Mobile Capture Flow v0.1

Date: 2026-07-03
Status: Product flow / source of truth
Workspace area: 02-product/sales-companion/
Related files: mobile-field-companion-principle-v0.1.md, product-principles.md, WORKING_CONTEXT.md, 01-company/decision-log/decision-log.md

Core Idea

The Mobile Field Companion principle defines what Sales Companion is.

The Mobile Capture Flow defines the first concrete working moment:

A sales person leaves a visit, opens Companion on a mobile device, speaks naturally, adds photos and lets Companion turn the raw field moment into structured operational intelligence.

This is the bridge between the product philosophy and the later dashboard.

The Field Moment

The flow starts in the real-life moment after a customer visit, store walk, showroom conversation, architect meeting or planning-studio introduction.

The user is not sitting at a desk.

They may be standing on the street, in a car, on a tram platform, between appointments or at the end of a long field day.

Therefore the capture experience must be:

• voice-first,
• fast,
• tolerant of incomplete input,
• photo-aware,
• context-aware,
• structured after capture, not before capture,
• and designed for confirmation instead of manual form-filling.

What The User Gives Companion

A useful capture can contain any combination of:

• spoken visit notes,
• photos of the store, studio, showroom, products or visual environment,
• names mentioned in the conversation,
• company or location names,
• product interests,
• objections,
• planning context,
• next actions,
• files that need to be sent,
• samples that need to be prepared,
• timing signals,
• uncertainty markers.

The user does not need to know the final data structure.

The user only needs to tell Companion what happened.

Example Input — Studio Phyne, Vienna

A natural input could be:

End of Day 2 in Vienna. I visited studio phyne. I presented Pedestal for their planning work and did a short product training. It feels like a relevant lead. Follow-up: send a complete email with the key information, Masterdata, product files, assets and pricing. Also send colour samples. Contact details still need to be verified.

Companion should preserve the raw input and then extract structure from it.

Companion Processing Model

1. Capture

Companion records the raw voice note, typed note, photos and optional location/time context.

2. Normalize

Companion cleans up rough language, resolves obvious structure and keeps uncertainty visible.

3. Classify

Companion identifies the type of field event:

• customer visit,
• lead discovery,
• planning-studio introduction,
• product training,
• store walk,
• no-fit observation,
• industry conversation,
• market signal.

4. Extract

Companion extracts operational objects:

• company / account,
• contact person if known,
• location,
• visit date,
• product focus,
• relationship status,
• interest level,
• follow-up actions,
• file requirements,
• sample requirements,
• open uncertainties.

5. Connect

Companion links the visit to existing memory:

• customer profile,
• city route,
• product context,
• relationship history,
• field report,
• similar leads,
• open tasks.

6. Generate

Companion prepares usable outputs:

• visit summary,
• relationship memory note,
• follow-up checklist,
• email draft,
• file pack checklist,
• sample task,
• field-report update,
• market-learning signal,
• dashboard card.

Minimum Structured Output

For the Studio Phyne example, the structured output could look like:

visit:
  account_name: studio phyne
  city: Vienna
  country: Austria
  visit_context: End of Day 2 field route
  event_type:
    - planning studio introduction
    - product presentation
    - short product training
  product_focus:
    - Pedestal
  relationship_status: potential lead
  confidence: medium
  uncertainty:
    - direct contact name to be verified
    - email address to be verified
  follow_up:
    - send complete follow-up email
    - attach Masterdata
    - attach relevant product files and assets
    - include pricing information
    - send colour samples
  outputs_to_prepare:
    - email draft
    - file checklist
    - sample shipment task
    - relationship memory note
    - field report addendum

This is not meant to be a final database schema.

It is the first product-readable structure that turns field speech into operational memory.

Human Confirmation Gates

Companion should not silently act on uncertain details.

Before sending or finalising anything, the user should confirm:

• correct company spelling,
• contact name,
• email address,
• sample delivery address,
• correct price file / Masterdata version,
• which product assets to include,
• whether the note should be added to a Field Report,
• whether the relationship should be marked as a lead, prospect, partner or no-fit.

Follow-up Output Logic

A captured visit can create several output layers at once.

Email Draft

A short, relevant follow-up email based on the visit context.

File Pack

A checklist of files to attach or link:

• Masterdata,
• price list / terms where appropriate,
• product sheets,
• image assets,
• 2D / 3D files if relevant,
• showroom or planning material,
• colour or finish information.

Sample Task

If samples are needed, Companion creates a clear task with open address confirmation.

Relationship Memory

The visit becomes part of the account history, not just a completed task.

Product-Market Signal

The visit can teach Companion which product narrative works for which environment.

Example:

Pedestal works as a planning-office tool, not only as a retail object.

Field Report Update

If the visit is strategically relevant, Companion proposes an addition to the current Field Report or city knowledge layer.

Design Principle

The user should never be forced to think like a database while standing outside after a meeting.

The product should ask for natural field reality first and turn it into structure afterwards.

The user captures the moment. Companion carries the structure.

MVP Version

The first MVP can be extremely simple:

1. Open mobile capture.
2. Speak or type what happened.
3. Add photos.
4. Companion extracts a structured summary.
5. User confirms or corrects.
6. Companion creates follow-up actions and draft outputs.
7. The confirmed note is saved as Markdown / relationship memory.

No heavy CRM interface is needed for the first proof.

Later Dashboard Implication

The dashboard should be designed from the capture outputs, not the other way around.

The dashboard should answer:

• What happened today?
• Which visits need follow-up?
• Which leads are warm?
• Which samples must be sent?
• Which files are missing?
• Which field reports need updates?
• Which product-market patterns are emerging?
• Which relationships are strategically important?

Non-Negotiables

• Preserve raw input.
• Mark uncertainty.
• Separate facts from interpretation.
• Never invent contact details.
• Keep Markdown as source of truth.
• Use HTML as the visual surface.
• Treat the field moment as the beginning of intelligence, not as administrative burden.