AI-Assisted Build / Project note
MountFlow
A working shop-operations prototype connecting customer intake, multi-specimen work orders, deposits, production status, notes, history, and search.
- Stage
- Functional validation prototype · Synthetic data
- Primary users
- Taxidermists / Outfitters / Studio administrators
Evidence first
What is proven, what I changed, and what remains open.
A 90-second evidence map for hiring review. Synthetic product views explain mechanics; they never count as outcome proof.
- Verified now
- A functional multi-route browser prototype with deterministic demo data across intake, specimen-level work, status, deposits, notes, history, search, and dashboard visibility.
- Next evidence gate
- Operator discovery and usability evidence before backend, authentication, communications, or AI assistance investment.
User / workflow evidence
Supported synthesisThe prototype models specialist studios that coordinate complex physical jobs through paper, spreadsheets, messages, and memory; direct operator interviews remain the next validation step.
Product decision
Verified recordI made each specimen the atomic operational record inside a customer work order and deliberately validated the deterministic workflow before adding AI.
AI / system boundary
Verified recordCurrent behavior is deterministic. Reminders, note summaries, search assistance, and communication drafts remain future hypotheses until the operator record and workflow earn validation.
Evaluation → change
Supported synthesisThe working prototype provides a concrete discovery surface for testing intake, multi-specimen state, deposits, history, and exception visibility; no operator test result is claimed yet.
Outcome / boundary
Verified recordThe browser prototype exists. Production deployment, customers, backend, authentication, and active AI capability are not claimed.
AI product decision record
The choices behind the product, architecture, and release gate.
Prove the operational data model before adding intelligence; a trustworthy record is the prerequisite for useful AI.
AI fit
Verified recordDo not use AI in the current core; validate intake, specimen state, deposits, history, and exception visibility first.
The highest-risk assumption is whether the studio's work is modeled correctly, not whether notes can be summarized.
Alternative consideredLeading with an assistant would hide a weak operational record behind fluent copy.
Knowledge & context
Verified recordUse structured customer, work-order, specimen, payment, and history records rather than RAG.
The workflow depends on exact current state, dependencies, and timestamps.
Alternative consideredSemantic retrieval may help later with notes and search, but cannot replace transactional truth.
System architecture
Supported synthesisKeep a conventional vertical application with clear entities and routes; integrate only after operator workflow validation.
Simple architecture makes discovery findings easier to translate into the product model.
Alternative consideredMCP, agents, and cross-system automation are premature before customers, backend, and integration jobs exist.
Evaluation & release
Supported synthesisTest operator task completion, state errors, exception visibility, and whether the atomic specimen model matches real work.
The prototype exists to uncover workflow and information-architecture failures before scaling.
Alternative consideredFeature count or an AI demo would not validate the operating model.
Evidence I still need from the private project record3 open items
- 01
Operator interview and task-observation findings.
- 02
Usability completion and error evidence for intake and status changes.
- 03
Which workflow changed after operator feedback.
The product bet
Vertical workflow software for taxidermy and outfitting operations
A working vertical-software prototype that proves disciplined product scoping before adding AI.
Specialist studios often manage complex physical work through paper, spreadsheets, messages, and memory, increasing errors and obscuring status.
- Primary user
- A specialist studio operator managing complex physical jobs across customers and multiple specimens.
- Trigger
- A customer arrives with one or more specimens requiring intake, deposits, production steps, communication, and delivery.
- Inputs
- Customer details, specimen records, services, measurements, deposits, notes, dependencies, and status history.
- Product action
- Create one customer work order with specimen-level records, track state and money, preserve history, and expose operational exceptions.
- Output
- A searchable work record, production queue, deposit status, timestamped history, and delivery checklist.
- Decision enabled
- What work is ready, blocked, overdue, awaiting payment, or ready for customer communication.
Public-safe product demonstration
A complete workflow, without protected data.
The names, records, amounts, dates, scores, and thresholds inside this surface are fictional. The product logic is the point.
A fictional preservation studio receives a shoulder-mount request and needs one traceable record from intake to delivery.
- Primary user
- Studio coordinator
- Decision enabled
- Keep customer expectations, specimen custody, craft work, and delivery aligned.
Create one trusted record at the handoff.
Customer preferences, condition notes, custody acknowledgement, and reference photos enter the same job record.
- Pose
- Left turn
- Finish
- Natural habitat
- Custody
- Studio
- Intake
- Work order
- Production
- Delivery
Reference photos and condition notes stay with the item.
Customers, specimens, dates, and production records are synthetic. This is a public-safe workflow demonstration.
Product decision and trade-off
AI features are attractive, but the real product risk is whether the shop's operational record is modeled correctly.
- Product choice
- Make each specimen the atomic record inside a customer work order and validate the deterministic workflow first.
- Rejected alternative
- Leading with an AI assistant before intake, status, deposits, and history are trustworthy.
- Product consequence
- The prototype becomes a credible discovery surface and future AI has a durable operating context.
A status update at the customer level hides that one specimen is blocked while another is ready.
Specimen-level state, dependency and deposit checks, timestamped events, and a customer-level roll-up.
every job state can be traced to one specimen, one owner, one timestamp, and one next action before delivery.
The leadership pattern in its appropriate form.
This case is not retroactively enlarged into a Director mandate. It shows which product-lead behaviors were already present at this scope.
Translated an unfamiliar vertical workflow into a concrete multi-route prototype.
Used coding agents as a delivery team while retaining scope, information architecture, and acceptance responsibility.
Director-ready judgment appears in sequencing: validate the operating model, then invest in intelligence and scale.
Product founder and AI-assisted builder defining the workflow, information architecture, page system, and product direction.
Evidence boundary
Impressive because the boundary is clear.
A functional multi-route browser prototype exists with deterministic demo data, intake, statuses, deposits, notes, history, search, and dashboard visibility.
- Built a functional multi-route validation prototype with deterministic demo data.
- Connected intake, status changes, deposits, notes, timestamped history, search, and dashboard visibility.
- Created a concrete test surface for operator discovery without implying production deployment.
Customers, specimens, prices, dates, and the walkthrough record are fictional.
No production deployment, customer adoption, backend, authentication, or active AI capability is claimed.