| Posture |
| Primary use case | AI front door for complex catalogs: buyer or partner arrives with a task, receives a validated configuration and commercial quote without waiting on a rep. Full CPQ workflow in one deterministic engine. | SAP CPQ is a cloud configure–price–quote engine for rep, partner, and channel quoting with SAP-stack handoff — discovery is often external to CPQ UI or commerce-led. | Coexistence |
|---|
| Primary user persona | Buyer self-serve (full / guided / sales-assisted); partner portals; internal presales paste-in. — Escalation target: < 10% of sessions. | Rep, deal desk, and partner users in CPQ UI by default. — External buyer UX typically requires Commerce RFQ, Configurations V2 headless build, or a separate product. | Lean Talkulate AI CPQ |
|---|
| Deployment model | Standalone without an SAP CPQ tenant: chat widget, iframe, JS snippet, embedded page, headless API, messengers. Connects to any CRM or ERP stack. | Cloud SaaS CPQ tenant on SAP infrastructure (Azure migration program). Requires SAP CPQ subscription and modeling — not a lightweight embed-only SKU. | Lean Talkulate AI CPQ |
|---|
| Entry UX | Plain-language task entry with optional persona routing (B2B technical / B2C / installer / fleet). — Contextual goals, not rigid decision trees. | Step-driven admin-modeled flows in CPQ UI. Configurations V2 enables custom external UIs on the rules core — you build NL entry separately. | Lean Talkulate AI CPQ |
|---|
| Real-time option swaps | Buyer swaps a compatible alternative → Engineer Agent revalidates the full BOM and reprices instantly. Out-of-rule swaps are blocked at the engine level. | Real-time repricing on option change is possible when UI and rule latency allow; behavior depends on product rules, VC, and integration timing. | Lean Talkulate AI CPQ |
|---|
| Self-serve modes | Three built-in modes: full, guided, and sales-assisted; benchmark < 10% escalation. | Commerce RFQ plus Configurations V2 headless patterns enable external self-serve — typically SI-built portals, not a native buyer-conversation core in CPQ-primary docs. | Lean Talkulate AI CPQ |
|---|
| Discovery & pricing |
| Discovery model | Interviewer Agent: contextual goals — 6–10 questions, 4–8 minutes to validated BOM in reference case. | Discovery is external to CPQ (rep-mediated) or commerce-channel framed. — CX AI on quotes is ML cross-sell / discount recommendations — not buyer NL discovery for configured lines. | Lean Talkulate AI CPQ |
|---|
| CPQ workflow coverage | Entry → Discovery → Selection (Engineer + DB) → Pricing → Output (BOM + proposal + reasoning) → Handoff (CRM / ERP / exception routing) in one engine. | Configure, price, quote in CPQ tenant; subscriptions, services, and hardware on quotes; downstream CRM/ERP/Commerce connectors per product FAQ. | Coexistence |
|---|
| Pricing engine depth | Volume tiers, bundles, regional lists, SLA adders, contract rebates, promotions, multi-currency — validated in the same Engineer Agent pass. | Rules-driven pricing; usage-based subscription models on quotes (2511 release); CX AI pricing recommendations apply to simple (non-configured) lines and non-recurrent prices only. | Lean SAP CPQ |
|---|
| Variant configuration (VC) | Encodes compatibility as product rules in optimized PostgreSQL — electrical, physical, regulatory, and commercial constraints. | Professional edition: utilize existing VC models and MTO, BTO, ETO rules in CPQ; 2511 improves (A)VC / VCPS conflict resolution UX. | Lean SAP CPQ |
|---|
| Output formats | Interactive commercial proposal + PDF + per-line reasoning + compatible swaps + branded PDF. | Quote and proposal templates in CPQ; buyer interactive proposal with per-line AI reasoning is not established in SAP CPQ-primary materials reviewed here. | Lean Talkulate AI CPQ |
|---|
| Approval & discount governance | Discount threshold approvals available as add-on scope per tenant. | Workflow/Approvals documented in Setup and Administration Guide — workflow actions, item actions, approvals, auto-reminder, likelihood of approval. | Lean SAP CPQ |
|---|
| Validation |
| Validation method | Deterministic Engineer Agent against live PostgreSQL via MCP — no document retrieval for core compatibility. | Rules engine plus VC for configured products. CX AI does not prove configured-line constraint validity — different layer from constraint engine. | Coexistence |
|---|
| Invalid configuration handling | Each BOM line is checked against the live catalog via structured queries (not document similarity). — Out-of-rule builds are refused with per-line reasoning before optional handoff into SAP CPQ. A published hardware reference customer (~3,400 SKUs) dropped the routine engineering compatibility pass on standard quotes after go-live. | SAP CPQ relies on variant configuration and product rules in the tenant. — Invalid combinations are blocked when models and master data are complete; residual risk is addressed through rule maintenance and test cycles — not through buyer-side similarity search over PDFs. | Program-dependent |
|---|
| AI on quotes (CX AI) | LLM only in Interviewer Agent; Engineer Agent is non-generative for configuration choices. | CX AI: ML cross-sell and discount recommendations from historical data; requires CX AI cross-platform app, ingestion, and inference mapping in CPQ Providers. | Coexistence |
|---|
| Dual-agent architecture | Interviewer and Engineer fully separated; Engineer only runs allowlisted DB queries. | SAP CPQ core is rules/VC engine — not LLM-fronted for configuration. Prompt-injection is not the core CPQ attack surface. | Lean Talkulate AI CPQ |
|---|
| Catalog source connectors | PostgreSQL, MySQL, MSSQL, SAP, NetSuite, Dynamics, Salesforce CPQ, Excel, PDF specs, XML, REST, file drops. | CPQ tenant catalog plus Product REST API (2502 bulk sync); ERP/VC replication and Sales Area alignment with S/4HANA (2511). | Program-dependent |
|---|
| Per-line audit trail | Per-line reasoning: which constraint triggered each selection — buyer and rep facing. | Audit Trail: admin, user, product, attribute, pricebook, and custom table logs — platform change logging, not buyer-facing “why this line” UX. | Lean Talkulate AI CPQ |
|---|
| Observability & tracing | Langfuse per session (Interviewer and Engineer traces separately); rate limits on chat/overview endpoints. | Tenant Swagger on *.cpq.cloud.sap; performance tuning docs exist — no equivalent LLM trace layer in CPQ core. | Lean Talkulate AI CPQ |
|---|
| Channels & time to value |
| Embed channels | Widget, iframe (~10 min deploy), JS snippet, embedded page, headless API, internal-tool mode. | Configurations V2 (“Headless CPQ”, 2511): external UIs call configuration logic and pricing without CPQ UI login — SI builds the shell. | Coexistence |
|---|
| Async messenger channels | WhatsApp, Telegram, Teams, Slack, Messenger, LinkedIn — same validation engine, scoped per channel. | Messenger-native CPQ intake is not documented as a standard SAP CPQ feature. — Not established in SAP CPQ public CPQ materials reviewed (May 2026); verify with SAP CPQ. | Lean Talkulate AI CPQ |
|---|
| Time to production buyer surface | Envelope 5 days–6 weeks; typical 3–5 weeks; reference case 5 weeks. | Full tenant modeling plus ERP/BRIM integration are multi-workstream programs; calendar depends on catalog complexity and SI scope. | Lean Talkulate AI CPQ |
|---|
| Customer effort | 10–15 hours over 2–3 weeks (vendor claim for standard deployment). | Heavy stakeholder alignment, VC/rule modeling, integration testing, and hypercare — effort scales with catalog and ERP scope. | Lean Talkulate AI CPQ |
|---|
| "No structured catalog" path | Data structuring service ($3,450–$17,250 one-time) for Excel, PDF, and tribal knowledge. | Catalog and VC modeling are admin- and partner-scoped; no public fixed-fee structuring SKU on the product page. | Program-dependent |
|---|
| Verticals | Eight first-class vertical packs on one engine. | Manufacturing / high-tech / telecom-style positioning on product page — partner-led verticalization, not eight named packs. | Lean Talkulate AI CPQ |
|---|
| Security & governance |
| Error mode | Refuse: incompatible combinations are blocked before the buyer accepts output; the engine does not publish out-of-rule builds from document similarity. | Rules and VC block disallowed combinations when modeled; gaps in rules can produce plausible but incorrect quotes. | Program-dependent |
|---|
| Regulatory constraint encoding | CE, FDA, UL, NSF, RoHS, and similar rules encodable in validation engine. | Implementation-dependent in product and price rules — depth depends on SI and customer team. | Program-dependent |
|---|
| Anti-prompt-injection | Engineer Agent allowlisted queries only; vendor does not need internal margin data. | SAP CPQ core quoting is not LLM-fronted; CX AI is a separate integration with its own data handling — confirm in DPA. | Lean Talkulate AI CPQ |
|---|
| GDPR / PII posture | DPA; no training on client catalog or conversations; data minimization. | Trust Center and SOC 2 Type 2 landing for CPQ; full report access restricted (NDA). — CX AI data handling: confirm with SAP. | Program-dependent |
|---|
| Multi-tenant isolation & hosting | Cloud AWS/Azure EU regions; per-tenant DB. — On-prem: $69,000 enterprise license. | Multi-tenant SaaS on Microsoft Azure (tenant-by-tenant migration program). — SOC 2 page lists SAP and Azure DC geographies. | Program-dependent |
|---|
| Built-in analytics | Conversation funnel analytics and demand-sensing export. | No standard SAP Analytics Cloud integration for CPQ reporting per SAP expert guidance — custom Quotes API integrations typical. | Lean Talkulate AI CPQ |
|---|
| Outcomes & commercial |
| Quote cycle (reference metrics) | See the reference deployment band below. | Vendor claim: quotes of 10,000+ lines in seconds — KPI definitions and tenant modeling matter; not directly comparable without alignment. | Program-dependent |
|---|
| First-pass accuracy | See the reference deployment band below. | No matched public study on equivalent catalog complexity in this publication pass. | Lean Talkulate AI CPQ |
|---|
| Quote capacity | See the reference deployment band below. | Capacity uplift for buyer-facing self-serve not published in SAP CPQ materials reviewed here. — Not established in SAP CPQ public CPQ materials reviewed (May 2026); verify with SAP CPQ. | Lean Talkulate AI CPQ |
|---|
| Conversion uplift | An unrelated buyer-facing pilot (not the server-reseller reference) saw better site conversion and fewer abandonments while waiting for quotes versus that pilot’s starting point — directional only; confirm on your own funnel. | Buyer-facing conversion uplift from external CPQ UX is not published in SAP CPQ product page pass. — Not established in SAP CPQ public CPQ materials reviewed (May 2026); verify with SAP CPQ. | Lean Talkulate AI CPQ |
|---|
| RFQ unit economics | Directional RFQ economics (illustrative, not an offer): teams often cite ~$230–$460 for a heavy manual RFQ versus up to about $12 when self-serve automation is at scale — model your tenant on Talkulate AI CPQ pricing / ROI. | License economics are quote-based with no public per-seat grid on the SAP CPQ product page — contact SAP. Capacity scales with licenses, PS, and integration programs. | Lean Talkulate AI CPQ |
|---|
| Pricing model | Implementation ($18,400) + monthly ($1,725, 600 dialogs) + per-dialog overage + integration ($115 / hour). Enterprise on-prem: $69,000. | SAP CPQ product page does not publish numeric list prices (May 2026) — pricing section routes to community/contact. License, PS, and ERP/BRIM integration are separate workstreams. SAP Business AI Base vs Premium (AI Units) applies to some premium AI features — CX AI entitlement is a separate cross-platform application. | Program-dependent |
|---|