Multimodal workflow context, one job end to end: a walk-in freezer service call

Short answer

Multimodal workflow context means the photos, notes, forms and readings a job produces stay attached to that job, in order, with its outcome. In an ordinary field-service call, a technician photographs a fault, writes what it means, gets a repair approved, photographs the fix and closes the job; a month later the system shows whether it held. Each step leaves a record. Together they show how what someone saw led to what they did.

Visual briefFull-size diagram
A seven-step workflow for a walk-in freezer service call plays from left to right, intake, dispatch, arrival photo, diagnosis note, customer approval, repair photo and 30-day check, and each step drops a record card below it, with the coil photo, technician note, signature and 30-day outcome highlighted as carrying the signal.
Illustrative job at a fictional commercial refrigeration service company.

This walkthrough is part of how to assess multimodal workflow records without uploading raw assets. Where the main guide uses a short rooftop-unit case, this page follows a different job through every step, including the decision points, and marks which records carry the signal an AI-data review looks for.

The job: a walk-in freezer that will not hold temperature

Illustrative and fictional. A 380-person commercial refrigeration service company has a maintenance contract with a restaurant group. At 06:50 one restaurant's kitchen manager reports that the walk-in freezer reads above its set point and ice is building on the evaporator. Stock is at risk, so the job is urgent.

The company runs a field service app on technicians' phones. Like many such apps, it lets technicians attach images and text notes to the booking and take pictures inside inspection forms with a caption; Microsoft documents both features for its Dynamics 365 Field Service mobile app. The company's own app may differ, which is a question for the owner, not an assumption.

Step by step: decisions and the records each one leaves

StepTimeDecision madeRecord left behindMode
1. Intake06:52Classify as urgent under the contractWork order with asset ID, problem code and priorityStructured text
2. Dispatch07:05Send the nearest certified technicianAssignment with technician ID and travel startStructured, time
3. Arrival07:48What to inspect firstPhoto of the iced evaporator coil, taken in the appImage, capture time
4. Diagnosis08:10Defrost timer failed; heater intactTechnician note and inspection form answersText, form
5. Approval08:25Repair now versus quote a full controller upgradeQuote line and the kitchen manager's signatureStructured, signature
6. Repair09:30Replace timer, run a manual defrostPhoto of the cleared coil, parts used, closing readingImage, structured
7. CheckDay 30Did the fix hold?No repeat call for the asset; next routine reading in rangeSystem status

Seven steps, records in several forms (structured fields, photos, notes, a form and a signature), one job number joining them. Remove the job number and the arrival photo is an anonymous picture of ice.

Which records carry the signal

Not every record is equally useful. An AI-data review looks for the moments where an observation turns into a judgment, and for proof of what happened next.

High signal.

  • Step 3 photo with step 4 note. The photo shows what the technician saw; the note says what it meant. Neither explains itself alone. The pairing is the core of multimodal workflow context.
  • Step 5 approval. The technician offered two options and the customer chose one. That is a recorded decision with a reason (cost, downtime) attached.
  • Step 7 outcome. No repeat call within 30 days labels the diagnosis as probably right.

Supporting.

  • Step 6 after-photo and reading. Confirms the end state matches the note.
  • Step 1 problem code. Lets many similar jobs be grouped and compared.

Low signal on its own.

  • Step 2 dispatch. Useful for scheduling studies, but it says nothing about the fault.

Timing holds the pairing together. Camera files can carry an Exif DateTimeOriginal tag, the time the image data was generated, defined separately from the time it was digitized. If the app keeps that capture time, the review can confirm the arrival photo came before the diagnosis note and the after-photo came after the repair. If only the upload time survives, the order can blur.

What this job looks like when a step is missing

Three common breaks, each removing a different piece of the chain:

Missing stepWhat the company still holdsWhat the case loses
Arrival photo taken on a personal phoneNote, approval, outcomeThe observation the note interprets
Approval given verballyPhoto, note, repair, outcomeThe decision between two options
No asset link on repeat callsEverything up to closureWhether the repair held

A complete version of this job across thousands of calls is the strong case. A company where step 5 is usually verbal still holds useful observation-to-action pairs, and should be described that way. The field-level review packet lists the documents that describe these records before anyone looks at a file.

Recognizing the same workflow at a client

The refrigeration call is one instance of a common shape: someone inspects a physical thing, records what they saw in a picture and words, a decision follows, and a later event confirms the result. The same shape appears in property maintenance, elevator service, fleet repair, roofing and insurance inspection. Ask the operations director, at metadata level:

  1. Where do job photos live, and are they attached to the job inside the app?
  2. Is the note written by the technician on site, in the same job?
  3. Are repair options and customer approvals recorded?
  4. Does a repeat visit link back to the asset or the original job?
  5. How many years of jobs does the system hold, and who owns it?

If the answers are mostly yes, the multimodal workflow context scorecard helps decide whether to introduce. The broader record type is described under multimodal operating records, and why a linked chain beats a pile of files is explained in the guide to useful workflow context instead of isolated files. A non-field example of the same idea is in the logistics workflow example pack.

Who owns these records, and what stays unknown

The service manager knows how technicians use the app. The owner, CEO or COO usually decides whether to discuss licensing. Client contracts matter: photos taken on a customer's premises may be restricted, and images can show people, vehicles or addresses that need removal. Some data marketplaces make this an entry condition; Defined.ai's partnership page, checked on October 18, 2026, requires partners to hold rights to commercialize image, text or multimodal data for AI training.

A partner never asks to see photos or notes. What stays unknown until SourceX reviews it: rights to client-site imagery, image quality after any compression by the app, how many jobs have all seven steps, and whether a buyer wants this record type. The company also needs 50+ full-time employees at peak (contractors excluded), documented operating history, rights to license and an authorized contact. Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English. Other record families are under data types.

Next step

Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

If a client's jobs leave a photo, a note and an outcome under one job number, register as a partner, confirm the company against who qualifies and read the terms before you introduce its owner.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

Does the job need video or audio to count as multimodal?

No. A photo and a written note tied to the same job is already two modes. Video, audio or sensor readings add more, but the value comes from the link between modes, not the number of them.

What if the customer approval was verbal?

Then the decision step leaves no record, and the case loses its middle. Note it as a gap. Many companies capture a signature or an emailed quote for larger repairs and not for small ones.

Is this example based on a real company?

No. The company, the job and every value are fictional, built to show how records accumulate along one ordinary workflow.

Sources

  1. learn.microsoft.com/en-us/dynamics365/field-service/mobile/get-w…
  2. learn.microsoft.com/en-us/dynamics365/field-service/inspections
  3. cipa.jp/std/documents/e/DC-008-2012_E.pdf
  4. defined.ai/partnership-programs
  5. refer.sourcex.si/terms

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment