Support: explanation and next steps
Diagnosis and proposal to review and communicate.
What we doYour need.
Your need.
Our technology.
Your backlist, content and knowledge can underpin new businesses. We combine OCR, search and connected assistants to create services you can sell and reduce the cost of producing them.
01 · Connected knowledge · MCP
Connect my content to AI
Publishers · Libraries · Museums
Catalogues, works and contentQuestions answered with source references
MCP connectors bring knowledge and services into AI assistants, in natural language and with the access you choose.
- Semantic search, relationships and knowledge graphs
- Passages and answers with source references
- Access for teachers, researchers and the general public
Your knowledge can earn revenue through specialist access. Together we design the service, its audience and its terms of use.
Explore this for my organisation See it in a projectbaiq
SOURCESDocument and context
WorkPassageReference
QUESTION / CONTEXT
Which document contains this idea?
THIS STAGE’S RESULT
Query of authorised sources
Fictional example · natural-language question
01 / 03 · Ask
People start with their need.
The assistant interprets the question and selects available tools to consult sources.
Query of authorised sources
baiq
SOURCESDocument and context
WorkPassageReference
STARTING WITH
Query + permitted tools
THIS STAGE’S RESULT
Passage + document + location
MCP retrieves information; the assistant writes.
02 / 03 · Connect
MCP brings the source closer.
The protocol connects tools and sources according to defined permissions; retrieval provides evidence to the assistant.
Passage + document + location
baiq
HOW THE PROCESS CONNECTSFrom sources to a grounded answer
How should I store a watercolour?
ORGANISATION CONTENTConservation guidePaper · watercolours · care
MCPFinds the passage
and its reference
and its reference
RETRIEVED SOURCE“Store on an acid-free support, away from direct light.”
[1] Conservation guide · p. 12
ASSISTANT SYNTHESIS
Use an acid-free support and keep it away from direct light. [1]
The answer draws on a passage you can check.03 / 03 · Check
Responses lead back to the source.
The assistant prepares the answer from retrieved material. People review references and make decisions.
Answer + sources to check it
Technology prepares and organises. People review and decide.1 / 3
02 · Specialist support
Support and resolve with data
Customer support · Internal help desk · Operations
Issues, guides and authorised recordsPrior analysis and resolution proposals
Agents analyse issues before a person handles them. We connect guides, logs, records and error signals through MCP to prepare diagnosis and proposals using the necessary information, according to team permissions.
- Prior analysis and proposals when the issue arrives
- Specialist queries of relevant logs and records
- Reviewable replies or evidence-based escalation
Less repeated investigation and waiting between teams: an opportunity to handle more enquiries with the same resources and improve service margins.
Explore this for my organisation See it in a projectbaiq
baiqFICTIONAL SCENARIO
“The import does not finish.”DEMO-1044
QUESTION / CONTEXT
“The import does not finish.”
THIS STAGE’S RESULT
Affected job: demo-job-27
Fictional scenario · relevant data only
01 / 04 · Identify
An issue has context.
The affected task and necessary information are identified according to team permissions.
Affected job: demo-job-27
baiq
baiqFICTIONAL SCENARIO
demo-job-27
Status: stopped
Signal: manifest reader failure
Source: authorised recordSTARTING WITH
Job + guide + error signal
THIS STAGE’S RESULT
Manifest reader failure
Only the information needed for this issue, according to team permissions.
02 / 04 · Retrieve
Authorised sources provide evidence.
MCP connects guides, records and error signals relevant to that issue.
Manifest reader failure
baiq
baiqFICTIONAL SCENARIO
demo-job-27
Status: stopped
Signal: manifest reader failure
Source: authorised recordSTARTING WITH
Stopped import + manifest error
THIS STAGE’S RESULT
Bug identified · diagnosis prepared
Analysis distinguishes evidence from proposals.
03 / 04 · Analyse
AI prepares the diagnosis.
The agent checks information and prepares proposals before a person handles the issue.
Bug identified · diagnosis prepared
baiq
baiqFICTIONAL SCENARIO
demo-job-27
Status: stopped
Signal: manifest reader failure
Source: authorised recordSTARTING WITH
Processed diagnosis
THIS STAGE’S RESULTCUSTOMER SUPPORT ENGINEERING
Issue created by the agent DEMO-BUG-027
Engineering: reproduction + expected/observed + evidence
04 / 04 · Orchestrate
Each team receives its part.
Support receives the diagnosis and proposed reply. When a bug is found, the agent directly creates the engineering issue.
Engineering: reproduction + expected/observed + evidence
Technology prepares and organises. People review and decide.1 / 4
03 · Digitisation · OCR
Recover and digitise my collection
Publishers · Libraries · Archives
Scans, columns and footnotesStructured text ready for review
OCR for complex documents. Columns, chapters, notes and references become content you can edit, find and reuse.
- Layout detection and reading order
- Paragraph and footnote reconstruction
- Visual verification, manual editing and traceability
AI transcribes; your team verifies. Less recovery work to explore new digital editions and licensed research collections.
Explore this for my organisation See it in a projectbaiq
THIS STAGE’S RESULT
Real scan sample · illustrative workflow reconstructionRecognised lines + coordinates
Page → regions → reading order
01 / 05 · Recognise
The scan retains its reference.
Recognising lines and their positions supports reconstruction while retaining the source page.
Recognised lines + coordinates
baiq
ILLUSTRATIVE RECONSTRUCTIONA recovered chapter
The text joins the page’s lines and retains the footnote marker.1
1 Footnote linked to the paragraph.
Heading · Paragraph · Footnote · ReferenceSTARTING WITH
Page ending + beginning of the next
THIS STAGE’S RESULT
Continuous paragraph · linked footnote
Heading / body / marker / footnote
02 / 05 · Reconstruct
The paragraph continues across pages.
Neighbouring-page context helps join lines and distinguish body text from footnotes.
Continuous paragraph · linked footnote
baiq
ILLUSTRATIVE RECONSTRUCTIONA recovered chapter
The text joins the page’s lines and retains the footnote marker.1
1 Footnote linked to the paragraph.
Heading · Paragraph · Footnote · ReferenceSTARTING WITH
Recognised lines and footnotes
THIS STAGE’S RESULT
Chapter → heading → paragraphs → footnotes
Each block retains its source reference.
03 / 05 · Organise
Structure becomes editable again.
Headings, paragraphs and notes are assembled into a structure editors can compare against the document.
Chapter → heading → paragraphs → footnotes
baiq
ILLUSTRATIVE RECONSTRUCTIONA recovered chapter
The text joins the page’s lines and retains the footnote marker.1
1 Footnote linked to the paragraph.
Heading · Paragraph · Footnote · ReferenceSTARTING WITH
Source and text side by side
THIS STAGE’S RESULT
Checked reading · recorded change
Automatic reading proposes; editors decide.
04 / 05 · Check
Review returns to the original.
Editors check uncertain readings and correct the text while retaining revision history.
Checked reading · recorded change
baiq
ILLUSTRATIVE RECONSTRUCTIONA recovered chapter
The text joins the page’s lines and retains the footnote marker.1
1 Footnote linked to the paragraph.
DOCXSTARTING WITH
Checked text with footnotes
THIS STAGE’S RESULT
Word · native footnotes
Reissue / review / translation preparation
05 / 05 · Export
Text returns to editorial work.
Editable output supports reissues and footnote reuse without transcribing them again.
Word · native footnotes
Technology prepares and organises. People review and decide.1 / 5
04 · Digital assets · Semantic DAM
Find and reuse my assets
Publishers · Libraries · Culture
A digital asset repositorySearch by meaning and reuse
Digital asset repositories explored in natural language. In OLA, built for Grupo Anaya, teams can describe an image, find related assets and reuse them in new content.
- Image search by meaning
- Catalogue filters from natural language
- Editorial reuse with asset traceability
Make existing assets discoverable for new publications or explore licensing the resources you can sell.
Explore this for my organisation See it in a projectbaiq
JPG · IllustrationQUESTION / CONTEXT
“A wetland to explain biodiversity”
THIS STAGE’S RESULT
Subject: wetland · use: educational resource
Fictional example · search within your collection
01 / 03 · Describe
Describe the image you need.
The request expresses the subject and use conditions even when the filename is unknown.
Subject: wetland · use: educational resource
baiq
JPG · IllustrationSTARTING WITH
Subject + format + catalogue metadata
THIS STAGE’S RESULT
Related assets for selection
Teams check result relevance.
02 / 03 · Find
Meaning and metadata together.
Semantic search gathers related assets and natural-language requests become catalogue filters.
Related assets for selection
baiq
SEMANTIC DAMOne idea connects your whole archive.
IllustrationYOUR REPOSITORYDifferent formats. Connected knowledge.
Resources to explain the biodiversity of a wetland
Wetlands and biodiversityImage · LandscapeEcosystem
The water that connects usVideo · Relevant excerptNatural cycle
Listen to an ecosystemAudio · With transcriptBiodiversity
Wetland atlasDocument · Related pageConservation
From your collection to your next creationSelection · Reuse · Source reference
03 / 03 · Reuse
Assets retain their origin.
People select an asset, review its conditions and retain the original reference when reusing or adapting it.
Asset ready for editorial work
Technology prepares and organises. People review and decide.1 / 3
05 · Intelligent tutors
Support every learner
Educational publishers · Schools · Training
Your curriculum contentPersonalised practice and guidance
Socratic tutoring, personalised practice and assisted grading grounded in curated content. More support and more time to teach.
- Questions and hints guided by progress
- Handwritten answer recognition
- Teacher oversight and judgment
More support for learners and a new offering for your customers: educational services built on your content and teaching approach.
Explore this for my organisation See it in a projectbaiq
CONSTANT SPEED · FICTIONAL EXAMPLEFrom two hours to one.
2 h 60 km÷ 21 h ? km
QUESTION / CONTEXT
60 km in 2 hours. How far in one hour?
THIS STAGE’S RESULT
Relate time and distance
Fictional example · constant speed
01 / 04 · Establish
Start with what the learner understands.
Goals and prior ideas guide support grounded in selected content.
Relate time and distance
baiq
CONSTANT SPEED · FICTIONAL EXAMPLEFrom two hours to one.
2 h 60 km÷ 21 h ? km
QUESTION / CONTEXT
From two hours to one: what changes?
THIS STAGE’S RESULT
Time ÷ 2 → distance ÷ 2
At constant speed, both quantities change in the same proportion.
02 / 04 · Ask
A hint invites reasoning.
Questions and hints help build explanations while preserving the learner’s reasoning.
Time ÷ 2 → distance ÷ 2
baiq
CONSTANT SPEED · FICTIONAL EXAMPLEFrom two hours to one.
2 h 60 km÷ 21 h 30 km
STARTING WITH
60 km ÷ 2 = 30 km
THIS STAGE’S RESULT
One hour → 30 km
Apply the idea: 120 km in 4 h → 30 km in 1 h
03 / 04 · Practise
The next question builds on progress.
Practice adapts to emerging questions and the content being studied.
One hour → 30 km
baiq
04 / 04 · Support
Teachers follow the reasoning.
Indicative progress shows prior ideas, questions and next steps for teacher review.
Prior ideas · questions · next steps
Technology prepares and organises. People review and decide.1 / 4
06 · AI-assisted publishing
Speed up creation and review
Publishers · Production teams · Training
Manuscripts and editorial criteriaMaterials ready for review
Drafts, activities, reviews, metadata and marketing materials. Technology prepares the work; your team brings purpose, judgment and voice.
- Activity generation and review
- Reports, style and proofreading
- Multiple formats and accessibility
Reclaim editorial time to develop new titles and supplementary materials. Less repetitive work, more capacity to expand your offering.
Explore this for my organisation See it in a projectbaiq
SCIENCE · FICTIONAL EXAMPLEWho produces its own food?
SunlightPlantDeer
STARTING WITH
Sunlight → plants → deer
THIS STAGE’S RESULT
Goal: recognise a food chain
Fictional example · science content
01 / 04 · Start with text
Content guides the activity.
Editorial material and level guidance define what each proposal should cover.
Goal: recognise a food chain
baiq
SCIENCE · FICTIONAL EXAMPLEWho produces its own food?
SunlightPlantDeer
STARTING WITH
Goal + level + activity type
THIS STAGE’S RESULT
Identify the producer / explain the relationship
Answer: plants are the producers.
02 / 04 · Generate
Variants for the same goal.
Batch proposals support the same content through different activities.
Identify the producer / explain the relationship
baiq
SCIENCE · FICTIONAL EXAMPLEWho produces its own food?
SunlightPlantDeer
STARTING WITH
Activity + proposed answer
THIS STAGE’S RESULT
Consistent question · checked answer
Structure / content / answer review
03 / 04 · Check
The proposal must be reviewable.
Structure is validated and questions and answers are reviewed, repairing detected problems.
Consistent question · checked answer
baiq
HOW THE PROCESS CONNECTSOne lesson. Purposeful activities.
SCIENCE · SOURCE LESSON01
Where does the energy come from?SunlightProducerConsumer
Plants use light to make their food. Herbivores obtain energy by eating them.
TWO WAYS TO PRACTISE
Teacher judgement before publication
Review level, wording and expected answer.
04 / 04 · Validate
The team retains judgment.
People review and adapt materials before incorporating them into publication.
Material ready for editorial review
Technology prepares and organises. People review and decide.1 / 4
07 · Translation and localisation
Bring my content to more languages
Publishers · Culture · Training
Text, images and glossariesConsistent drafts ready for review
Translation memories, glossaries and author profiles. We also localise text in images and learning materials, reducing repeated production work.
- Retrieval of previous translations
- Meaning and consistency checks
- Image localisation and visual review
Reduce repetitive translation preparation and assess new editions and markets. Your team keeps control over meaning and voice.
Explore this for my organisation See it in a projectbaiq
Al atardecer, la luz atraviesa la vidriera.
At dusk, light passes through the stained-glass window.
STARTING WITH
Source + reference translations
THIS STAGE’S RESULT
Precedents related to the passage
Reference retained to check the choice.
01 / 05 · Retrieve
Earlier decisions provide context.
Search gathers similar sentences and the publisher’s reference translations.
Precedents related to the passage
baiq
Al atardecer, la luz atraviesa la vidriera.
At dusk, light passes through the stained-glass window.
STARTING WITH
Term: vidriera
THIS STAGE’S RESULT
Equivalent: stained-glass window
Fictional example · glossary and editorial context
02 / 05 · Contextualise
Terminology has a memory.
Glossaries and author profiles help preserve meaning and consistency across works.
Equivalent: stained-glass window
baiq
Al atardecer, la luz atraviesa la vidriera.
At dusk, light passes through the stained-glass window.
STARTING WITH
Al atardecer, la luz atraviesa la vidriera.
THIS STAGE’S RESULT
At dusk, light passes through the stained-glass window.
Fictional ES → EN example · draft awaiting review
03 / 05 · Draft
A draft to work with.
The translation workflow or connected assistant prepares a contextual draft; MCP provides references.
At dusk, light passes through the stained-glass window.
baiq
Al atardecer, la luz atraviesa la vidriera.
At dusk, light passes through the stained-glass window.
STARTING WITH
Source ↔ proposed translation
THIS STAGE’S RESULT
Meaning / omissions / terminology
Translators assess and refine proposals.
04 / 05 · Check
Meaning is checked against the source.
Omissions, meaning and consistency are checked before accepting the draft.
Meaning / omissions / terminology
baiq
HOW THE PROCESS CONNECTSMeaning travels too.
EDITORIAL CONTEXTA cultural guide
Keep the imagery, tone and terminology.
Al atardecer, la luz atraviesa la vidriera y proyecta reflejos azules sobre el suelo de piedra.
¶ 04At dusk, light passes through the stained-glass window and casts blue reflections on the stone floor.
¶ 04SHARED TERMINOLOGY
vidriera ↔ stained-glass window
The final reading is human.Check meaning · refine style · approve
05 / 05 · Deliver
Each change can be assessed.
Proofreading and explained changes prepare the text for editorial decisions.
Reviewed text + explained changes
Technology prepares and organises. People review and decide.1 / 5
Related solutions · Bespoke developmentThe same bridge.
The same bridge.
More independent teams.
Bespoke applications for your sources, tools and procedures.
Customer support
Customer issue + guides + service status.
Receive diagnosis and a proposed reply before handling the case. When a bug is found, the agent creates the engineering issue.
What the team receivesPrior analysis, a reviewable reply and a next step.
Less waiting for customers. Fewer cross-team requests.
Design this for my teamInternal help desk
Team procedures, documentation and tools.
Resolve tool, access and process questions using current procedures. Route each request to the right owner.
What the team receivesPractical steps, source and owner.
Fewer interruptions for specialists. More team autonomy.
Design this for my teamSystems incidents
Logs, error events, operational records and technical guides.
Connect logs, errors and jobs to prepare the diagnosis. Open a technical issue with reproduction steps, relevant evidence and a proposed intervention.
What the team receivesTraceable diagnosis and a prepared engineering handoff.
Less repeated investigation. More useful technical discussions.
Design this for my teamCompany onboarding
Welcome guides, procedures and role-specific knowledge.
A role-specific plan: what to learn, where to find it and who to ask. A starter checklist and requests following team approvals.
What the team receivesA plan, checklist and contextual help.
Fewer blocking questions. More time to support the person.
Design this for my teamRelevant information. Defined permissions. Human review.
MCP connects authorised sources. Each person receives the necessary context; changes and approvals follow the team’s procedure.
Specific questions.
What does MCP bring to a publisher or library?
MCP is a protocol connecting AI assistants to sources and tools. It enables book discovery, passage retrieval or service access from compatible assistants such as ChatGPT or Claude, depending on configuration and permissions. You decide which content is exposed and which operations are allowed.
Can you work with scans and historical documents?
Yes. We have developed workflows that detect layout, recognise text, reconstruct paragraphs and footnotes, check results against the image and retain source references. We assess each collection because document condition and composition affect results.
Who keeps editorial control?
Your team. We design AI to prepare, suggest and check; people review decisions that require judgment. Together we define validation points and permissions.
Your knowledge could open a new business opportunityYou have the knowledge.
Let’s explore your opportunity ↗