a system where every decision is explicit and every run makes the same one.

Suburb analysis

Interprets every indicator in context — what it means, what it means against the other indicators, and what it means for this particular brief. The cross-referencing that used to be the senior analyst’s job.

Suburb ranking

The filter cascade, then the weighted ranking, then the brief-specific overlays. Hard gates first, in hierarchy order; differentiation across the full qualifying set. The top score is a candidate, never auto-adopted.

Strategy report

Produces the argument — reasoning, comparison set, weighting disclosure, comparables, negotiation position, risks and caveats — in house style, white-labelled to the agency being served.

Together they execute the research deterministically. Same brief, same data, same answer, every time — held to the standards the team spent a decade developing.

End of chapter 3 · The method

Chapter 04 / 06

The workflow

A paid request is checked, researched, criticised, drafted and handed to a researcher for release.

Follow the work from request to reviewed draft.

Codifying the method was the first step. Running it was the point. The spine below is the path a healthy run takes; the branches are the two places Morpheus refuses to continue, and the point where the work is handed to a person.

Trigger

A paid request arrives

A research request begins as a form submission and arrives as an email. That email is the trigger — the work starts when the demand does.

Agent

Read the request and parse the Brief Lock

Morpheus reads the submission through the Gmail connector and extracts the structured brief — client, budget band, objective, yield floor, risk profile, horizon, property type, geography, deliverables, existing portfolio, brand context.

Source threadStructured briefMissing-field list

Gate

Is the brief complete?

The brief is checked against what the methodology actually requires. The test is binary — there is no partial-credit path.

If a field is missing

The run stops here, and the team is told exactly what is missing. Research never starts on a guess.

Agent

Gather the data, run the method

Morpheus calls the HtAG Developer Portal APIs for the proprietary market data — roughly 150 indicators per market — and reaches out through web search for the external context: the agency’s brand, the client’s business, whatever the report has to be right about beyond the numbers. Then it filters, ranks, overlays and interprets.

Artefact packageWorkspace pathsEvidence trail

Agent

Critic check

A separate reviewing step, not the one that produced the work. It tests the package against the brief that authorised it and returns a pass or fail with the issues attached.

Gate

Did the critic pass?

The second binary gate. Its purpose is to keep an unverified or misaligned package from ever becoming something a person might send in a hurry.

If the critic fails

The package is held before a draft is created, and the team is told why. There is nothing sitting in an outbox to be sent by accident.

Connector

Create the draft — for review only

The connector writes a formatted Gmail draft with its artefacts — the branded document, the visual summary, the research-process disclosure, the suburb shortlist. Draft-only is a design decision, not a limitation.

Notify

Tell the team it is ready

Two notifications go out in parallel: one into the workspace channel, one by email. Both say the same thing — the package is prepared and waiting on a person.

Human

A researcher reads it and decides

The reviewer opens the draft, checks it against the artefacts, and sends it or does not. This is the release gate, and it is the reason the rest of the workflow is safe to run at volume.

Gmail

Receives and interprets the research request, and delivers the reviewed package.

HtAG Developer Portal APIs

The proprietary market data. HtAG’s data stays where it is — Morpheus operates above it.

Web search

The external and client-specific context the analysis needs.

Three connections into the business — and nothing about HtAG’s own systems changed to accommodate them.

End of chapter 4 · The workflow

Chapter 05 / 06

Checks and records

Evidence checks, required brief fields and the record make the run inspectable before and after drafting.

Inspect the operating record.

The run these numbers come from — the step record, the timeline and the flow with its gates, straight off the instance. Click any frame to read it at full size.

Run 93752982 · Steps

8 of 8 steps, on the record. The Brief Lock parsed from a real payment email.

One inspected run: 8 of 8 steps, 512 seconds — and this particular run took the critic-fail branch and stopped, exactly as designed.

Six things checked before anything is drafted.

The critic is the difference between a system that produces work and a system that is willing to reject its own work. Each check can hold the run on its own.

Brief alignment

Does the output answer the brief that was paid for, or a nearby question the research drifted into?

Evidence

Does every figure trace to a real query, and are the gaps named rather than smoothed over?

Branding

Does the artefact match the agency brand context captured in the brief?

Mandatory sections

Is anything the deliverable is required to contain missing from the package?

Wrong-client leakage

Has any material from another engagement reached this one? This check exists because the risk is real, not theoretical.

File outputs

Do the artefacts actually exist at the paths the draft is about to reference?

Ten required fields establish the brief.

This is the structure that makes the first gate possible. A brief is either complete against this list or it is not, and the workflow behaves differently in each case.

Client

Who the work is for.

Recipient

Who actually receives it — not always the same person.

Geography or market

The area the research covers.

Goal

What the client is trying to decide.

Budget

The envelope the advice must sit inside.

Dwelling type

The asset class in scope.

Timeframe

The horizon the decision runs over.

Constraints

What is ruled out before work begins.

Requested deliverables

What the client was told they would get.

Brand context

Whose name the artefact carries.

End of chapter 5 · Checks and records

Chapter 06 / 06

Capacity and questions

The workflow carries repeatable work while the human result and approval boundary stay visible.

Four repeatable parts of the run.

Each one is the kind of work where consistency beats inspiration — where a tired analyst at six o’clock on a Thursday is measurably worse than a rested one on Tuesday morning. That is precisely the work a deterministic system should own.

The administration

Reading the request, extracting the brief, checking it was complete, chasing down what was missing.

The data gathering

Pulling roughly 150 indicators for every market on the list, every time.

The analysis

Filtering, ranking, cross-referencing, overlaying — the codified reasoning the team had already proven over a decade.

The production

Assembling that reasoning into a document, in house style, to the same standard, on every single brief.

None of it was the work the team was hired for.

The numbers that went up are the human ones.

Research only becomes a decision when someone understands it. That explanation is human work — it cannot be codified, and nobody at HtAG wants it to be.

$350per brief, down from $750

50briefs a week, from 3–5

~$1Mannual run rate, from ~$150k

50/daymeasured capacity — one inspected run: 8 of 8 steps, 512 seconds

More walkthroughs

A person on camera for every piece of research, taking the client through the reasoning, the alternatives considered and rejected, and what the recommendation means for their position. Production used to compete with this for the same hours — and production always won. That competition is over.

More coaching, more conversations

The team now runs more coaching sessions and holds more client conversations than at any point in the firm’s history — not despite the volume of research going out the door, but because of it.

A person on every release

Every package is drafted, checked, held for review and released by a researcher. Autonomy is earned — and it was never earned on the last step, because the last step is where a client’s trust is on the line.

The price went down. The revenue went up six times.

When the marginal cost of producing a brief collapsed, HtAG cut the price from $750 to $350 — and won the volume that was always waiting behind it. Fifty briefs a week at less than half the old price is a run rate approaching $1,000,000, from roughly $150,000. Cheaper for every client, and substantially larger as a business. This case shows how a documented research method changed the economics of this service.

The constraint moved — because capacity did

Measured capacity now stands at fifty briefs a day — at today’s price, more than $6M of annual delivery headroom. Demand no longer throttles at the point of delivery, so marketing spend converts into revenue instead of backlog. The constraint is customer acquisition — a commercial priority the team can address with the delivery capacity in place.

The multiple moved with it

Research that once depended on the founder now runs as governed operating memory — inspectable, transferable, independent of any one person. That converts key-person risk into an asset of the business, which is the difference between selling hours and selling a system. Every vertical has a version of this: the multiple follows the method out of the founder’s head.

The five autonomy stages, in full The other build: the learning loop All eight domains this runs beside

What people ask about this build.

Answered against the run record, not the intention.

Are the figures in this case study real?

Yes, with their baselines. Before: three to five briefs a week at $750 each — roughly $150,000 a year against a structural ceiling of $195,000. After: fifty briefs a week at $350 each — a run rate approaching $1,000,000 — with measured capacity for fifty a day, which at today’s price is more than $6M of annual capacity. This is customer zero — HtAG Analytics, run in the open on the product.

Did the workflow replace the research team?

No. It replaced four categories of work — administration, data gathering, codified analysis and document production — each of which has a right answer. The human numbers went up: the team now records more client walkthroughs, runs more coaching sessions and holds more client conversations than at any point in the firm’s history, and a researcher still releases every package.

Does this workflow send email to clients?

No. It creates a Gmail draft and notifies the team. Sending is a manual human action. There is no path through the workflow that releases a client email automatically.

What happens if the brief is incomplete?

The run stops before any research is produced. Morpheus posts a workspace notification and emails the team naming the missing fields, so research never starts on a guess.

What did this do to the value of the business itself?

Before the build, most of the research was performed by the founder — and a business whose core process depends on a key person carries a structural discount on its sale multiple. That process now runs as governed operating memory: documented, inspectable and independent of any individual. Scalable revenue on a transferable system is what buyers pay multiples for — the same reason operators in any vertical embed technology into the business rather than around it.

Is this workflow autonomous?

It prepares unattended and releases nothing. The request email triggers the run, and the analysis, ranking and document production execute end to end — but it stops at two hard gates and again at release. On the five-stage autonomy ladder it sits at Prepare: it drafts the artefact and stops before any external action.

Map a workflow like this

End of chapter 6 · Capacity and questions

Apply it to your business

Start with one piece of work.

Bring the process, the evidence and the person who owns the decision.

Map a workflow ↗