Live
Co-Pilot
Conversational buying, Smart Buy Again, sourcing preparation, price evidence, budgets, orders and supported spend questions, with buyer confirmation.
Explore Co-Pilot19 buyer-facing capabilities, each shown as a picture of what it does and where a person decides. Co-Pilot is live. Sift, Scout, Sentry and WhatsApp Ordering & Approval are live for approved beta testers. All visuals use sample data, not customer records.
Co-Pilot and RapidSource are capabilities, not extra products to choose between. This guide brings their workflows and the specialist products together without replacing your purchasing authority.
Lapasar Procurement OS + B2B Marketplace: a full procure-to-pay suite with no licensing fee. Catalogue buying, requisitions, budgets, multi-level approvals, purchase orders, goods receipt, three-way matching and payment readiness, with AI controls and scoped ERP PunchOut. Available now.
Real vendors and products, AI search, photo and list matching, price benchmarks and Co-Pilot, connected to Procurement OS controls. Lapasar Procurement OS + B2B Marketplace: a full procure-to-pay suite with no licensing fee. Available now.
An AI sourcing agent for items outside the catalogue. It invites sellers, collects quotes in one format and bargains within your limits. You send and you pick. Live for approved beta testers.
Reads your spend files, matches the same item across names and entities, and shows paid, Lapasar and benchmark prices side by side. Live for approved beta testers.
Runs 14 procurement checks across your purchases, quotations, contracts and suppliers, then routes flags to people who decide. Live for approved beta testers.
Search, reorder, approve and confirm delivery in chat, with the same catalogue, budgets, roles and approvals as the platform. Live for approved beta testers.
Co-Pilot is live. Scout, Sift, Sentry and WhatsApp Ordering are live beta products, open only to approved beta testers. Requesting access does not automatically grant it.
Live
Conversational buying, Smart Buy Again, sourcing preparation, price evidence, budgets, orders and supported spend questions, with buyer confirmation.
Explore Co-PilotLive beta · Approved testers only
AI sourcing agent: invites sellers, collects quotes, bargains within your limits.
Explore Scout by LapasarPast lines matched for comparison
Live beta · Approved testers only
Matches items across your spend files and compares paid, Lapasar and benchmark prices.
Explore Sift by LapasarLive beta · Approved testers only
Runs 14 procurement checks and routes flags to people who decide.
Explore Sentry by LapasarLive beta · Approved testers only
Search, reorder, approve and confirm delivery on WhatsApp.
Explore WhatsApp Ordering & ApprovalEach one is labelled by what it does: advises, explains, prepares an action for you to confirm, or reads without changing anything. Nothing here awards, approves or orders on its own, and no capability guarantees savings.
Showing all 19 capabilities.
A conversational buying workspace. Describe what you need; Co-Pilot pulls out items, quantities and budget, searches the real catalogue, compares options and prepares carts, quotations and RFQs for you to review.
Boundary: Not an autonomous purchasing agent. Cart additions and purchasing actions need your confirmation, and approval rules still apply.
Reorder from your authorised purchase history. Select up to 20 items, check current offers for your delivery location and quantities, and see current against previous prices.
Boundary: Hybrid: purchase frequency and price checks come from records, not AI. No automatic repeat orders and no silent substitutions.
Answer cites only the results shown. Prices come from your account.
A short explanation alongside marketplace search results, grounded in the products that actually matched and the prices that apply to your account.
Boundary: Does not set prices and must not mention products that are not in the results.
Visible clues
Upload a picture of a part or product. AI reads visible brand and model clues, tries candidate searches and returns catalogue results or a clear fallback.
Boundary: Recognition is not proof of exact model, specification or compatibility. You review the match.
Paste a list or upload a spreadsheet, image or document. Line items and quantities are extracted, matched to catalogue candidates, and unmatched or ambiguous lines are set aside for you.
Boundary: Extracted quantities, units and models need review. An extracted list is not an approved order.
Dots: recent internal purchases of the same listing
Checks the offer you actually selected against internal purchasing evidence and explains whether better eligible options may exist.
Boundary: Internal evidence is not the external market. A cheaper item is not automatically equivalent, and item price is not delivered-cost savings.
Looks for external price evidence on comparable products, shows the evidence and its limits, or says plainly that comparability is insufficient.
Boundary: A benchmark is not a supplier quotation. Unknown shipping, tax, stock or pack contents stay unknown.
Suggests Vendor K: lowest price. Warranty not stated: flagged.
Summarises supplier responses to an RFQ: price spread, lead times, warranty and responsiveness where recorded, a recommended vendor with reasons, risk notes and negotiation suggestions.
Boundary: The recommendation does not award the RFQ. Missing supplier facts are flagged, not filled in.
Interprets old purchase descriptions and proposes catalogue matches so past prices can be compared with current ones.
Boundary: Similar names do not prove identical products. Savings are only as reliable as the accepted match, unit and pack basis.
A concise brief for the approver, built from the current purchasing record: what is being bought, why it may deserve attention, and price and procurement context where available.
Boundary: AI advice is not an approval. Your approval permissions and stages remain the authority.
Only codes in the list above can be suggested
Suggests GL codes for purchase lines, and at order level where applicable, using only your company's allowed codes and the item descriptions.
Boundary: Suggestions stay within your configured codes and do not replace finance review.
Metric: spend · Filter: category · Period: 6 months
Ask a reporting question in plain language. It becomes a constrained report on permitted metrics, filters and periods, explained in words with a chart you can save.
Boundary: Uses an allowed reporting vocabulary, not free-form queries. It cannot see another company's data and declines unsupported questions.
AI writes the summary. Numbers come from calculation.
A readable periodic summary of procurement outcomes, with breakdowns and suggested next actions.
Boundary: Totals come from calculation logic. AI writes the explanation and must not add savings of its own.
Describe what you are looking for in the activity log and it becomes a supported search and filter, within the records your role can see.
Boundary: Search help does not establish cause. Recorded events are the evidence.
Scope: Finance dept · 1-31 Mar
For authorised reviewers: traces a specific question through filtered activity timelines and sessions, and returns a cited explanation whose references are checked.
Boundary: Read-only, inside an approved scope. It cannot repair anything or create missing history, and gaps in telemetry are disclosed.
Support chat for your orders: order details, proof of delivery, existing tickets, shipping questions, reminders and cancellation requests, with a human agent when needed.
Boundary: A cancellation request is not an immediate cancellation. Every lookup is limited to your company, and a person is always reachable.
Source: Training ch. 6.2
Answers how-to questions from the platform's training chapters, step by step, with sources cited. Where your role allows, it can add facts from your own records.
Boundary: Documented training is the source. It does not invent controls, procedures or record facts.
For company administrators: describe a setup change, such as a new cost centre, budget allocation or approval assignment, and get a change plan to preview before anything is applied.
Boundary: Nothing is applied without your review, and server validation rejects invalid or conflicting plans. No open database access.
Behind the scenes, AI helps catalogue staff categorise products, spot likely duplicates and flag unit or pack sizes that could mislead buyers.
Boundary: Findings go to people for review. A finding is not a correction, and no finding is not proof of correctness.
These are documented capabilities, not a certification that every feature is enabled for every account. What you can use depends on your permissions, your company configuration and the data available. Examples on this page are illustrative, not customer records.
Co-Pilot is for buying. Support, training and setup each have their own assistant with its own scope.
Buyers
Co-Pilot
Buy, source and understand your procurement position.
Buyers needing help
Lapasar Assistant
Support for orders, deliveries, tickets and cancellations.
Any user learning
Training assistant
How the platform works, cited from training material.
Company administrators
Configuration assistant
Reviewed change plans for company setup.
A
Approaches eligible suppliers for better prices on your behalf, in controlled rounds under the acceptance rules your company sets. Suppliers can accept, counter or decline, and qualifying accepted prices become available when buyers purchase.
A4 copier paper 80gsm, box of 5 reams
Eligible by policy · max 2 rounds · accept at 3% or more off
Was RM62.00
Next buyer to add this item sees the accepted price in the catalogue.
B
Looks across recent requests for purchases that appear to have been broken into smaller orders to stay under a per-purchase approval limit. It flags them for the approver rather than blocking them, so a person makes the call.
Requester: Site maintenance · Category: Electrical cabling · Approval limit RM10,000.00
Each bar is measured against the RM10,000.00 limit
Possible split purchase. 3 orders from one requester within 48 hours, each just under the limit. Combined RM29,400.00 would route to the next approval tier.
C
Interprets item descriptions to suggest supported categories and eligible company GL codes. These suggestions reduce manual classification work without replacing required finance review.
Codes come from your own chart of accounts. Finance can review and correct any line.
D
Explains the selected offer using supported internal price evidence or external market benchmarks. Item matching and pack comparability need review; prices and totals come from records and calculations, not AI invention.
Safety helmet, ANSI Z89.1 type I, white · Qty 24 · Matched specification
Supplier Alowest unit price
RM41.50 x 24 + delivery RM85.00 · Cash on delivery
RM1,081.00
Supplier B
RM43.20 x 24 + delivery free · 30-day credit
RM1,036.80
Lowest landed
Supplier C
RM42.10 x 24 + delivery RM40.00 · 30-day credit
RM1,050.40
The lowest unit price is not the lowest total once delivery and payment terms are counted.
AI features run inside the same roles, budgets and approval rules as the rest of the platform.