01Bible Translation AI Advisory

AI is already entering Bible translation. The question is whether it is entering wisely.

AI can accelerate research, checking, drafting, reporting, and other parts of the translation process. But adopting a tool is not the same as improving the work.

Responsible AI adoption starts with the translation process itself. Before asking what AI can do, we establish what quality, human accountability, community ownership, consultant oversight, and publication integrity must remain protected.

Translation leaders still have to answer harder questions:

  • Where should AI participate?
  • Where should human judgment remain decisive?
  • How will translation quality be protected?
  • Who is accountable for AI-influenced decisions?
  • How will we know whether AI is creating value — or simply moving work downstream?

Human + Machine Works helps Bible translation organizations answer those questions before, during, and after AI adoption.

The sequence
  • Assess before you adopt.
  • Pilot before you scale.
  • Govern before complexity grows.
  • Assure before value quietly leaks away.
Take the AI Readiness Assessment

5 minutes · Preliminary self-assessment · No cost

Talk with an advisorFind my starting point

03Practical Tools

Start with evidence, not assumptions.

You may not be ready for a consulting engagement yet. Use these diagnostic tools to identify the question your organization needs to answer first.

AI Readiness

Is this AI use actually ready?

Assess one proposed Bible translation AI use across purpose, community authority, translation integrity, governance, data, people, technology, evaluation, and value.

Results

  • Preliminary nine-domain profile
  • Potential critical-review areas
  • Strengths and priority gaps
  • Recommended next step
About 5 minutes
Take the Readiness Snapshot

Preliminary self-assessment. Formal readiness requires evidence review and consultant calibration.

AI Policy

Which AI policies should you build first?

Answer questions about your organization, data, AI and agent use, and current security controls. Receive a prioritized policy roadmap rather than another generic checklist.

Results

  • Risk exposure
  • Security maturity
  • Build First priorities
  • Build Next priorities
  • Policy sources
About 6 minutes
Build My Policy Plan

Governance guidance, not legal advice. Policy applicability depends on organizational structure, activities, contracts, jurisdiction, and risk.

04The roadmap

One framework. Four decisions.

  1. 01AssessWhat must AI protect — and where does it belong?Package 1 — AI Readiness & Integrity Assessment
  2. 02ImplementHow should humans and AI work together?Package 2 — Human–AI Workflow Implementation
  3. 03GovernHow should AI operate across the organization?Package 3 — Organizational AI Governance
  4. 04AssureIs the system still operating as intended?Package 4 — Ongoing AI Translation Assurance

You can enter at any stage.

  • A project considering AI may begin with Package 1.
  • An organization already piloting AI may begin with Package 2.
  • An agency with fragmented AI use may begin directly with Package 3.
  • An organization with established policies and multiple active projects may begin with Package 4.
01

AI Readiness & Translation Integrity Assessment

Know what must remain true before deciding where AI belongs.

The problem

AI readiness cannot be assessed in isolation from the translation process it will enter.

Before leadership decides whether AI should assist research, checking, back translation, or drafting, it needs to know whether the existing translation process is strong enough, what quality and publication controls already exist, who holds decision authority, whether community and church participation are protected, what problem AI is expected to solve, and how improvement will actually be measured.

The real question is not simply, “Is AI good for Bible translation?” It is: is this project ready for this particular AI task, using this technology, with this team, at this stage, under these safeguards?

Start with the translation — not the technology.

Before assessing AI, we establish the controls the translation process must preserve.

Professional quality

Was it done and checked well?

Translation team, exegetical review, linguistic review, consultant checking.

Community legitimacy

Do the people understand and own it?

Representative speakers, comprehension testing, community participation.

Ecclesial accountability

Have the appropriate churches participated and accepted it?

Church review and defined authority.

Publishing integrity

Can the exact text be responsibly released?

Rights, correction closure, final version, release authorization.

These four forms of authorization are distinct — one cannot substitute for another. Package 1 evaluates the evidence around these controls before recommending where AI should enter.

What can you prove today?

Establish your current-state baseline across the evidence chain.

  • Governance
  • Community
  • Provenance
  • Exegesis
  • Language
  • Comprehension
  • Church Review
  • Consultant Check
  • Corrections
  • Rights
  • Production
  • Release
  • Stewardship
Client receives

Translation Integrity Baseline

A current-state view of what your project can already demonstrate, what is incomplete, and what must be strengthened before AI adoption expands.

Define value before choosing the AI.

What is AI actually expected to improve?

Don't stop atMeasure instead
AI draft speedTime to consultant-acceptable material
Number of outputsUseful work accepted after review
AI usageTotal workflow improvement
Machine scoreQuality evidence plus human judgment
Initial time savedRework and downstream correction
More automationHuman capability retained or increased

We establish the baseline before implementation so that a later pilot can distinguish real value from faster output that simply transfers work downstream.

Client receives

AI Value & Measurement Baseline

One project. Different AI-readiness decisions.

A project does not receive one generic AI-readiness score.

  • ResearchReadyConditionalNot Ready
  • Back TranslationReadyVerification RequiredNot Ready
  • CheckingReadyPilotRestricted
  • Alternative DraftingReady After Human DraftRestrictedNot Ready
  • AI-First DraftingPilot CandidateConditionalNot Recommended

How the assessment works

Phase 1

Establish the translation baseline

We assess:

  • Existing translation workflow
  • Quality and publication controls
  • Decision rights
  • Consultant oversight
  • Community and church participation
  • Documentation and evidence

Outcome: Translation Integrity Profile

Phase 2

Determine AI fit and value

We assess:

  • Problem AI is expected to solve
  • Current performance baseline
  • Project type
  • Translator capability
  • Language resources
  • Cultural and theological complexity
  • Proposed AI tasks
  • Technology and data risk
  • Human-development risk

Outcome: AI Readiness & Value Profile

Phase 3

Make the decision

We determine:

  • Which AI uses are appropriate
  • Which require safeguards
  • Which should remain pilot-only
  • Which require redesign
  • Which should not proceed
  • What should be measured next

Outcome: AI Use Decision + Roadmap

See your assessment develop as we work.

Clients receive access to a secure assessment workspace where they can:

  • Complete live diagnostic checks
  • Provide evidence
  • See current findings
  • Review translation-integrity gaps
  • Establish measurement baselines
  • See readiness by AI task
  • Track safeguards
  • Review consultant recommendations
  • Access final deliverables

Translation Integrity Check

What can your project prove today?

AI Value Check

What problem are you actually trying to solve?

Measurement Check

Are you measuring activity or real value?

Human Authority Check

Who decides, reviews, approves, and remains accountable?

Preview the Assessment Experience
What you receive

Translation Integrity Profile

See where your existing translation and eventual publication controls are strong, incomplete, or missing.

Project Classification

Understand the translation context in which AI would operate.

AI Value & Measurement Baseline

Define what AI is expected to improve and how improvement should be measured.

Task-Readiness Profile

Separate findings for research, checking, back translation, alternative drafting, and AI-first drafting.

AI Use Decision Matrix

See what is approved, conditional, pilot-only, restricted, or not recommended.

Human Accountability Map

Identify who recommends, evaluates, decides, approves, and can stop AI use.

Risk & Safeguard Plan

Know what must change before adoption expands.

90-Day Pilot Recommendation

Know what should be tested next and what evidence should be collected.

Leadership Decision Brief

Give senior leadership a concise decision document.

What will leadership be able to decide?
  • Approved

    Evidence supports the proposed use.

  • Approved with safeguards

    Proceed once specified controls are established.

  • Pilot only

    Promising, but insufficient evidence exists for broader adoption.

  • Redesign

    AI may be appropriate, but not in the proposed task or workflow.

  • Not recommended

    Current quality, governance, capability, data, or formation risks do not justify the use.

The value

Make the right decision before investing in implementation.

Protect against
  • Unsuitable AI investment
  • Premature AI-first drafting
  • Downstream rework
  • Increased consultant correction
  • Sensitive-data exposure
  • Translator dependence
  • Weakened community ownership
  • Unclear publication accountability
Identify opportunities for
  • Lower-risk AI uses
  • Consultant capacity recovery
  • Stronger checking
  • Better research access
  • Repetitive-task reduction
  • Measurable pilots
  • Clearer organizational decisions

The goal is neither to approve AI nor reject AI. The goal is to know where it produces defensible value and where human judgment must remain decisive.

What this assessment does not replace

This service does not:

  • Certify a Bible translation as accurate or inspired
  • Replace qualified consultant checking
  • Replace community or church authorization
  • Grant publishing approval
  • Treat a machine score as proof of translation quality
  • Claim certification on behalf of FOBAI, SIL, UBS, NIST, ISO, or UNESCO
02

Human–AI Workflow Implementation

Move from experimentation to a tested operating model.

The problem

Selecting an AI tool does not answer when AI enters the process, what humans must complete first, who evaluates AI output, which decisions require consultant involvement, what should be documented, or whether the workflow actually improves the work.

We do not just train people on a tool. We design how the work should operate — mapping the current workflow for bottlenecks, repeated work, missing controls, quality problems, unclear responsibilities, work suitable for automation, and work essential to translator development.

Then we design the future workflow: human-first drafting, controlled AI-first drafting, or AI-assisted checking. The technology follows the translation need — not the other way around.

The critical difference
  • We measure time to acceptable work, not time to machine output
  • AI can produce Draft Zero fast while increasing translator revision
  • …and consultant correction, rechecking, and source verification
  • …and community retesting, training, and downstream repair
  • Polished output can transfer effort downstream instead of creating productivity
  • So Package 2 evaluates the whole workflow, not generation speed
What you receive

Human–AI Workflow

See exactly where AI participates and where humans retain authority.

Task & Decision Matrix

Know who generates, reviews, interprets, recommends, approves, escalates, and suspends.

Project AI Use Policy

Define approved, restricted, and prohibited uses.

Approved Tool Framework

Evaluate tools against the actual use case, language, data, and security requirements.

Translator & Consultant Training

Train people not merely to operate AI — but to challenge it.

Controlled Pilot

Test the workflow on representative translation material.

Performance Dashboard

Measure more than initial drafting speed.

Final Adoption Recommendation

Adopt, modify, limit, retest, suspend, or discontinue.

What will leadership be able to decide?
  • Adopt

    The workflow demonstrated appropriate value.

  • Adopt with safeguards

    Benefits exist, but additional controls are required.

  • Expand gradually

    Use the workflow only for selected projects, books, genres, or tasks.

  • Limit AI

    Use AI for checking, research, or back translation — but not initial drafting.

  • Redesign & retest

    There is potential, but the current model failed important criteria.

  • Suspend or discontinue

    The demonstrated risks outweigh the benefits.

The value

Capture AI's useful efficiencies without transferring hidden costs to translators, consultants, or communities.

  • Time to consultant-acceptable material
  • Human rework
  • Critical errors
  • Consultant capacity
  • Translation quality
  • Translator capability
  • Community response
  • Workflow compliance

Starting at $45,000

03

Organizational AI Governance

Stop governing AI project by project.

The problem

As AI adoption expands, organizations accumulate personal AI accounts, unapproved tools, different standards across projects, bots and agents with unclear access, inconsistent data handling, duplicated software, unmanaged vendor relationships, undocumented AI influence, inconsistent consultant capability, and unclear accountability.

Package 3 moves the organization from asking “Should we use AI?” to asking “Where is AI already being used, what can it access, what is it permitted to do, and who is accountable?”

What we assess
  • AI landscape — what tools, models, agents, bots, integrations, and vendors exist
  • Projects — which AI uses suit which translation contexts
  • Data — what AI can access
  • People — who owns each consequential decision
  • Translation quality — what standards AI-assisted work must satisfy
  • Governance maturity — are your rules visible, enforceable, measurable, repeatable
What we build

Organizational AI Policy

Define appropriate use across the agency.

AI and Agent Inventory

Know what is connected to organizational work and data.

Approved & Prohibited Tool Framework

Stop every project from independently deciding what is safe.

Project & Data Classification

Connect ministry context with AI and information risk.

Human Accountability Model

Establish named decision owners.

Translation Quality Standard

Make quality criteria explicit for AI-assisted work.

AI and Agent Security Guardrails

Control data access, identity, permissions, and consequential actions.

Consultant Competency Framework

Define the capability required to govern AI-assisted translation.

Pilot Governance

Require evidence before wider deployment.

Incident Response

Know how to respond when something fails.

Executive & Board Dashboard

Give leadership visibility into material AI risks and decisions.

12-Month Governance Roadmap

Move from policy creation into operating practice.

What leadership will finally be able to answer
  • 01

    Which AI systems are we using?

    And which are approved.

  • 02

    What data can they access?

    And what agents are allowed to do.

  • 03

    Which project types may use AI drafting?

    And which decisions require human approval.

  • 04

    Who owns the risk?

    What requires escalation, and what would cause us to stop AI use.

The value

Replace fragmented experimentation with one organizational system.

  • Duplicate tools
  • Shadow AI
  • Inconsistent policies
  • Repeated project-level governance
  • Excessive permissions
  • Avoidable security exposure
  • Unclear ownership
  • Duplicated evaluation effort
  • Unmanaged AI agents
04

Ongoing AI Translation Assurance

Governance should not end when the policy is published.

The problem

AI environments change continuously. Models change. Teams change. Projects enter new genres. Consultants change. Vendor terms change. Agents gain new connections. Exceptions become habits. Documentation declines.

What was responsibly approved six months ago may no longer be the system operating today. That is governance drift.

What we monitor
  • Translation quality — are quality risks increasing
  • Workflow compliance — are teams following the approved human–AI process
  • Translator development — is AI developing people or creating dependency
  • Consultant oversight — are consultants exercising meaningful judgment
  • AI tools & models — have model or vendor changes altered the risk
  • Security & data — are approved restrictions still operating
  • Community participation — is local ownership being preserved
  • Corrective actions — are identified problems actually being fixed
What you receive

Quarterly Project Assurance Reviews

Independent, risk-based review.

Portfolio Quality & Risk Dashboard

One view across active projects.

Workflow Compliance Reporting

Evidence that approved process is followed.

Consultant Coaching

Sustained capability, not one-off training.

Translator Development Review

Capability growth, not dependency.

Model & Tool Advisory

Reassessment when the technology changes.

Incident & Corrective Action Tracking

Problems closed, on the record.

Executive & Board Risk Reporting

Material risks stated plainly.

Every project gets a clear assurance status
  • Effective

    Controls are working as designed.

  • Effective with observations

    Working, with noted improvements.

  • Corrective action required

    Specific fixes are tracked to closure.

  • Restricted

    AI use narrowed until controls are restored.

  • Suspended

    AI use stops pending remediation.

The value

Make sure AI's promised value survives real-world use.

  • Rework
  • Quality degradation
  • Consultant overload
  • Tool proliferation
  • Shadow AI
  • Data exposure
  • Translator dependency
  • Community-trust loss
  • Governance drift
  • Model changes

We call this value leakage — the difference between the benefit AI was expected to produce and the value actually realized.

Starting at $7,500 / month

08Side by side

Choose the decision you need to make.

Decision1. Assess2. Implement3. Govern4. Assure
Primary questionShould we use AI here?How should we use it?How do we govern it?Is it still working responsibly?
ScopeProjectProjectOrganizationPortfolio
Readiness assessmentTranslation + AI readinessIncludedOrganization-wideContinuous
Translation integrity baselineFullIncludedStandardizedMonitored
Workflow designFullStandardsAudit
Tool assessmentInitial risk reviewFullEnterpriseContinuous
TrainingRecommendationsFullOrganizationalCoaching
PilotRecommendedRunGovernedMonitored
AI policyProject recommendationProjectOrganizationUpdated
AccountabilityMap current decision rightsImplementInstitutionalizeVerify
Quality measurementIntegrity + value baselinePilotStandardizeMonitor
Leadership reportingDecision briefAdoption decisionGovernance dashboardRecurring dashboard
Starting investment$12.5K$45K$95K$7.5K/mo

Pricing shown is proposed starting pricing and is confirmed during scoping.

09Value

AI activity is not the same as AI value.

That is why measurement begins in Package 1 — not after the AI workflow has already been implemented.

An organization can increase every one of these and still create little meaningful value. The important question is: what changed because AI was introduced?

  • AI users
  • Tokens consumed
  • AI-generated drafts
  • Agents deployed
  • Content produced
Value created
  • Time recovered
  • Appropriate automation
  • Increased consultant capacity
  • Stronger visibility
  • Faster research
  • Better consistency detection
  • Reduced repetitive work
  • Increased organizational capability
Value leaked
  • Downstream rework
  • Consultant correction
  • Quality problems
  • Tool duplication
  • Unapproved AI
  • Sensitive-data exposure
  • Translator deskilling
  • Community distrust
  • Policy failures
  • Excessive agent permissions

We measure financial return, risk-adjusted return, and mission or human-capability return separately, rather than inventing questionable monetary values for theological formation or community trust.

10Our position

We are not selling AI adoption. We are helping organizations decide where technology belongs.

Machine may
  • Suggest
  • Generate
  • Compare
  • Organize
  • Detect
  • Retrieve
Humans remain responsible for
  • Interpreting
  • Discerning
  • Evaluating
  • Deciding
  • Approving
  • Accepting accountability

That principle runs through the whole package architecture: the consultant and responsible human actors govern the process rather than being displaced by it.

You don't need another AI strategy document. You need to know what decision to make next.

Not sure where you belong? Start with a 30-minute AI Adoption Conversation.

  • Your current stage
  • The problem you are actually trying to solve
  • Whether an assessment is needed
  • Which package fits
  • Whether Human + Machine Works is the right partner