Skip to content
  1. Canada · English
  2. Four services, practical examples.

Four services, practical examples.

Explore process diagnostics, governed AI adoption, supervised document automation and incremental modernization through illustrative scenarios for Canada and international projects.

The interactive journey below follows one governed AI adoption example, from initial assessment to decision records. The browser check compares fictional records with preserved references.

Illustrative example with fictional data. The organization, addresses and volumes do not represent a real client.

Input · shadow-scan · compatible sources

Start with the logs you have.

This example starts with seven days of proxy logs. Shadow-scan fit depends on the available sources and the product documentation.

Output7 days of proxy logs

proxy.log · 2026-10-02

  • 09:14:03 · 10.0.4.17 · POSTchat.assistant-a.example/files2.1 MB
  • 09:15:40 · 10.0.7.32 · GETapp.assistant-b.example/—
  • 10:02:11 · 10.0.4.17 · POSTchat.assistant-a.example/upload840 KB
  • 10:47:55 · 10.0.9.05 · POSTapi.model-c.example/v1/chat12 KB

Discovery · shadow-scan

An inventory to investigate.

Traffic to AI services helps build an inventory of tools, departments and frequency. Logs are signals for investigation; they do not establish the content sent.

OutputIllustrative inventory: 14 services, 6 departments

Illustrative inventory · 5 of 14 services
Service7-day requestsDepartmentsFiles
Chat assistant A1,284HR, Sales, Legalyes
Chat assistant B642Marketingyes
Model API C3,910Developmentno
Corporate assistant5,106Allno
Transcription D211Leadershipyes

Classification · DezoTech team + client

Context changes the assessment.

The team and client review contracts, workflows and possible personal information to classify uses. This is a human, contextual assessment, not an automatic regulatory conclusion from traffic.

OutputUses prioritized for human review

Illustrative human assessment

  • High
    HR documents in an assistant without a contractPossible personal information; contextual review required.
  • High
    Leadership meetings in an external servicePotential strategic information.
  • Medium
    Development API use without reviewReview code and test data.
  • Low
    Queries without attachments in the corporate assistantBased on the contract and conditions in this example.

Decision · Client AI committee

An owner and a decision for each use.

The committee decides whether to allow, allow with conditions or block each use. In this example, every decision has an owner, rationale and date.

OutputDecisions documented by the committee

  • Block

    Uploads to assistants without a contract

    AI committee · October 8, 2026
  • Allow with conditions

    Corporate assistant with single sign-on

    AI committee · October 8, 2026
  • Review in 30 days

    Development API use

    CTO · October 8, 2026

Control · DezoTech team + client IT

Put the decision into practice.

The team and client IT implement the example rules, with single sign-on and human review of exceptions. A rescan helps assess the effect.

OutputIllustrative policy and rescan

AI-use-policy-v1 · example

  • AllowCorporate assistantSingle sign-on required
  • BlockUploads to unapproved servicesRules implemented by IT
  • ReviewExceptions through the AI committeeWithin 2 business days
  • RescanOctober 8 to 150 uploads to unapproved services

Evidence · SHA-256 verification in your browser

Compare with a preserved reference.

Each record is checked individually against its preserved seal. Change record 2’s summary: that record diverges and the set becomes inconsistent. Later records are not rebuilt.

OutputIntegrity checked relative to preserved seals

Calculating SHA-256 in your browser…

  1. Record 1 · discovery

    …

    14 AI services identified across 6 departments; 3 with file uploads.

    IT / Security

    Previous reference
    000000000000…00000000
    Preserved seal
    bdd66a12f77a…9f64e0a9
    Calculated now
    …
    View record data
    {
      "seq": 1,
      "ts": "2026-10-06T14:02:11Z",
      "tipo": "discovery",
      "origem": "proxy logs, September 29 to October 5",
      "resumo": "14 AI services identified across 6 departments; 3 with file uploads.",
      "responsavel": "IT / Security",
      "hash_anterior": "0000000000000000000000000000000000000000000000000000000000000000"
    }
  2. Record 2 · decision

    …

    Uploads to assistants without a contract: block. Corporate assistant: allow with single sign-on.

    AI committee

    Previous reference
    bdd66a12f77a…9f64e0a9
    Preserved seal
    5b2a009e13c7…223ec90a
    Calculated now
    …
    View record data
    {
      "seq": 2,
      "ts": "2026-10-08T17:30:00Z",
      "tipo": "decision",
      "origem": "AI committee meeting minutes",
      "resumo": "Uploads to assistants without a contract: block. Corporate assistant: allow with single sign-on.",
      "responsavel": "AI committee",
      "hash_anterior": "bdd66a12f77afece5a4dd8b3a023378b88e1bd30df7589780d293c589f64e0a9"
    }
  3. Record 3 · control

    …

    Proxy policy active. Rescan: 0 uploads to unapproved services.

    IT / Security

    Previous reference
    5b2a009e13c7…223ec90a
    Preserved seal
    704a953af901…c7069985
    Calculated now
    …
    View record data
    {
      "seq": 3,
      "ts": "2026-10-15T09:00:00Z",
      "tipo": "control",
      "origem": "AI-use-policy-v1 + 7-day rescan",
      "resumo": "Proxy policy active. Rescan: 0 uploads to unapproved services.",
      "responsavel": "IT / Security",
      "hash_anterior": "5b2a009e13c7dbd112b172aee643e80e1c44a6c583b0f5d7b27197a7223ec90a"
    }

Trust conditions: reference seals must be kept separately and obtained from a trusted source. If records and seals are replaced together, this comparison cannot establish the integrity of the original. SHA-256 alone does not authenticate an author or date. This demo does not send or save any example data.

How each service starts with the process.

These are illustrative scenarios, not client outcomes. They describe inputs, review and outputs to assess in a project; they do not run OCR, integrations or migrations on this page.

Process diagnostics and intelligence

Illustrative input
Fictional requests, service steps and records of handoffs between teams.
Planned review
The team checks the workflow, exceptions and assumptions. A baseline must be measured in a real engagement.
Expected output
A process map, dependencies and a prioritized opportunity, with criteria for testing an improvement.
Explore the service: Process diagnostics and intelligence

AI adoption with practical governance

Illustrative input
A fictional list of AI tools and the tasks where a team plans to use them.
Planned review
Responsible owners assess data, terms of use, review needs and exception handling.
Expected output
Approved uses and conditions, named owners and a task-based training plan.
Explore the service: AI adoption with practical governance

Supervised document automation

Illustrative input
A fictional order and attachments, with expected fields and a discrepancy to review.
Planned review
The responsible person checks critical fields and routes items that do not meet the rules.
Expected output
A proposed record and an exception queue, assessed against a human reference.
Explore the service: Supervised document automation

Integration and incremental modernization

Illustrative input
Fictional records from two systems with different identifiers and formats.
Planned review
The team validates mappings, rules and continuity before authorizing a change.
Expected output
A defined integration scope with reconciliation criteria and a testing, transition and rollback plan.
Explore the service: Integration and incremental modernization

The example changes with the sector and market.

Public sector is the priority in Brazil. For Canada and international projects, the initial focus is professional services, distribution, logistics and manufacturing. The applications below are possibilities to assess against data, language and local requirements.

Financial services and healthcare require specialist assessment. The proposed starting point is administrative work; sensitive financial decisions and clinical decisions fall outside this scope.

Explore financial services and healthcare

Which workflow in your organization is worth testing?

Describe how the work happens, which documents and systems are involved and who validates the result. The conversation helps define an assessment or pilot suited to your context.

Discuss your own use case