Room to grow.

Companies, contacts, and a pipeline that lives in one place.

• Companies, contacts, and deals on one timeline, with an owner and a stage on every record
• A pipeline you configure: your stages, your qualification fields, your close reasons
• Email on the record through configured SMTP and IMAP, so sent messages and captured replies stay with the customer
• Notes, activity history, and follow-up tasks against the customer instead of someone's inbox
• Accounts, contacts, leads, deals, meetings, approvals, contracts, renewals, sequences and campaign records

SMTP and IMAP require configuration. WhatsApp is not implemented. HubSpot and Shopify adapters exist, but live provider operation is not yet a public promise.

Dated customer engagements with a controlled commercial state.

• CRM-linked bookings with inquiry, confirmed, completed, and cancelled states
• Fees, advances, balances, refunds, and block dates held on the booking
• Availability checks with refusal when the requested period is already booked
• The booking stays attached to the customer record and its commercial history

Bookings does not yet provide a service catalogue, staff qualification or assignment, shift-based capacity, a time-slot engine, or a no-show state.

A real double-entry ledger, not a folder of invoices.

• Double-entry ledger with a chart of accounts you control
• Customer invoices, credit notes, receipts, and part-payments against an open balance
• Purchase order to goods receipt to supplier bill, with a three-way match
• GSTR-1 and GSTR-3B summaries plus a Tally-compatible export
• Bank reconciliation against an imported statement
• Recurring billing and automated dunning on overdue accounts

Pharos prepares the numbers; you or your CA file them. There is no direct GSTN connection and no live bank feed.

Products, variants, price lists, and the identifiers the counter uses.

• Products with variants, SKU, barcode, units, media, options, and attributes
• List, MRP, and cost price held separately on the sellable record
• Tax-code links keep HSN or SAC treatment with the tax definition
• Customer-group price lists and product kits

Price lists do not yet have store or channel scope or effective-date scheduling.

What is on the shelf, at which location, right now.

• Movement-derived on-hand quantity by retail location
• Transfers with separate dispatch and receipt states
• Batch, expiry, FEFO, and serial-unit tracking
• Stock takes with counted-versus-system variance
• Moving-average or FIFO valuation and reorder analytics

There is no explicit reserved or available-to-sell field, per-location reorder setting, or supplier-return workflow.

The register covers the cart, tender, receipt, and the ledger entry behind it.

• Touch-first counter screen: scan or search, edit the cart, apply discounts, hold a bill
• Split tender across cash, card, and UPI, with change and rounding handled
• Returns referenced to the original bill
• Every sale records the stock and tax consequences, then shows whether the accounting entry posted or needs recovery
• Shift open and close with a cash-drawer count and a variance somebody explains

There is no atomic exchange record. Payment-terminal and device-provider integrations are not claimed as verified.

The retail customer file, available at the counter.

• Customer details, GSTIN, credit limit, phone lookup, and price-list assignment
• Purchase history, return history, and payment-derived khata
• Deterministic RFM segmentation from recorded transactions
• A link to the CRM customer context

Phone and email are not globally unique, and the module does not yet provide contact-preference or consent management.

GST computed on the line, not bolted on at the end.

• GST tax codes with HSN or SAC and CGST, SGST, and IGST rates
• Line-level tax calculation stored on the invoice or bill
• Interstate treatment supplied explicitly for the transaction
• Period tax summaries that feed the GST returns prepared in Finance

Place of supply is not universally inferred, and price lists do not yet carry an inclusive or exclusive tax setting.

Points and tiers that settle correctly at the till.

• Points earned per rupee or per product, on rules you set
• Programmes, tiers, multipliers, redemption bonuses, and corrections
• Redemption at the Pharos till
• A derived points ledger with balances and expiry

Loyalty is native to Pharos and its point of sale; external omnichannel synchronisation is not claimed.

Offers a cashier cannot get wrong.

• Percentage, amount, threshold, and buy-X-get-Y offers with date windows
• Variant, category, and customer-segment scope
• Preview and application at the register
• Priority, stacking rules, and coupons

Promotion rules do not yet have store or channel scope.

Getting the order out of the door with a custody trail.

• Handover to a courier or aggregator recorded as a custody event, with a timestamp and a named actor
• Daily manifests with sign-off
• Custody status and silence-window exceptions
• Adapter refresh and a Shopify handover event where configured

There is no pick-and-pack workflow or return-to-stock flow. Courier and Shopify providers are not yet claimed as live-verified integrations.

General materials, shelves, and an append-only movement record.

• Items, units, shelves, lots, expiry, and FEFO controls
• Opening, receipt, issue, count, and adjustment movements
• Append-only evidence with reasoned reversals and actor history
• Unit cost, derived on-hand, reorder visibility, and a retail crosswalk

General Inventory does not yet provide native cross-location transfers, serial tracking, or a stock-take session document. Retail Stock remains a distinct ledger.

Works orders against real bills of material.

• Bills of material with component quantities and material-cost estimates
• Work orders and operations with planned production
• Material issue or backflush from Inventory
• Good and scrap quantity reported against the work order
• Quality gates and conditional finished-output movement

Cost is material-only, not full actual production cost. Labour, conversion cost, OEE, and operation-level yield are not included. Output posting can be blocked after a report and remains visible for recovery.

The asset register the plant floor actually uses.

• Assets, lines, and tools in the plant registry
• Declared machine status with actor and reason
• Archive and restore with history preserved
• An open-downtime indicator when Maintenance has an active event

Meter readings and the downtime lifecycle belong to Maintenance. A declared machine status is not the same as observed downtime.

An inspection gate, not a form somebody files.

• Work-order and asset quality checks
• Manual or measured characteristics with minimum, maximum, target, and tolerance
• Derived pass or fail outcomes, with controlled waivers
• Required-check completion gates and completed-check first-pass yield

There are no reusable inspection-plan templates, sampling or batch inspection, NCR disposition workflow, or per-unit defect rate.

Planned upkeep that competes for the same hours as production.

• Preventive schedules by calendar or by meter reading
• Six-state maintenance work orders
• Meter readings, downtime events, and derived availability
• Parts issued from Inventory against the work order
• Breakdown, repair, and closure history against the machine

There is no OEE, MTBF or MTTR calculation, labour-hours ledger, full repair cost, asset hierarchy, failure-code library, or warranty workflow.

The org: who works here, in which team, under whom.

• Employee records with business job role, business unit, department, team, manager, and location
• Employment status, joining, and termination details
• A login bridge from the employee to their workspace identity
• Competency and readiness data used by workforce workflows

There is no position or grade model and no dedicated dated history for organisation assignments. Business job roles are not security roles.

Versioned rosters checked before they are published.

• Shift patterns and templates per site and per role
• Schedule versions, diffs, publishing, locking, and late amendments
• Compliance preflight for overlap, leave, rest, consecutive days, and competency
• Coverage warnings and governed overrides before publication
• Notifications through a configured delivery path

Employee self-service availability is not yet consumed by scheduling, and there is no roster-acknowledgement model. Shift swaps are handled in Marketplace.

Hours that survive an argument.

• Clock and break events with source, location code, time, and actor
• Late, early, and absent derived from the roster rather than judged by hand
• Daily overtime and break-rule derivation
• Regularisation requests with an approval trail
• The paid-days feed payroll actually runs on

GPS or device attestation is not verified, weekly overtime is not fully applied, and manual paid-days overrides remain possible and visible.

Balances that are computed, not remembered.

• Leave types and policies with monthly accrual, rollover, and carry-expiry rules
• Balances accrued automatically and visible to the employee
• Apply, approve, and cancel with the roster and attendance kept in step
• Holiday calendars per location

There is no encashment calculation or payment workflow. Attendance synchronisation can be blocked by a locked period.

Open shifts the right people can pick up.

• Publish unfilled shifts to the workers qualified and available for them
• Claims, offers, withdrawals, swaps, bids, auctions, and gigs
• Eligibility and rule-based recommendations using skills and readiness
• A history of who was offered a shift and who took it

Matching is deterministic and rule-based, not AI. Readiness depends on Team data.

Demand-driven staffing instead of last month's roster.

• History import, patterns, assumptions, scenarios, versions, approval, and locking
• Required hours translated into the roles and headcount to schedule
• Moving-average or seasonal-naive forecasting with held-out algorithm selection
• Planned-versus-actual comparison and an explicit feed-plan action into Scheduling

Forecasting is statistical and deterministic, not generative AI. Insufficient history remains unknown or uses a disclosed pattern replay.

Statutory Indian payroll driven by attendance you can audit.

• Earnings and deductions structured per employee, with a pay register per cycle
• Paid days derived from real attendance and approved leave, not typed in at month end
• Confirmed EPF, ESI, and professional-tax tables applied to the cycle
• Approval, finance posting, and settlement states
• Payslips per employee for the period

Monthly TDS and ECR are not supported. There is no Form 16, 24Q, gratuity, bonus, LWF, reimbursements, loans, advances, arrears, or external payroll connector.

Employee documents with bytes, versions, checksums, and expiry evidence.

• Real file bytes stored against the employee document record
• Checksums and downloads for document evidence
• Expiry dates, versions, and uploader history
• Archive and restore without erasing the record

This is not a full HRIS. Employee master data belongs to Team; broad profile self-service, onboarding, e-signature, absence, and case management are not included here.

The hiring process, in the system where the hire will exist.

• Requisitions that need an approval before a role can be opened
• Pipelines per role, with candidates moving through your stages
• Interviews with panelists and structured scorecards
• Structured scorecards, so a decision has evidence behind it
• Offers that create the employee record on acceptance: nobody re-types the hire

A hiring-process tool, not a sourcing suite: no candidate merge, job-board posting, CV parsing, assessments, talent pool, external calendar or email integration, interviewer-availability check, or offer-letter generation.

Training assigned, completed, and provable.

• Courses and modules with employee enrolments
• Admin-recorded completion, progress, overdue state, and scores
• Competency awarded or revoked from recorded completion
• Training history attached to the employee record

There is no hosted learning content, quiz engine, learner self-service path, role or team bulk assignment, or expiry-driven automatic enrolment. Scores are entered by an administrator.

Client work with hours, cost, and a bill at the end.

• Projects with phases, tasks, owners, and due dates
• Time and expenses booked to project tasks
• Billable and non-billable hours separated at entry, not argued about later
• Budget versus actual per project, in hours and in rupees
• A human invoice action creates the Finance invoice from project evidence

Time entries do not have an approval state. There are no retainers, advances, labour tax, or employee-role rate cards.

Tickets with an owner, a clock, and a resolution.

• Tickets raised by email or by an agent, against a customer record
• Queues, assignment, and priority you configure
• SLA clocks for response and resolution, with a breach you can see coming
• Configured IMAP email-to-ticket with threading and attachments
• Resolution, reopen, and closure history per ticket

A requester-visible product comment does not itself send outbound email. This is not a broad portal, CMDB, IT asset, or field-service system.

The first screen: what needs you today.

• A Mission Control landing screen shaped by role and enabled module packs
• Employee self-service entry points backed by live workspace APIs
• Shortcuts into the modules this workspace has actually turned on

Executive indicators appear only when the underlying workspace data exists; example figures are never presented as measured business data.

Control who can see what, with proof of who did what.

• Users, roles, and permissions, denied by default until granted
• Module packs switched on or off per workspace
• An audit trail of who changed what and when
• Every record scoped to your organisation and enforced on the query, not just the screen

Selected business events routed into visible actions and failures.

• A defined allowlist of business-event triggers
• Recorded workflow runs with their current state
• Email or pending-notification actions where delivery is configured
• Task, activity, and ticket actions with failed, unavailable, and pending-human states kept visible

There are no arbitrary conditions, branches, record mutations, approvals, or money movement. Outbound webhook delivery is not implemented and reports unavailable.

One set of numbers, drawn from the tables everyone works in.

• Dashboards, widgets, KPIs, trends, and charts
• Report templates, saved views, schedules, email delivery, and exports
• Drilldown on supported reports, with filters and source detail
• Retail, purchasing, benchmark, and executive analytics surfaces

Drill-through and real-time behaviour depend on the report; universal cross-module comparison is not claimed.

Files that live on the record they belong to.

• Upload, download, checksum, ownership, folders, and tags
• Versions with restore, expiry, retention, disposal, and legal hold
• Access logs and full-text BM25F search
• PDF text layers, image thumbnails, EXIF, and a knowledge base

There is no OCR for scanned PDFs, PDF page preview, e-signature, checkout, document approval, or bulk disposal. Folder, tag, and owner controls are narrower than the storage engine.

Companies, contacts, and a pipeline that lives in one place.

• Companies, contacts, and deals on one timeline, with an owner and a stage on every record
• A pipeline you configure: your stages, your qualification fields, your close reasons
• Email on the record through configured SMTP and IMAP, so sent messages and captured replies stay with the customer
• Notes, activity history, and follow-up tasks against the customer instead of someone's inbox
• Accounts, contacts, leads, deals, meetings, approvals, contracts, renewals, sequences and campaign records

SMTP and IMAP require configuration. WhatsApp is not implemented. HubSpot and Shopify adapters exist, but live provider operation is not yet a public promise.

Dated customer engagements with a controlled commercial state.

• CRM-linked bookings with inquiry, confirmed, completed, and cancelled states
• Fees, advances, balances, refunds, and block dates held on the booking
• Availability checks with refusal when the requested period is already booked
• The booking stays attached to the customer record and its commercial history

Bookings does not yet provide a service catalogue, staff qualification or assignment, shift-based capacity, a time-slot engine, or a no-show state.

A real double-entry ledger, not a folder of invoices.

• Double-entry ledger with a chart of accounts you control
• Customer invoices, credit notes, receipts, and part-payments against an open balance
• Purchase order to goods receipt to supplier bill, with a three-way match
• GSTR-1 and GSTR-3B summaries plus a Tally-compatible export
• Bank reconciliation against an imported statement
• Recurring billing and automated dunning on overdue accounts

Pharos prepares the numbers; you or your CA file them. There is no direct GSTN connection and no live bank feed.

Products, variants, price lists, and the identifiers the counter uses.

• Products with variants, SKU, barcode, units, media, options, and attributes
• List, MRP, and cost price held separately on the sellable record
• Tax-code links keep HSN or SAC treatment with the tax definition
• Customer-group price lists and product kits

Price lists do not yet have store or channel scope or effective-date scheduling.

What is on the shelf, at which location, right now.

• Movement-derived on-hand quantity by retail location
• Transfers with separate dispatch and receipt states
• Batch, expiry, FEFO, and serial-unit tracking
• Stock takes with counted-versus-system variance
• Moving-average or FIFO valuation and reorder analytics

There is no explicit reserved or available-to-sell field, per-location reorder setting, or supplier-return workflow.

The register covers the cart, tender, receipt, and the ledger entry behind it.

• Touch-first counter screen: scan or search, edit the cart, apply discounts, hold a bill
• Split tender across cash, card, and UPI, with change and rounding handled
• Returns referenced to the original bill
• Every sale records the stock and tax consequences, then shows whether the accounting entry posted or needs recovery
• Shift open and close with a cash-drawer count and a variance somebody explains

There is no atomic exchange record. Payment-terminal and device-provider integrations are not claimed as verified.

The retail customer file, available at the counter.

• Customer details, GSTIN, credit limit, phone lookup, and price-list assignment
• Purchase history, return history, and payment-derived khata
• Deterministic RFM segmentation from recorded transactions
• A link to the CRM customer context

Phone and email are not globally unique, and the module does not yet provide contact-preference or consent management.

GST computed on the line, not bolted on at the end.

• GST tax codes with HSN or SAC and CGST, SGST, and IGST rates
• Line-level tax calculation stored on the invoice or bill
• Interstate treatment supplied explicitly for the transaction
• Period tax summaries that feed the GST returns prepared in Finance

Place of supply is not universally inferred, and price lists do not yet carry an inclusive or exclusive tax setting.

Points and tiers that settle correctly at the till.

• Points earned per rupee or per product, on rules you set
• Programmes, tiers, multipliers, redemption bonuses, and corrections
• Redemption at the Pharos till
• A derived points ledger with balances and expiry

Loyalty is native to Pharos and its point of sale; external omnichannel synchronisation is not claimed.

Offers a cashier cannot get wrong.

• Percentage, amount, threshold, and buy-X-get-Y offers with date windows
• Variant, category, and customer-segment scope
• Preview and application at the register
• Priority, stacking rules, and coupons

Promotion rules do not yet have store or channel scope.

Getting the order out of the door with a custody trail.

• Handover to a courier or aggregator recorded as a custody event, with a timestamp and a named actor
• Daily manifests with sign-off
• Custody status and silence-window exceptions
• Adapter refresh and a Shopify handover event where configured

There is no pick-and-pack workflow or return-to-stock flow. Courier and Shopify providers are not yet claimed as live-verified integrations.

General materials, shelves, and an append-only movement record.

• Items, units, shelves, lots, expiry, and FEFO controls
• Opening, receipt, issue, count, and adjustment movements
• Append-only evidence with reasoned reversals and actor history
• Unit cost, derived on-hand, reorder visibility, and a retail crosswalk

General Inventory does not yet provide native cross-location transfers, serial tracking, or a stock-take session document. Retail Stock remains a distinct ledger.

Works orders against real bills of material.

• Bills of material with component quantities and material-cost estimates
• Work orders and operations with planned production
• Material issue or backflush from Inventory
• Good and scrap quantity reported against the work order
• Quality gates and conditional finished-output movement

Cost is material-only, not full actual production cost. Labour, conversion cost, OEE, and operation-level yield are not included. Output posting can be blocked after a report and remains visible for recovery.

The asset register the plant floor actually uses.

• Assets, lines, and tools in the plant registry
• Declared machine status with actor and reason
• Archive and restore with history preserved
• An open-downtime indicator when Maintenance has an active event

Meter readings and the downtime lifecycle belong to Maintenance. A declared machine status is not the same as observed downtime.

An inspection gate, not a form somebody files.

• Work-order and asset quality checks
• Manual or measured characteristics with minimum, maximum, target, and tolerance
• Derived pass or fail outcomes, with controlled waivers
• Required-check completion gates and completed-check first-pass yield

There are no reusable inspection-plan templates, sampling or batch inspection, NCR disposition workflow, or per-unit defect rate.

Planned upkeep that competes for the same hours as production.

• Preventive schedules by calendar or by meter reading
• Six-state maintenance work orders
• Meter readings, downtime events, and derived availability
• Parts issued from Inventory against the work order
• Breakdown, repair, and closure history against the machine

There is no OEE, MTBF or MTTR calculation, labour-hours ledger, full repair cost, asset hierarchy, failure-code library, or warranty workflow.

The org: who works here, in which team, under whom.

• Employee records with business job role, business unit, department, team, manager, and location
• Employment status, joining, and termination details
• A login bridge from the employee to their workspace identity
• Competency and readiness data used by workforce workflows

There is no position or grade model and no dedicated dated history for organisation assignments. Business job roles are not security roles.

Versioned rosters checked before they are published.

• Shift patterns and templates per site and per role
• Schedule versions, diffs, publishing, locking, and late amendments
• Compliance preflight for overlap, leave, rest, consecutive days, and competency
• Coverage warnings and governed overrides before publication
• Notifications through a configured delivery path

Employee self-service availability is not yet consumed by scheduling, and there is no roster-acknowledgement model. Shift swaps are handled in Marketplace.

Hours that survive an argument.

• Clock and break events with source, location code, time, and actor
• Late, early, and absent derived from the roster rather than judged by hand
• Daily overtime and break-rule derivation
• Regularisation requests with an approval trail
• The paid-days feed payroll actually runs on

GPS or device attestation is not verified, weekly overtime is not fully applied, and manual paid-days overrides remain possible and visible.

Balances that are computed, not remembered.

• Leave types and policies with monthly accrual, rollover, and carry-expiry rules
• Balances accrued automatically and visible to the employee
• Apply, approve, and cancel with the roster and attendance kept in step
• Holiday calendars per location

There is no encashment calculation or payment workflow. Attendance synchronisation can be blocked by a locked period.

Open shifts the right people can pick up.

• Publish unfilled shifts to the workers qualified and available for them
• Claims, offers, withdrawals, swaps, bids, auctions, and gigs
• Eligibility and rule-based recommendations using skills and readiness
• A history of who was offered a shift and who took it

Matching is deterministic and rule-based, not AI. Readiness depends on Team data.

Demand-driven staffing instead of last month's roster.

• History import, patterns, assumptions, scenarios, versions, approval, and locking
• Required hours translated into the roles and headcount to schedule
• Moving-average or seasonal-naive forecasting with held-out algorithm selection
• Planned-versus-actual comparison and an explicit feed-plan action into Scheduling

Forecasting is statistical and deterministic, not generative AI. Insufficient history remains unknown or uses a disclosed pattern replay.

Statutory Indian payroll driven by attendance you can audit.

• Earnings and deductions structured per employee, with a pay register per cycle
• Paid days derived from real attendance and approved leave, not typed in at month end
• Confirmed EPF, ESI, and professional-tax tables applied to the cycle
• Approval, finance posting, and settlement states
• Payslips per employee for the period

Monthly TDS and ECR are not supported. There is no Form 16, 24Q, gratuity, bonus, LWF, reimbursements, loans, advances, arrears, or external payroll connector.

Employee documents with bytes, versions, checksums, and expiry evidence.

• Real file bytes stored against the employee document record
• Checksums and downloads for document evidence
• Expiry dates, versions, and uploader history
• Archive and restore without erasing the record

This is not a full HRIS. Employee master data belongs to Team; broad profile self-service, onboarding, e-signature, absence, and case management are not included here.

The hiring process, in the system where the hire will exist.

• Requisitions that need an approval before a role can be opened
• Pipelines per role, with candidates moving through your stages
• Interviews with panelists and structured scorecards
• Structured scorecards, so a decision has evidence behind it
• Offers that create the employee record on acceptance: nobody re-types the hire

A hiring-process tool, not a sourcing suite: no candidate merge, job-board posting, CV parsing, assessments, talent pool, external calendar or email integration, interviewer-availability check, or offer-letter generation.

Training assigned, completed, and provable.

• Courses and modules with employee enrolments
• Admin-recorded completion, progress, overdue state, and scores
• Competency awarded or revoked from recorded completion
• Training history attached to the employee record

There is no hosted learning content, quiz engine, learner self-service path, role or team bulk assignment, or expiry-driven automatic enrolment. Scores are entered by an administrator.

Client work with hours, cost, and a bill at the end.

• Projects with phases, tasks, owners, and due dates
• Time and expenses booked to project tasks
• Billable and non-billable hours separated at entry, not argued about later
• Budget versus actual per project, in hours and in rupees
• A human invoice action creates the Finance invoice from project evidence

Time entries do not have an approval state. There are no retainers, advances, labour tax, or employee-role rate cards.

Tickets with an owner, a clock, and a resolution.

• Tickets raised by email or by an agent, against a customer record
• Queues, assignment, and priority you configure
• SLA clocks for response and resolution, with a breach you can see coming
• Configured IMAP email-to-ticket with threading and attachments
• Resolution, reopen, and closure history per ticket

A requester-visible product comment does not itself send outbound email. This is not a broad portal, CMDB, IT asset, or field-service system.

The first screen: what needs you today.

• A Mission Control landing screen shaped by role and enabled module packs
• Employee self-service entry points backed by live workspace APIs
• Shortcuts into the modules this workspace has actually turned on

Executive indicators appear only when the underlying workspace data exists; example figures are never presented as measured business data.

Control who can see what, with proof of who did what.

• Users, roles, and permissions, denied by default until granted
• Module packs switched on or off per workspace
• An audit trail of who changed what and when
• Every record scoped to your organisation and enforced on the query, not just the screen

Selected business events routed into visible actions and failures.

• A defined allowlist of business-event triggers
• Recorded workflow runs with their current state
• Email or pending-notification actions where delivery is configured
• Task, activity, and ticket actions with failed, unavailable, and pending-human states kept visible

There are no arbitrary conditions, branches, record mutations, approvals, or money movement. Outbound webhook delivery is not implemented and reports unavailable.

One set of numbers, drawn from the tables everyone works in.

• Dashboards, widgets, KPIs, trends, and charts
• Report templates, saved views, schedules, email delivery, and exports
• Drilldown on supported reports, with filters and source detail
• Retail, purchasing, benchmark, and executive analytics surfaces

Drill-through and real-time behaviour depend on the report; universal cross-module comparison is not claimed.

Files that live on the record they belong to.

• Upload, download, checksum, ownership, folders, and tags
• Versions with restore, expiry, retention, disposal, and legal hold
• Access logs and full-text BM25F search
• PDF text layers, image thumbnails, EXIF, and a knowledge base

There is no OCR for scanned PDFs, PDF page preview, e-signature, checkout, document approval, or bulk disposal. Folder, tag, and owner controls are narrower than the storage engine.