Marlow & Finch - App & Productivity Intelligence Pack (Sample)

ConfidentialEvidence and architecture packUrgency: MEDIUM

What this is. The internal evidence spine behind the report, three journey maps, Workbench and roadmap. Charge-out, internal cost, conditional revenue and TBC are kept as separate economic types. Marlow & Finch Lawyers is a fictional firm. The method, scoring and structure are exactly what we produce for a real engagement.

Warning

Proposed architecture is not current production state. Model routing, data location, subprocessors, retention, logging, rollback and matter permissions require live verification before client material is used.

Example

Recoverable effort is $991 of the $1,586 modelled cost-to-serve per advice. $991 / $1,586 = 62.48%, displayed as 62.5%.

1. Executive snapshot

System of knowledge: Microsoft 365, SharePoint and Teams hold the precedents, matter plans, checklists and house document set. The precedent library is maintained by hand.
System of work: your practice management system holds matter state, key dates, time recording and the trust ledger. Several steps are updated by hand and the file-opening pack is assembled manually.
Document path: the document management setup, the PDF toolset, the e-signature platform and the searches and settlement platform carry a matter from draft to signed and settled. Scanned search results limit what any system can read.
AI estate: Microsoft Copilot is part-enabled without file connectors; a senior solicitor and the practice manager use ChatGPT and Claude for ad hoc drafting. There is no governed firm-wide path.
Proposed governed path: Australian-hosted AI, matter-scoped separation, no cross-matter context, human gates and a versioned eval record. This describes proposed architecture that awaits live verification.

2. Stack and decision surface

System	Current role	Decision	Integration / control note	Confidence
Microsoft 365 / SharePoint / Teams	Documents, collaboration and firm knowledge	Keep as approved knowledge boundary	Define matter-scoped stores, labels, permissions, retention and audit access before AI use.	High
Document management setup	Matter files, precedents and the house document set	Extend, then tidy selectively	Index approved precedents first. Do not expose client material without matter scoping.	High
Your practice management system	Matter state, key dates, time recording and the trust ledger	Keep, integrate narrowly	Read matter state and write approved file-opening and handover records. No agent is given trust ledger access.	High
Word / Outlook / Excel	Advices, client letters and engagement letters	Keep output formats	Generate from validated structured inputs; preserve version control and solicitor approval.	High - artefact reviewed
PDF toolset	Contracts, searches and executed documents	Test extraction quality first	Text quality on scanned searches and older titles sets the ceiling on what a draft can rely on.	Medium
E-signature platform	Costs agreements, authorities and client sign-off	Keep, human-triggered	A named person sends for signature and confirms execution. No agent sends or countersigns.	High
Searches and settlement platform	Title, rates, council results and settlement lodgement	Keep, read-only for AI	Results may be collated and summarised. Ordering, lodgement and settlement stay with a person.	High
Copilot	Part-enabled, no file connectors	Assess tenant controls before extending	Data paths, model routing and permissions require live tenant verification.	Medium
Australian-hosted model endpoint	Blue Seas AI recommended model path	Candidate, subject to architecture approval	Confirm model profile, logging, retention, subprocessors and matter separation.	Medium

3. Productivity ledger

Metric	Value	Economic type	Source	Use
Client intake internal time base	$37,200/yr	Internal cost model	600 hrs x $62/hr loaded support cost	Alternative view of the same intake time; held outside the ~$569k
Client intake cost per matter	$579	Charge-out cost-to-serve estimate	1.0 fee-earner hr x $480 + 1.6 support hrs x $62	$344 recoverable per matter; about 400 new matters a year; conversion rate TBC
Client intake annual recovery	$137,520	Projected productivity value	240 fee-earner hrs x $480 + 360 support hrs x $62	600 hours returned a year
Matter delivery cost per matter	$1,066	Charge-out cost-to-serve estimate	1.6 fee-earner hrs x $480 + 4.8 support hrs x $62	$666 recoverable; about 255 property matters a year; 6.4 hrs of file work per matter
Matter delivery annual recovery	$169,830	Projected productivity value	255 fee-earner hrs x $480 + 765 support hrs x $62	1,020 hours returned a year
Advice cost per document	$1,586	Charge-out cost-to-serve estimate	3.2 solicitor hrs x $480 + 0.8 support hrs x $62	About 264 advices, letters and client updates a year
Advice annual recovery	$261,624	Projected productivity value	528 solicitor hrs x $480 + 132 support hrs x $62	$991 per document; 62.5% recovery leaves 1.5 hrs of solicitor time on every document
Total current mapped recovery	$568,974/yr	Projected productivity value	$137,520 + $169,830 + $261,624	Stated as ~$569k and ~2,280 hours; recoverable charge-out cost-to-serve, which excludes revenue, cash, margin and any guarantee
Additional property matter scenario	$168,000-$224,000/yr	Conditional revenue	60-80 additional x $2,800 fixed fee	Not additive; demand and available delivery time required
Retainer client scenario	$288,000/yr	Conditional revenue	4 additional x $6,000 x 12 - calculated from workshop inputs	Not additive; current client base and available time TBC

Assumptions behind every dollar

Fee-earner time is valued at the firm's realised charge-out rate of $480/hr across 9 solicitor fee earners. Support, paralegal and administrative time is valued at a loaded cost of $62/hr. The model returns 100% of a fully automated step, 75% of an AI-assisted step and 0% where the work stays human. Every figure is an ESTIMATE.

4. Build / buy / extend verdicts

Capability	Verdict	Why	Gate
Advice and document production	Extend the firm's precedents and house document set	Strong artefact evidence and explicit solicitor QA.	Redacted golden set and validated legal positions.
Client update drafting	Wrap approved file sources with governed drafting	The output format repeats; the legal position stays human.	Matter separation and provenance eval.
Client intake and engagement	Extend SharePoint-grounded engagement letter templates	About 80% reusable with matter-specific tailoring.	Precedent library, scope criteria and costs disclosure check.
Orchestrator	Build last	Routing adds value only after workers, gates and evals are proven.	Two workers live with stable contracts and rollback.
Trust account movements and court filings	Exclude	The firm's hard boundary, and the line the whole design is built around.	No exception in this roadmap.

5. Measurement framework

Workflow	Baseline now	Controlled test	90-day gate
Matter delivery	About 6.4 hrs of file work per property matter; about 255 matters a year	One closed matter through a governed draft + two human QA steps	Time, material error, source traceability and reviewer correction rate
Advice and document production	4 hrs per document; about 5 advices and client updates a week	Approved-source draft with the solicitor writing the legal position	Minutes per document, provenance pass, correction rate, no cross-matter contamination
Client intake and engagement	2.6 hrs per new matter; about 8 enquiries a week; conversion rate TBC	One completed engagement replayed blind against the letter that was sent	Time, scope coverage, costs disclosure exceptions and reviewer edits

6. Open data gaps

TBC before a full economic model

Enquiry volume per year, conversion rate and conflict-check effort.
Hands-on and waiting time across the mapped intake, matter delivery and document stages.
Current retainer client count, documents per client and solicitor hours per client.
Observed validation of the matter delivery stage allocations, reviewer mix and waiting time; the current model is anchored to the supplied aggregate effort.
Partner-led advisory volumes, hours and workflow.
Live tenant controls, model routing, data location, retention and logging.
The practice manager's current template and automation inventory, and the evidence attached to each.

Validation board

Gap	Evidence needed	Decision this evidence supports
Client intake baseline	Annual enquiries, conversions and conflict-check effort	Validate the intake model and the hours returned
Matter delivery stage timing	Hands-on time, waiting time and reviewer changes	Replace the 6.4-hour estimate with observed delivery data
Advice baseline	Retainer clients, document mix and solicitor effort by document type	Test the advice model without assuming volume
Production controls	Tenant, routing, retention, logging and rollback evidence	Approve or stop live client-matter use

Close order

1. Name an owner for each evidence set. 2. Capture a small observed sample before changing the model. 3. Re-run the economic ledger and roadmap gate using the observed result.