Competitive Landscape — CRM + AI for Tea Export / Private-Label
Date: 2026-06-17 Status: Research synthesis for discovery / proposal work. Pricing and AI-feature details are 2026 ballparks from secondary sources and should be re-verified on vendor sites before quoting them to the client. Purpose: Understand what already exists — both tea/agri-commodity-specific tools and mainstream CRMs — so we can position our proposal honestly and learn from incumbents.
1. Headline finding
There is no dominant, sales-and-relationship-centric CRM built for the tea export / private-label business. The market splits into buckets that each leave the same gap:
- Tea / plantation ERP — strong on estate operations, manufacturing, regulatory back-office and traceability; weak-to-absent on B2B sales pipeline, sample→quote→order workflow, and buyer relationship management.
- Agri-commodity CTRM (commodity trade & risk management) — strong on contracts, position, risk, sampling, quality and traceability; built for trading desks, not a private-label manufacturer’s sales team, and heavyweight/expensive. This is the closest analog and the best source of feature ideas.
- Tea-auction & AI-grading platforms (Sri Lanka-specific) — digitize the Colombo auction and quality grading; adjacent infrastructure, not a CRM, but this is where local “tea + AI” mindshare and partnerships sit.
- Generic CRM (Pipedrive, which Empire uses today) — easy pipelines, now with credible AI, but zero tea-domain logic.
The white space is precisely what the discovery docs propose: a sales/private-label CRM with tea-domain workflow (sample tracking, blend/quote, certifications, export docs) wrapped in modern LLM productivity AI. None of the incumbents combine all three.
2. Tea / agri-commodity-specific tools
2.1 The closest analog — iRely (i21 CTRM, dedicated Tea solution)
iRely is the single most relevant industry-specific competitor. It is a commodity trade & risk management platform with an explicit Tea vertical (also coffee, cocoa, cotton, sugar), 35+ years old, and claims use by “some of the largest tea companies in the world.” Documented modules that map directly to Empire’s workflow:
- Sampling: record samples received/sent, capture quality results against user-defined criteria, sample inquiries to chase pending results, and contract inquiries that flag contracts missing expected samples.
- Blending / production: blend management, blend-sheet generation, recipe management, lot-level blend tracking.
- Quality management: unlimited tests/properties with spec limits and in/out-of-spec flags.
- Traceability: full forward/backward lot trace and mock recalls.
- Plus demand planning, procurement, logistics, inventory, finance and costing.
What it leaves open: it is a heavyweight trader/ERP-grade system, not a lightweight, sales-rep-friendly CRM. No modern conversational AI is confirmed (AI references are marketing-level), and it is likely costly and slow to deploy. It is the best feature benchmark to study — especially the “sample inquiry” and “contract missing sample” patterns — but it is not how Empire’s sales team would want to work day to day.
2.2 Other agri-commodity CTRM platforms (feature inspiration)
- Eximware (Partner XM): cloud CTRM for importers/traders of coffee, tea, cocoa, nuts, spices; built-in forecasting, purchase contracts, logistics, QC, sales. Importer/buy-side oriented; no confirmed AI.
- Agiboo (Agiblocks): agri/soft-commodity CTRM with recently strengthened sampling features and an integration with CommodityAI for AI-driven shipment management and workflow automation — one of the clearer real AI stories in the space. Notably, the Agiboo ecosystem’s CRM piece (AgCRM) is built on Salesforce — i.e., even CTRM vendors bolt on a generic CRM rather than build a native one. That reinforces our white space.
- DycoTrade: CTRM on Microsoft Dynamics 365 with a “Commodity Trade Copilot” built on MS Copilot — a good reference for what an AI assistant looks like for traders/logistics/finance.
- Fendahl, ION Aspect, ComFin, Mindsprint, CTRM Next: enterprise CTRM platforms, risk/position-centric, not tea-specific, not sales CRMs.
Cross-cutting gap: all are trade/risk/position systems for trading desks. None is a nimble sales + private-label-project CRM, and where AI exists it targets price/risk, not sales productivity, sample-chasing, buyer comms or quote drafting.
2.3 Tea / plantation ERP (mostly estate & manufacturing)
- CICRA Solutions (Sri Lanka): ~20 years in Sri Lankan tea software; built the Tea Board back-office system tSyst and the original Colombo Tea e-Auction platform. Regulatory/auction infrastructure, not an exporter’s commercial CRM. Architecturally dated (C#/ASP.NET, Crystal Reports).
- OKLO (Sri Lanka) — SmartAuction: current operator of the upgraded Ceylon Tea SmartAuction for the Colombo Tea Traders Association, with a factory segment for valuing/grading and catalogue printing. Associated with Microsoft; AI is aspirational rather than confirmed. The incumbent local “tea-tech” brand — a credible potential partner as much as a competitor.
- Blue Lotus 360 (Sri Lanka): cloud ERP with a tea-plantation configuration and generic sales-force automation; markets “AI-powered” ERP (depth unverified). Worth watching as a local competitor.
- TEAlink (India), Likemind (India), EBiz Clouds / BIZ-TEA, Sage X3, O2B/Odoo, ERPNext, BatchMaster: tea-estate and tea-manufacturing ERP — green-leaf intake through withering/rolling/fermentation/drying/grading/blending/packing, inventory, traceability, finance. EBiz Clouds notably claims export documentation. All are production/ops ERP, not sales relationship management; little verifiable AI.
Two unverified leads to discard or confirm: a “3i Infotech tea product” and a vendor named “Logicabeans” could not be substantiated as real tea software. Treat as misremembered until confirmed.
2.4 Export documentation / compliance (Sri Lanka)
No tea-specific export-documentation product surfaced. Sri Lankan tea exporters register with Customs, submit CUSDEC declarations and supporting documents to the Tea Board’s Tea Exports section via the national ASYCUDA system, and assemble the usual export set (commercial invoice, packing list, B/L or AWB, Certificate of Origin via a registered chamber, phytosanitary/health certs, lab reports) plus LC or DA/DP payment handling. Generic export-doc tools exist (e.g. Shipping Solutions in the US) but are not Sri Lanka- or tea-localized.
Opportunity: no tool ties buyer/order → required cert & doc set → LC milestones → Tea Board / ASYCUDA submission into one workflow for a Sri Lankan tea exporter. A CRM that auto-assembles the correct document/cert checklist per destination and tracks LC deadlines would be genuinely differentiated. (ASYCUDA integration depth is unverified and needs local diligence.)
2.5 Where AI actually exists in the tea world today
Real tea-domain AI in Sri Lanka is operational, not commercial: national projects for green-leaf quality grading via image classification and AI visual grading of auction samples. The local AI narrative is entirely about grading and auction. That means a sales/CRM productivity AI story is fresh and complementary, not me-too — and potentially partner-friendly with the grading/auction players rather than competitive.
2.6 How Sri Lankan tea exporters typically run sales (anecdotal)
Inferred from industry sources, to validate with the client: larger players run heavyweight ERP (Empire’s group already implemented IFS Applications for Empire Tea Kenya), while the sales/CRM layer is generic or ad-hoc — Pipedrive, spreadsheets, and WhatsApp/email for buyer comms, sample dispatch and quoting. The tea export sale is sample-driven in a way no generic CRM models natively. That last mile — the buyer relationship and the sample→quote→doc workflow — is the unserved layer.
3. Mainstream CRM + AI options
If we configure an existing CRM rather than build, these are the realistic candidates. All figures are 2026 ballparks — re-verify before quoting.
3.1 Pipedrive (their current tool)
Pipedrive has gone “agentic” in 2026: an AI Sales Assistant (next-step suggestions, on the entry tier), Pipedrive Pulse (intent/sentiment analysis to surface high-intent deals), an agentic email experience (drafts, thread summaries, urgency flagging), AI call/meeting summaries, and Insights AI (natural-language reporting). Pricing starts ~$14/user/mo (annual) across four tiers (Lite / Growth / Premium / Ultimate); exact upper-tier prices need confirming. Quotes/proposals come via Smart Docs (bundled in Premium/Ultimate or a ~$32.50/mo add-on) which auto-inserts deal/contact/product data and supports e-signatures, but there is no deep native CPQ for configurable private-label specs. Full REST API + webhooks (May 2026 added deal-product/installment webhooks), but engineers flag friction: custom fields exposed as 40-char hashes, split pagination, tight burst rate-limits.
Verdict: easy, cheap, surprisingly good agentic email AI, and lowest switching cost since they already use it. Weak as an operations backbone (no inventory/lot/shipment bridge), shallow CPQ, and a moderately awkward API for a heavy custom AI layer.
3.2 Zoho CRM — strongest cost-sensitive / South Asia fit
Zia AI provides deal predictions, anomaly detection, sentiment analysis and more (gated to Enterprise+, with agentic agents on Ultimate). Pricing is notably lower than rivals: Professional ~$23, Enterprise ~$40, Ultimate ~$52 per user/mo (annual). Crucially, Zoho has the strongest native fit for our workflow: built-in Products, Quotes, Sales Orders, Invoices, a native CPQ (line items, discounts, bundling), and Blueprint (drag-and-drop sales-process enforcement) ideal for standardizing sample→quote→order→shipment. The wider Zoho One suite (Inventory, Books, Sign, Analytics) can cover operations cheaply, and Zoho is widely adopted in South Asia (local familiarity and support). Limits: Zia behind Enterprise+, UI polish below HubSpot, agentic AI less mature today.
3.3 Odoo — CRM + operations bridge, self-host option
Odoo is an ERP-adjacent modular suite, so CRM, quotations/sales orders, inventory/lots, purchasing, invoicing and shipping live in one system — the operational integration no pure CRM offers natively. Community edition is free and self-hostable (CRM, Sales, Inventory, Accounting); Enterprise is ~$25–37/user/mo and unlocks richer AI (Community AI is modest: text auto-fill and semantic search). Full external API and deep customizability. Limits: weakest sales-AI of the group, self-hosting carries DevOps burden, UX trails dedicated CRMs. Best if operations integration matters more than cutting-edge AI.
3.4 HubSpot
Best-in-class UX and a strong agent story (Breeze Copilot + Customer/Prospecting/Data agents GA). But the features we need — custom objects (to model Sample Request, Blend Spec, Lot) and CPQ — push to Enterprise (~$150/seat/mo) plus a CPQ add-on (~$84/user/mo), onboarding fees (~$3,500), and Breeze usage credits stacked on top. Powerful, developer-friendly API, but cost-misaligned for a single SME exporter unless they want the unified marketing+sales+service suite.
3.5 Salesforce
The most mature agentic platform (Agentforce + Einstein) and unmatched capability, but base Enterprise ~$165/user/mo and a realistic AI-enabled config runs ~$560/user/mo before implementation. Salesforce itself steers SMBs toward lighter products. Almost certainly over-engineered and over-priced for Empire’s size unless large multi-entity growth is anticipated.
3.6 Custom-build with an AI layer
2026 patterns: a thin data layer (Postgres/Supabase) + web app + LLM workflows behind human-in-the-loop approval gates, orchestrated with frameworks like LangGraph (stateful agents with approval gates), Semantic Kernel (governance-leaning) or AutoGen (built-in HITL). Upside: exact fit to tea export/private-label workflow, no per-seat fees, full control of the AI UX, own the data. Downside: you rebuild everything a CRM gives free (auth, audit, mobile, email sync, reporting, dedupe, permissions, uptime) and carry ongoing maintenance/security burden.
4. Side-by-side
| Option | AI maturity (2026) | Ballpark $/user/mo | Native quote/sample fit | Ops/ERP bridge | Cost fit for SME exporter |
|---|---|---|---|---|---|
| Pipedrive (incumbent) | Good (agentic email/Pulse) | ~$14–50+ | Smart Docs add-on; no real CPQ | None | Strong |
| Zoho CRM | Good, improving | ~$23–52 | Native CPQ + inventory + Blueprint | Via Zoho One | Strong |
| Odoo | Modest | Free / ~$25–37 | Native quotes + sales orders | Best (built-in ERP) | Strong (esp. self-host) |
| HubSpot | Strong (Breeze agents) | $100–150 + credits + onboarding | Custom objects = Enterprise; CPQ add-on | None | Weak (expensive at needed tier) |
| Salesforce | Strongest (Agentforce) | ~$165 base, ~$560 AI-enabled | Full (CPQ add-on) | Via add-ons | Weak (over-engineered) |
| Custom + AI | As good as we build | infra + dev cost | Exact-fit | Build it | Depends |
| iRely (Tea CTRM) | Thin/marketing | enterprise (high) | Native sampling/blending/quality | Full (it is the ops system) | Weak (heavy, slow, costly) |
5. Recommended reading of the field
For a cost-sensitive Sri Lankan exporter needing quote/sample workflows tied to operations:
- Zoho CRM and Odoo are the strongest “configure SaaS” candidates (native CPQ/inventory/Blueprint and South-Asia familiarity for Zoho; CRM↔ERP integration and self-host for Odoo).
- Pipedrive is the low-friction incumbent-upgrade with surprisingly good agentic AI but weak operations depth.
- HubSpot / Salesforce are powerful but cost-misaligned at the tier their features require.
- iRely is the best feature benchmark (sample inquiries, blend sheets, contract-missing-sample flags) even though it is not how the sales team would want to work.
The most defensible recommendation is a hybrid: configure Zoho or Odoo as the system of record for contacts/deals/quotes/inventory, and add a thin, human-in-the-loop AI layer for blend-spec parsing, quote drafting, sample follow-up nudges, and export-doc checklists — capturing most of the custom-build upside without rebuilding CRM plumbing. This should be presented as a hypothesis to test in discovery, not a locked decision.
5.1 Why custom beats Zoho/Odoo for Empire Teas — updated recommendation
Research basis: deep-research workflow run July 2026, adversarially verified against official Zoho and Odoo documentation. Confidence levels noted per claim. The section 5 recommendation to configure Zoho/Odoo was the correct hypothesis to test; the findings below show it does not survive contact with Empire Teas’ specific requirements.
Bottom line: By the time you’ve worked around Zoho’s Blueprint limits, Odoo’s missing stage gating, the certification mismatch gap in both, and the Claude integration work in both — you’ve done the same amount of custom development as a greenfield build. The difference is that the greenfield gives you a data model you own, a workflow that matches the tea domain exactly, and no per-seat fees. Zoho or Odoo gives you the same dev cost but a platform that actively fights you.
The five gaps that close the case
1. Sample tracking with multi-round approval
Empire Teas’ private label pipeline needs 8+ stages, each with an approve/reject/revise gate. This is not an edge case — it is the core of the private label workflow.
- Zoho: Blueprint (the only tool that enforces stage gating) is unavailable on Free and Standard. Professional tier allows 3 blueprints with 10 transitions maximum (confirmed from official Zoho docs, 3-0 vote). An 8-stage pipeline with approve/reject/revise at each round exceeds this limit before you even reach stage 5. Enterprise is required — and even then, Blueprint operates on the Deals module only.
- Odoo: No native stage approval gating whatsoever. The feature is absent from core Odoo CRM documentation. Multiple competing third-party paid modules (confirmed via Odoo Apps Store, 2-1 vote) add approve/reject buttons — which would be redundant if the feature were native. The gap persists in Odoo 18.0 (Jan 2025 module update confirms this is still a workaround, not a native feature).
Consequence: multi-round sample tracking requires custom development on both platforms regardless. You are not configuring — you are building. The only question is whether you build on top of a platform you do not control.
2. Two truly independent pipelines with different fields
Empire Teas needs two fundamentally different pipelines: private label (8+ stages, 4–6 month cycle, sample tracking, artwork approval, cert checks) and bulk (3–4 stages, weeks, price-driven). These need different fields, different stage names, different alert thresholds, and different automation rules.
- Zoho: Multiple pipelines confirmed native (3-0 vote), but with three hard constraints (all confirmed from official Zoho docs): (a) pipelines are restricted to the Deals module only — cannot be applied to custom modules; (b) if the same stage name appears in two pipelines, the probability value is shared not independently configurable (2-1 vote); © pipeline assignment during deal creation is manual — there is no automatic association at creation time (2-1 vote). Layout rules can conditionally show/hide fields per pipeline (3-0 vote), but this is conditional display logic, not fully independent field schemas.
- Odoo: Pipelines are implemented via Sales Teams, not as native independent pipeline objects. Each team gets its own stages. Third-party modules attempt to address this but were not confirmed to support per-pipeline custom fields or automation rules.
Consequence: you can approximate two pipelines on Zoho, but the shared-probability constraint and manual assignment are friction points that require workarounds. Odoo’s team-based pipeline model is a structural mismatch with the private label vs. bulk distinction.
3. Certification mismatch detection
One of the most valuable features for Empire Teas: when a salesperson creates a deal for a buyer who requires EU Organic, the CRM should automatically flag if the product being offered does not hold EU Organic certification.
- Zoho: Validation rules exist (confirmed) but cannot be applied to multi-select lookups, multi-picklists, or formula fields (confirmed from official Zoho validation rules documentation). Certification data is inherently multi-value on both sides (a buyer may require 3 certs; a product may hold 7). This means the cross-reference check — “does this product hold every cert this buyer requires?” — cannot be implemented as a native validation rule. It requires a custom Deluge function or an external webhook.
- Odoo: No native certification mismatch detection. Would require a custom module.
Consequence: this feature is custom development on both platforms. There is no configuration path.
4. AI customisation — tea domain knowledge
The AI features in both platforms are generic. Neither can be tuned to tea grades, Ceylon Tea certification names, private label workflow terminology, or Incoterms without developer work.
- Zoho Zia: available on Enterprise and above only. Provides deal predictions, anomaly detection, sentiment analysis — all based on CRM data patterns. No mechanism confirmed for injecting domain-specific context (tea grades, cert names, stage terminology) into AI behaviour. Zoho MCP (launched July 2025, 3-0 vote) allows Claude to read and write Zoho CRM records conversationally, but operates as a conversational bridge only — not an event-based automation trigger. This means the AI layer must be driven by a human opening an MCP-connected chat, not by a new inquiry email arriving.
- Odoo: natively supports only OpenAI (ChatGPT) and Google Gemini (confirmed from official Odoo 19.0 documentation, 3-0 vote). Claude is not a built-in provider option. Connecting Claude requires a custom module or middleware — confirmed by three separate third-party Anthropic modules on the Odoo Apps Store and a community forum thread noting no native Claude support as of mid-2026.
Consequence: calling Claude from a backend and writing AI-extracted data back into CRM records — which is the core of Empire’s AI layer — requires custom API integration work on both platforms. The custom build calls Anthropic directly from its own backend with zero middleware tax.
5. API quality for a custom AI layer
Even if you accept both platforms, bolting the AI layer on via API has real constraints.
- Zoho: credit-based daily API model. Enterprise/Zoho One gets 50,000 base credits plus 1,000 per user per 24-hour window (confirmed from official Zoho CRM v8 API docs, 3-0 vote). Critically, resource-intensive operations — including bulk upsert over 10 records, Search, Query, Composite, and Convert Lead — share a sub-concurrency cap of 10 across all CRM editions (3-0 vote). Any background AI process writing results back in bulk hits this cap.
- Odoo: external API access is restricted to the Custom pricing plan only — the highest SaaS tier, approximately $49/user/month (confirmed from official Odoo 18.0 documentation, 2-1 vote). The API is not available on One App Free or Standard plans. For 15 users, this is ~$735/month just for API access — before any implementation or custom module cost.
The real cost of “configure Zoho/Odoo”
The section 5 hybrid recommendation assumed configuration was the majority of the work. The research shows it is not. Here is what custom development would be required on either platform:
| Feature | Zoho | Odoo |
|---|---|---|
| Multi-round sample approval gating | Custom Deluge function + Blueprint workaround (Enterprise) | Third-party paid module or custom module |
| Cert mismatch detection | Custom Deluge function or external webhook | Custom module |
| Claude API integration | MCP (conversational only, not event-triggered) or custom webhook middleware | Custom module (no native Claude support) |
| Two truly independent pipelines with separate field schemas | Layout rules workaround; shared probability is a hard constraint | Sales Team workaround; no native independent pipeline objects |
| Automatic pipeline assignment at deal creation | Workflow automation post-creation (not at creation) | Manual or custom logic |
By the time this work is done, you have paid Enterprise licensing (Zoho) or Custom plan fees (Odoo) on top of the same custom development hours as a greenfield build. The platform gives you auth, mobile, and reporting — but also gives you constraints you cannot remove, a data model you cannot fully own, and per-seat costs that compound as the team grows.
A custom build at Empire’s scale (10–20 users) has:
- No per-seat fees after build
- A data model designed specifically for private label + bulk tea export
- Claude called directly from the backend — no middleware, no provider restrictions, no MCP-only conversational limitation
- Stage approval gating built exactly as the workflow requires, not approximated around Blueprint limits
- Certification mismatch detection as a first-class feature, not a workaround
- Estimated AI running cost: ~$7/year at Claude Haiku rates for Empire’s inquiry volume
Updated recommendation: proceed with custom build. Use Zoho CRM and Odoo as feature benchmarks (especially iRely’s sample inquiry and contract-missing-sample patterns), not as the implementation target.
6. How to use this with the client
- Lead with the gap, not a product. “Nobody has built the sales/private-label/AI CRM for tea — the existing tools are either estate/manufacturing ERP, trader CTRM, or generic CRM with no tea logic.” This is credible and differentiating.
- Borrow CTRM credibility. Reference iRely’s sample-inquiry and blend-sheet patterns to show we understand the domain, while positioning lighter and AI-native.
- Position the local AI players (OKLO, grading projects) as complementary/partners, not threats — our story is sales productivity, theirs is grading/auction.
- Be honest about the build-vs-configure tradeoff. Present Zoho/Odoo hybrid as the pragmatic path and custom as the differentiated path; let discovery findings (data volume, ERP API, budget, IT appetite) decide.
7. Confidence flags
- Confirmed: the existence and positioning of iRely/Eximware/Agiboo/DycoTrade; CICRA and OKLO roles in the Colombo auction; Empire Tea Kenya’s IFS Applications implementation; the Sri Lanka export-doc/ASYCUDA process; the general 2026 AI/feature direction of Pipedrive, HubSpot, Zoho, Odoo, Salesforce.
- Thin / re-verify: all specific pricing tiers and AI add-on costs; depth of AI in iRely/Eximware/Blue Lotus/OKLO; whether vendors’ newest agentic features are GA in the Sri Lanka region.
- Discard unless confirmed: “3i Infotech tea product” and “Logicabeans”; the product name “TeaRetina” for AI grading.
- Local diligence needed: current auction operator nuance (CICRA vs OKLO), and any ASYCUDA / Tea Board API integration feasibility.
8. Sources
Tea / agri-commodity:
- iRely Tea CTRM — https://irely.com/solutions/commodity-trading-and-procurement/tea/ ; https://www.ctrmcenter.com/resources/irely/
- Eximware — https://www.eximware.com/pxm
- Agiboo / Agiblocks sampling & CommodityAI — https://www.ctrmcenter.com/resources/agiboo/ ; https://www.ctrmcenter.com/ctrm-community/commodityai-integrated-into-agiblocks-ctrm-modernizing-commodity-supply-chains-with-ai-powered-automation/
- DycoTrade Commodity Trade Copilot — https://marketplace.microsoft.com/en-us/product/dynamics-365-for-operations/dycotrade.dct-advanced-ctrm
- CICRA tSyst & e-Auction — https://solutions.cicra.lk/sri-lanka-tea-board-back-office-system/ ; https://www.ft.lk/front-page/CICRA-Solutions-Colombo-Tea-e-Auction-platform-marks-first-year-with-Rs-175-b-sales/44-715928
- OKLO SmartAuction — https://www.okloworld.com/blog/oklo-upgrades-the-ceylon-tea-e-auction-platform/
- Microsoft / Sri Lanka tea tech — https://news.microsoft.com/apac/2022/02/04/turning-over-a-new-leaf-how-tech-is-transforming-sri-lankas-tea-industry/
- Blue Lotus 360 tea ERP — https://bluelotus360.com/erp-for-tea-plantation/
- TEAlink — https://tealink.in/ ; EBiz Clouds — https://ebizclouds.com/tea-industry-erp/
- Empire Tea Kenya / IFS — https://www.dailymirror.lk/business/Inivos-facilitates-Empire-Tea-Kenya-go-live-with-IFS-Applications/215-262567
- Sri Lanka Tea Board export division — https://srilankateaboard.lk/sri-lanka-tea-board/divisions-and-units/tea-export-division/
- Sri Lanka export procedure / ASYCUDA — https://www.srilankabusiness.com/pdf/export-procedure-21-10-2014.pdf
Mainstream CRM + AI:
- Pipedrive agentic AI — https://www.pipedrive.com/en/newsroom/pipedrive-unveils-next-generation-ai-crm-with-agentic-experience ; Smart Docs — https://www.pipedrive.com/en/features/smart-docs-add-on ; pricing — https://www.g2.com/products/pipedrive/pricing
- HubSpot Breeze — https://www.hubspot.com/products/artificial-intelligence/breeze-ai-agents ; pricing — https://blog.hubspot.com/sales/hubspot-sales-hub-pricing
- Zoho CRM Zia / CPQ / pricing — https://www.zoho.com/crm/cpq.html ; https://saascrmreview.com/zoho-crm-pricing/
- Odoo pricing / AI — https://oec.sh/odoo-pricing ; https://www.erpclaw.ai/blog/odoo-enterprise-ai-pricing-2026/
- Salesforce / Agentforce pricing — https://saascrmreview.com/salesforce-pricing/
- AI agent frameworks 2026 — https://www.lindy.ai/blog/best-ai-agent-frameworks