Guide
Help visitors or staff find the relevant service, document, process or form without navigating the whole information structure unaided.
Chinkara Maven
An assistance layer with visible sources, limits and human control.
Chinkara Maven is GA Applications' concept and interaction layer for guided digital assistance. It is designed to help a person navigate approved knowledge, form a useful draft or reach the right next action. Maven is not presented as an infallible authority, autonomous employee or substitute for professional judgement.
Aska bounded question
Supportwith visible sources
Escalatewhen judgement is required
Why it matters
Guide
Help visitors or staff find the relevant service, document, process or form without navigating the whole information structure unaided.
CHAPTER 01 / What Maven is
A designed path between a question, trusted material and a controlled next step.
Maven gives the assistance experience a recognisable voice and interface while the underlying design remains practical: understand the request, find permitted context, make the boundary visible and keep important decisions with the appropriate person.
Explain
Turn approved source material into a concise answer that can point back to its governing page or record.
Prepare
Collect structured context or create a draft for review rather than silently committing the organisation.
Route
Move an approved request into contact, support, CRM or another workflow with the person's intent attached.
CHAPTER 02 / What Maven is not
Personality should never conceal uncertainty or authority.
Not an infallible source
Answers can be incomplete, out of date or wrong. Important information should expose its source and review path.
Not automatic professional advice
A conversational response does not become legal, financial, medical, engineering or other regulated advice merely because it sounds confident.
Not unlimited access
The interface should not imply that it may read every company record or act as every user. Permissions still govern retrieval and action.
Not silent autonomy
Material messages, record changes, approvals or external actions need explicit authority appropriate to their risk.
CHAPTER 03 / The control loop
Request → approved sources → draft → human review → permitted action.
This sequence is the central Maven design pattern. A lower-risk navigation answer may complete quickly; a sensitive operational request should stop at review or escalation. The system should record enough context to explain what source and permission shaped the result.
1. Request
Capture the person's intent while discouraging unnecessary confidential or sensitive data.
2. Retrieve
Search only sources approved for that audience and task.
3. Draft
Prepare an answer, summary, classification or next-step proposal with uncertainty visible.
4. Review
Ask an authorised person to confirm outputs or actions where impact requires it.
5. Act or route
Use a specifically permitted connection and preserve an appropriate record of what happened.
CHAPTER 04 / Demonstrable maturity
Separate the interface available now from a roadmap for later.
The current public Maven experience should be understood as a guided website entry and demonstration of the assistance pattern unless a page expressly identifies a supported integration. Future capabilities must remain labelled as roadmap until built, tested, authorised and supported.
Demonstrable now
Guided service discovery, transparent answer patterns, structured project context and routing to the relevant GA page or enquiry path.
Project-dependent
Connections to private knowledge, CRM, dashboards, document stores or operational actions require separate source, permission, security and support design.
Roadmap, not a promise
Broader multimodal input, deeper workflow assistance or additional channels remain possibilities until an approved release describes them as functioning.
CHAPTER 05 / Where it can live
One assistance pattern can take different roles without pretending they are identical.
Public website guide
Help a visitor understand services, compare paths and prepare a useful enquiry from approved public content.
Staff knowledge assistant
Find controlled internal procedures or reference material according to role and source permissions.
Document preparation
Extract structured context or prepare a draft for checking against the original source.
Workflow companion
Explain the next permitted step, identify missing information and route an authorised request into a connected system.
Customer guidance
Answer bounded product or service questions and escalate uncertainty to a person without pretending to resolve every matter.
CHAPTER 06 / Trust architecture
The useful intelligence is partly in what the system refuses to hide.
Source boundary
Show which approved material supports an important response and when that material was last reviewed where relevant.
Permission boundary
Retrieve records and expose functions according to authenticated role, task and least-needed access.
Action boundary
Separate an answer or draft from an external message, approval, transaction or record change.
Human boundary
Make escalation, correction and override visible before a user depends on an uncertain output.
Retention boundary
Decide what conversation or action history is retained, why, for how long and who may inspect it.
CHAPTER 07 / Relationship to GA services
Maven is the interaction layer; AI Assistants is the service that designs the system.
The AI Assistants service assesses use case, source quality, permissions, review, privacy, integration and failure handling. Maven provides a branded design language for selected assistance experiences. CRM can hold authorised context, automation can route an approved event, and dashboards can show operational signals—each connection remains separately scoped.
AI Assistants
Design the knowledge, permission, review and evaluation model.
Website design
Place the assistant within a clear visitor journey and accessible interface.
CRM and automation
Carry only authorised structured context into defined records and workflows.
Dashboards
Monitor selected usage, exceptions and outcomes without exposing private conversation content to analytics.
CHAPTER 08 / Fit assessment
Sometimes the best assistant is a better page, form or search function.
Maven is appropriate when people ask varied questions, relevant knowledge can be controlled and a conversational or guided path materially improves the task. It is a poor fit when the answer must always be deterministic, source material is unreliable, permissions cannot be enforced or a simple interface solves the need with less risk.
Good signal
Repeated information-finding work, multiple approved sources and a clear human escalation route.
Caution signal
High-impact decisions, weak records, unclear ownership, sensitive data or expectations of flawless autonomy.
Simpler alternative
Improve navigation, structured content, filters, deterministic rules or a well-designed form before adding generative behaviour.
Continue through the chinkara maven system.
Each route answers a different operating question while retaining the same standards for evidence, responsibility and a usable next step.
Route 01
About
See how GA Applications approaches websites, business systems, automation and responsible digital design as one connected project.Route 02Process
Follow GA Applications' eight-phase path from diagnosis and scope through design, build, launch, handover and measured improvement.Route 03Pricing
Understand how scope, content, integrations, risk and support shape a GA Applications estimate without misleading one-size-fits-all packages.Route 04Faq
Direct answers about GA Applications services, delivery, pricing, ownership, SEO, AI, support, locations and independent concepts.Route 05Support
Report an issue, request a controlled change or find the right support path for a GA Applications website or connected system.Route 06Warranty
Understand how a potential defect is distinguished from a change and assessed under the applicable GA Applications agreement and law.Route 07Refunds Cancellations
Understand how a cancellation or refund request is reviewed against completed work, commitments, third-party costs, written terms and law.Questions worth resolving
What to clarify before committing.
Is Chinkara Maven a separate AI service?
Maven is GA Applications' branded assistance concept and interaction layer. The AI Assistants service is the broader work of assessing and building the sources, permissions, review, integration and evaluation behind an appropriate assistant.
Can Maven access our private systems?
Not by default. Any private source or system connection requires explicit scope, authorised access, security controls, data handling decisions and testing.
Will Maven always give the correct answer?
No. AI-assisted outputs can be incomplete or wrong. Important use cases need approved sources, visible uncertainty, evaluation and a human review or escalation path.
Can Maven send messages or change records automatically?
Only a specifically designed implementation could take a permitted action. Material external actions should be separated from drafting and protected by appropriate confirmation, permissions and records.
When is Maven not the right solution?
It may be unsuitable when source information is untrustworthy, permissions cannot be enforced, the decision is too consequential for the available controls or a simpler deterministic interface would work better.
Design the boundary first
Bring one real assistance task and the sources it should trust.
We can assess whether Maven, another assistant pattern or a simpler interface is the responsible way to help the user reach the target.