← Assessment library

8 to 10 minutes

Document readiness for an AI assistant

Check whether your corpus is ready for a document or tender response AI assistant.

A reliable document assistant depends on corpus quality, access rights, freshness, structure, documents to exclude and human validation. This scanner helps decide whether to prepare the corpus, scope a pilot or defer.

ROI – Risk – Control matrix

Business value · ROI · data · risks · costs · dependencies · reversibility · implementation capacity.

Corpus

What is the primary intended use? *

One primary use helps limit the corpus, risks and test criteria.

Internal search and supportFind procedures, validated answers or internal knowledge.Customer or citizen supportPrepare answers that remain reviewed before sending.Tender responseReuse evidence, references and validated content in a controlled workflow.First-draft preparationSupport drafting without automating final approval.Other document use to scopeThe use case should be clarified before discussing a pilot.

Required field

What types of documents would be involved? *

What is the approximate document volume? *

Very limitedA few documents, easy to verify.LimitedSmall corpus usable manually.MediumSignificant volume but can be bounded for a pilot.HighHigh volume requiring preparation and sampling.Very highLarge, multi-source or complex document base.

Required field

Are formats usable? *

Hard to useScans, images, unstable exports or closed formats.FragileMixed formats or uncertain extraction quality.Usable with preparationConversion or cleaning needed.UsableFormats are mostly readable and recoverable.Pilot-readyStructured, readable formats adapted to testing.

Required field

Are documents up to date and dated? *

OutdatedOld or undated documents.Uncertain freshnessDates or versions are hard to verify.Partly up to datePart of the corpus seems reliable.Up to date on main scopeKey documents are recent or validated.Up to date and maintainedVersions, dates and validation are controlled.

Required field

Is a document owner identified? *

No ownerNo one clearly owns the topic.Unclear ownershipSeveral actors, but no clear ownership.Possible ownerA possible owner exists, without full mandate.Owner identifiedAn owner can decide or validate.Active ownerNamed, available and legitimate owner.

Required field

Quality and risks

Are access rights clear? *

UncontrolledRights are unknown or too open.FragileAccess exists, but without clear rules.PartialMain rules are known, but incomplete.ClearRights and profiles are mostly defined.ControlledRights are documented, reviewed and aligned with uses.

Required field

How sensitive are the documents? *

Non-sensitivePublic, generic or fictitious data.Low-sensitivity internalInternal data with limited confidentiality stakes.Regular internal / clientInternal or client data to protect, without extreme criticality.ConfidentialContracts, client data, HR, strategy, finance or sensitive elements.Highly sensitive / regulatedHealth, highly sensitive data, secret, security or strong regulatory exposure.

Required field

Are there duplicates or competing versions? *

No known duplicatesVersions and sources appear unique.Minor duplicatesSome variants without major impact.Likely duplicatesCompeting versions may exist.Many duplicatesSeveral sources or versions contradict each other.Contradictory corpusUncontrolled competing versions and risk of inconsistent answer.

Required field

Are documents structured and easy to segment? *

Very unstructuredLong, heterogeneous or hard-to-segment documents.Poorly structuredSome markers, but segmentation is difficult.Partly structuredUseful but incomplete titles, sections or metadata.Well structuredStructure is sufficient for a limited pilot.Very structuredUsable sections, metadata, versions and classification.

Required field

Are editorial quality and completeness sufficient? *

LowContent is incomplete, contradictory or hard to use.FragileQuality varies across documents.AdequateContent is mostly usable with review.GoodContent is clear and complete on the useful scope.Very goodContent is reliable, consistent and validated.

Required field

Are corpus languages controlled? *

UncontrolledLanguages are mixed or unidentified.Poorly controlledLanguages are known but translations or variants are uncertain.Partly controlledMain scope is clear, with some exceptions.ControlledLanguages and variants are identified.Very controlledMultilingual corpus is structured and maintained.

Required field

Is document update organized? *

Not organizedNo update rule.IrregularOccasional updates, without clear ownership.PartialA process exists on part of the corpus.OrganizedOwners and frequency are known.ManagedUpdates are tracked, dated and controlled.

Required field

Use case & pilot

Which additional uses or functional exclusions should be clarified? (Optional)

Are the use case and users clearly defined? *

Very unclearTopic is still a broad idea.Problem suspectedA pain point exists, but scope is unclear.Need identifiedProblem and users are mostly known.Need framedScope, objectives and expected decision are clear.Decision-readyThe right format can be chosen quickly.

Required field

What would the initial usage scope be? *

Very limitedA few people or isolated tests.LimitedA small team or occasional use.DiffuseSeveral teams or decentralized uses.BroadMany users or business areas involved.Very broad / uncontrolledMassive, unknown or uncontrolled usage.

Required field

How business-critical would the generated answers be? *

LowError or slowness has no major consequence.LimitedLocal or easily recoverable impact.SignificantNoticeable impact on quality, timelines or workload.HighImportant business, client, financial or reputational impact.CriticalCritical impact on decision, continuity or obligations.

Required field

What level of human validation would be maintained? *

No review plannedOutput would be used as-is.Occasional reviewReview exists, but depends on individuals.Review for sensitive casesImportant cases are reviewed, without exhaustive rules.Formalized reviewReview responsibilities and criteria are known.Systematic and tracked reviewCritical outputs are reviewed, validated and traceable.

Required field

Is the pilot scope formulated and testable? *

Not formulatedNo scope, user, corpus or exit decision is defined.Very genericAn intention exists, but the test cannot yet be prepared.Partly definedUse case and scope are identified, with incomplete criteria.Defined and testablePopulation, corpus, duration, exclusions and protocol are sufficiently clear.Ready for decisionSuccess criteria, thresholds, owner and go/no-go decision are established.

Required field

Are pilot success criteria measurable? *

Not measurableNo indicator or success criterion.Very qualitativeUseful intentions but not measurable.Some indicatorsMeasurement possible on part of the scope.MeasurableMain objectives and indicators are defined.Measurable and prioritizedIndicators, thresholds and priorities are decision-ready.

Required field

Are key stakeholders involved? *

Not involvedIT, DPO/CISO, HR or business teams are not aligned.Low involvementSome discussions, without coordination.Partial involvementSome actors are involved, but not all.Structured involvementKey actors are involved in decisions.Collective governanceRoles, decisions and responsibilities are shared.

Required field

Which documents or data should be excluded from the pilot? (Optional)

If exclusions are not yet known, leave this blank: the report will flag them for review.

Choose what happens next

You can generate the diagnostic anonymously. Contact details are only retained if you request email delivery or a conversation with BRAINYTECH.

Name

Organization

Role

Phone

Source

I would like to receive the link to this summary by email.

This authorization is limited to delivery and does not authorize commercial follow-up.

I would like BRAINYTECH to contact me about this diagnostic.

If you request delivery or contact, the contact details you provide are used only for that purpose. Privacy (2026-09-10)

This diagnostic is indicative. It does not replace a full audit. It is used to identify a first orientation and whether deeper scoping is useful.