Thirteen roles, four real Dynamics 365 scenarios each. For every one: how the person actually asks, the governed D365 action behind it, and how the assistant sharpens to that individual over time.
Every role also gets its D365 workflow approval queue on the phone — the approval work items D365 already routes to them (requisitions, invoices, expenses, journals, project approvals): approve, reject, request changes, or delegate by voice, online or offline, always audited.
cross-module read across Finance, Sales, and Supply Chain; high-value approvals.
“What changed overnight, and what needs me today?”
Nāda assembles one narrative from Finance (cash position, P&L vs plan), Sales (pipeline and forecast movement), and Supply Chain (OTIF, major order or production exceptions) and reads back the three or four things that actually moved — with the decisions waiting on the CEO surfaced at the top.
Gets better: learns which KPIs the CEO opens first, the segments and business units they care about, and how terse or detailed they like the brief — so over weeks the briefing reorders itself to match the CEO's attention rather than a fixed template.
“Build my board narrative for the quarter and the questions they'll ask.”
Nāda drafts the quarterly story from D365 actuals (Finance, Sales, operations) with drill-down evidence behind each claim, plus a list of anticipated board questions and the records that answer them.
Gets better: learns the board's recurring questions and the CEO's narrative voice, so each quarter the draft needs fewer edits and pre-loads the evidence the board always probes.
“Approve the Riverside capex but hold the discount exception until I see margin.”
Major capex, large-deal exceptions, and strategic spend route to the CEO; decisions are validated against Finance/Sales business logic and committed (or queued offline and reconciled on reconnect) with a full audit trail.
Gets better: learns the CEO's standing thresholds — what they always wave through versus always want context on — and pre-attaches the margin or cash detail they ask for every time.
“Approve the capex requisition for Riverside and delegate the rest to the CFO while I'm out.”
The big-ticket items that escalate to the CEO in D365 workflow — high-value purchase requisitions, capex and budget register entries, major contracts — arrive as approval work items; Nāda reads the request, the policy that routed it, and the supporting detail, and lets the CEO approve, reject, request changes, or delegate with a spoken comment, queued offline and committed with audit on reconnect.
Gets better: learns which requests the CEO delegates versus keeps, the comment they usually attach, and the thresholds that warrant a closer look.
Finance — AP, AR, GL, payment runs, period close.
“Approve the payment run, but pull the two invoices over fifty thousand for me to check.”
Vendor invoices, payment proposals, and journal entries route to the CFO; each is validated against Finance controls and either committed or held, online or offline, with audit on reconnect.
Gets better: learns which vendors and amounts the CFO fast-tracks versus inspects, and starts pre-flagging the exceptions that match the CFO's own scrutiny pattern.
“Where's our cash and who's overdue?”
Nāda reads AR aging and collections status, names the top overdue accounts and the cash on hand, and offers to send a chase or escalate.
Gets better: learns the customers the CFO watches, their preferred aging buckets, and the threshold at which a receivable is worth mentioning — tightening the briefing to the accounts that actually matter to this CFO.
“What's blocking close, and where are we off plan?”
During month-end Nāda reports subledger reconciliation status, open accruals, and GL variances against plan, with commentary only on the lines that breach the CFO's materiality threshold.
Gets better: learns the recurring close blockers and the CFO's variance threshold for narrative, so it surfaces the same trouble spots earlier each period and drafts the variance commentary in the CFO's style.
“Clear the vendor invoices under policy and hold the two that breach budget.”
Vendor invoice, payment journal, AP/AR journal, and expense-report approvals route to the CFO through D365 workflow; Nāda batches the work items, shows policy and budget context per line, and approves, rejects, or requests changes with comments — online or offline.
Gets better: learns which vendors and amounts the CFO clears on sight, the common reasons they reject, and the budget lines they always check.
Supply Chain (production, logistics) and Field Service operations across sites.
“How are operations running this morning?”
Nāda consolidates production order status and downtime (Supply Chain), service SLA performance (Field Service), and logistics exceptions into a single read across every plant and region.
Gets better: learns which sites and metrics the COO prioritizes and which they ignore, reordering the briefing toward the COO's real hot spots.
“A line's down at Plant 2 — what are my options?”
Nāda surfaces a production stoppage, late shipment, or SLA breach with the affected orders and proposes the response (authorize overtime, expedite freight, re-route) for one-tap action.
Gets better: learns which exceptions the COO wants pushed immediately versus batched, and which response the COO usually chooses for each failure type.
“Authorize the overtime for the weekend run.”
Capacity changes, expedite-freight authorizations, and overtime route to the COO, validated against the relevant Supply Chain and Finance logic and audited.
Gets better: learns the COO's approval thresholds by site and cost, pre-staging the operational and cost context the COO weighs before saying yes.
“Approve the PO for the line spares and the overtime, reject the rush-freight on order 5512.”
Purchase orders, capacity changes, and overtime that escalate operationally arrive as workflow items; Nāda presents each with its operational and cost impact for one-pass approve, reject, or delegate.
Gets better: learns the COO's standing approvals by site and cost band and pre-sorts the queue into routine and needs-you.
Sales — pipeline, forecast, deal desk; Finance credit for exceptions.
“Where's the quarter, and what slipped?”
Nāda reads pipeline by stage, forecast versus quota, and deals that moved out, naming the segments and reps driving the gap.
Gets better: learns whether the CRO thinks in commit or best-case, which segments and reps they drill into, and the forecast view they trust — and leads with that each time.
“Approve the discount on the Contoso deal but not the payment terms.”
Discount approvals, non-standard terms, and credit exceptions route to the CRO with margin and credit context (Sales + Finance), committed or queued offline.
Gets better: learns the CRO's discount tolerance by segment and deal size, pre-flagging the requests that sit outside the CRO's usual envelope.
“Which big deals are going quiet?”
Nāda flags stalled opportunities, single-threaded deals, and deals with no next step, with the history behind each.
Gets better: learns which risk signals the CRO actually acts on versus dismisses, and stops surfacing the noise the CRO never touches.
“Approve Contoso's discount, send the non-standard terms back for margin detail.”
Discount, pricing, and credit-override requests submitted by reps route to the CRO; Nāda shows margin and credit context and approves, rejects, or requests changes with a comment, syncing to Sales.
Gets better: learns the CRO's discount envelope by segment and the detail they always ask reps to add before approving.
the audit/evidence layer, adoption and data-quality telemetry, identity and access.
“Show me every AI-initiated action this week and who approved it.”
Nāda queries the action-packet evidence log — source context, the human who confirmed, validation status, timestamp, and provenance — for any role, app, or date range.
Gets better: learns which controls and risk categories the CIO reviews and assembles the same evidence cut automatically ahead of the CIO's review cadence.
“Where is adoption lagging and which harness needs tuning?”
Nāda reports usage by role and site, where users override or abandon drafts (a signal the harness needs work), and emerging D365 data-quality issues.
Gets better: learns the CIO's adoption KPIs and the thresholds that mean "intervene," surfacing the at-risk roles before the CIO has to ask.
“Who can approve payments, and was anything unusual this week?”
Nāda reports effective access (via Entra roles) for sensitive duties and flags anomalous actions against the evidence log.
Gets better: learns the CIO's review rhythm and the duties they treat as sensitive, pre-building the access cut for each periodic review.
“Approve the access request for the new controller, deny the standing export rights.”
Access requests, privileged-role grants, and IT or project spend that route for approval arrive in Nāda with the requester, the entitlement, and the risk; the CIO approves, denies, or time-boxes with a comment, all logged to the evidence layer.
Gets better: learns which entitlements the CIO grants routinely versus scrutinizes and pre-flags the requests that break the usual pattern.
Supply Chain — procurement, vendor collaboration, sourcing.
“How are our key suppliers performing?”
Nāda reads procurement spend by category and supplier on-time/in-full and quality scores (vendor collaboration), naming the suppliers trending the wrong way.
Gets better: learns the categories and strategic suppliers the CPO watches and the performance thresholds worth flagging, narrowing the briefing to the CPO's real risk list.
“Approve the PO and the new vendor, hold the price increase.”
Purchase orders above threshold, price/contract changes, and new-vendor onboarding route to the CPO with spend and risk context, committed or queued offline.
Gets better: learns the CPO's approval patterns by category and value and pre-attaches the contract or benchmark context the CPO reviews before approving.
“What's my supply risk this month?”
Nāda surfaces single-source items, expiring contracts, and suppliers with deteriorating delivery performance, with the affected items and spend.
Gets better: learns which risks the CPO prioritizes — continuity, price, or compliance — and leads the watch list with the category the CPO acts on most.
“Approve the requisition and the PO to Acme, hold the new-vendor onboarding for compliance.”
Purchase requisitions, purchase orders, purchase agreements, and vendor onboarding route to the CPO through D365 workflow; Nāda shows spend, contract, and risk context for approve, reject, request-change, or delegate.
Gets better: learns the CPO's approval patterns by category and value and the compliance checks they want attached to new-vendor approvals.
Field Service — work orders, bookings, parts, asset history. Built for weak connectivity.
“Log two hours on this work order, mark it resolved, and add the parts I used.”
On site and often with no signal, the technician updates the Field Service work order — time, status, parts, and notes — by voice; Nāda drafts the entries locally and writes them back, validated and audited, on reconnect.
Gets better: learns the technician's common parts, the way they phrase resolution notes, and the task patterns for each job type — so the draft is nearly complete before they edit a word.
“Do I have the right pump on the truck, or do I need a transfer?”
Nāda checks truck stock and warehouse availability (Field Service inventory / Supply Chain) and drafts a transfer or order request.
Gets better: learns which parts this technician uses per symptom and pre-checks the likely parts for the job before they ask.
“What's the history on this unit and what usually fixes this fault?”
Nāda pulls the asset's service history and prior resolutions and suggests the next action, with guides if available.
Gets better: learns this technician's preferred fixes per symptom and starts leading with the resolution they trust, not a generic list.
“Submit my timesheet and request approval for the warranty part.”
The technician is usually the requester, not the approver: Nāda lets them submit time, expenses, and parts or return requests into the right D365 workflow by voice, and tracks where each one sits.
Gets better: learns the technician's common submissions and pre-fills them, and learns to nudge when an approval is stuck so work isn't blocked.
Field Service scheduling and Customer Service cases.
“Who can take an emergency call in the north zone this afternoon?”
Nāda reads the Field Service schedule board, proposes the best available resource by skill and location, and reassigns on confirmation.
Gets better: learns the manager's dispatch preferences — which technicians they trust for which work, how they trade off travel versus skill — and proposes the assignment the manager usually makes.
“What's about to breach SLA?”
Nāda lists cases and work orders nearing SLA, ordered by account priority, and offers to escalate or reassign.
Gets better: learns the accounts the manager protects and the lead time they like before a breach, surfacing the right cases earlier.
“Approve the warranty claim and order the parts.”
Warranty claims, parts orders, and customer credits route to the manager with the work-order context, validated and audited.
Gets better: learns the manager's approval thresholds and the claims they always question, pre-flagging the exceptions.
“Approve the techs' timesheets, the warranty claim on case 412, and the parts return.”
Work-order, time, warranty/RMA, and parts-return approvals (Field Service and Power Automate) route to the manager; Nāda batches them with the job context for fast approve, reject, or request-change.
Gets better: learns whose time the manager rubber-stamps, the claims they question, and the return reasons they reject.
Supply Chain — production control, shop floor, quality.
“How's line 3 running and what's my scrap today?”
Nāda reads production order progress, WIP, downtime, and scrap for the lines the manager owns and reads back the exceptions.
Gets better: learns which lines and KPIs the manager watches and the shift rhythm, leading the read with the line most likely to be in trouble.
“Log a fifteen-minute stoppage on line 2 for a die change, and flag a quality hold on the last lot.”
The manager records a downtime reason or quality issue by voice from the floor; Nāda drafts the production journal and quality entries and writes them back on reconnect.
Gets better: learns the manager's common downtime and reject codes and the way they describe issues, turning a spoken phrase into the right coded entry.
“Approve the overtime and the material substitution for the rush order.”
Schedule changes, overtime, material substitutions, and quality dispositions route to the manager, validated against Supply Chain logic and audited.
Gets better: learns the manager's thresholds for substitution and overtime and pre-attaches the order and cost context they weigh.
“Approve the material requisition and the overtime, send the substitution back with a note.”
Material requisitions, production order releases, overtime, and quality dispositions route to the plant manager; Nāda shows order, cost, and quality context for one-pass decisions, offline-capable on the floor.
Gets better: learns the manager's substitution and overtime thresholds and the dispositions they always escalate to quality.
Supply Chain — warehouse management, inventory, shipments.
“What's still to ship today and what's stuck at the dock?”
Nāda reads open shipments, picks, putaway, and dock status (Warehouse Management) and names the at-risk orders.
Gets better: learns the supervisor's priorities — carrier cut-offs, key customers — and leads with the orders that matter to this shift.
“Count bin A-12 — I've got forty units, system says forty-three.”
The supervisor records counts and discrepancies by voice; Nāda drafts the inventory adjustment with reason code and writes it back, audited, on reconnect.
Gets better: learns the bins and items that drift and the supervisor's common adjustment reasons, pre-staging the count list and the right reason code.
“Reassign the night picks and approve the short-pick on order 8842.”
Work reassignment, short-pick approvals, and expedite requests route to the supervisor, validated and audited.
Gets better: learns the supervisor's reassignment patterns and which short-picks they accept, proposing the move they'd usually make.
“Approve the cycle-count adjustments under five units, escalate the big variance on bin C-04.”
Inventory adjustments, transfer orders, and short-pick exceptions route to the supervisor; Nāda groups them with reason codes and variance for approve or reject, audited.
Gets better: learns the variance the supervisor accepts without a look and the bins or items that always warrant escalation.
Supply Chain — master planning, purchase orders, vendor collaboration.
“What planned orders need firming this week, and any reschedules?”
Nāda reads master planning output — planned orders and action messages (reschedule in/out, expedite) — and walks the planner through the exceptions, firming on confirmation.
Gets better: learns which items and vendors the planner firms without a second thought and which they always review, pre-sorting the action messages into "likely firm" and "needs you."
“Confirm the open POs with Acme and expedite the casting line.”
Nāda creates and confirms POs, chases unconfirmed orders, and drafts expedite requests (vendor collaboration).
Gets better: learns the planner's preferred vendors, typical lead-time overrides, and expedite habits, drafting the PO the way the planner usually places it.
“What's going to stock out before the inbound lands?”
Nāda flags stockout risk, late inbound, and safety-stock breaches against demand, with the affected items.
Gets better: learns which items the planner protects and the buffer they like, leading the watch list with the parts this planner actually expedites.
“Firm these planned orders and push the requisitions through approval, expedite the casting line.”
The planner both submits and advances the chain — planned order to purchase requisition to PO approval; Nāda walks the requisitions into D365 workflow, tracks approver status, and chases stuck items.
Gets better: learns which requisitions sail through, which approvers are slow, and when to escalate so material isn't late.
Sales — opportunities, activities, quotes, orders; Supply Chain availability.
“Log the Contoso visit, move the opp to proposal, and set a follow-up for Tuesday.”
Walking out of a meeting, the rep updates the opportunity stage, logs the activity, and sets the next step by voice; Nāda drafts the records and writes them back to Sales.
Gets better: learns the rep's note style, their typical next steps by deal stage, and the cadence they keep — so a one-line spoken summary becomes a complete, on-brand CRM update.
“Quote 200 units of the standard kit — is it available, and what's my usual discount?”
Nāda configures the quote, checks price and availability (Sales + Supply Chain ATP), and drafts the order for submission.
Gets better: learns the rep's common products and discount habits and pre-fills the quote the rep usually builds, flagging when a discount needs approval.
“Brief me on Northwind before my call.”
Nāda assembles the account picture — open opportunities, recent orders, open cases, AR status — and the rep's own last notes, then offers a pipeline review.
Gets better: learns which accounts and which facts the rep wants before a call and pre-stages that briefing ahead of calendar meetings.
“Request approval for the 18 percent discount and the extended terms on Northwind.”
The rep submits discount, pricing, and credit requests into approval by voice and is notified the moment they clear, so they can close in the meeting.
Gets better: learns the rep's common asks, pre-fills the request with the justification the approver expects, and flags when an ask will exceed policy before they submit.
Project Operations — tasks, time/expense, budget, resourcing, billing.
“Approve this week's timesheets for the Riverside project and mark design 80 percent done.”
Nāda surfaces pending time and expense for approval and updates task progress and percent-complete in Project Operations, validated and audited.
Gets better: learns the PM's approval patterns — whose time they rubber-stamp, what they question — and the projects they update on which day, pre-staging the weekly approval run.
“Are we on budget for Riverside, and what's the forecast at completion?”
Nāda reads actuals versus budget, cost-to-date, and estimate-at-completion, naming the lines trending over.
Gets better: learns which projects and cost categories the PM watches and the variance that warrants a flag, leading with the job most likely to slip.
“Request a second engineer for July and approve the change order on scope.”
Resource requests, change-order approvals, and milestone billing route through the PM with the project context, validated and audited.
Gets better: learns the PM's resourcing patterns and change-order thresholds and pre-attaches the schedule and margin impact the PM checks before approving.
“Approve this week's timesheets and expenses for Riverside, approve the change order, hold the subcontractor invoice.”
Timesheet, expense, change-order, and project vendor-invoice approvals route to the PM through Project Operations and D365 workflow; Nāda batches them with budget and margin impact for fast, audited decisions.
Gets better: learns whose time the PM approves on sight, the expense types they question, and the change-order size that needs a second look.