Competitor Dossier — SMB Chatbot-Widget Builders

Scope: The self-serve / SMB chatbot-widget builder category — our direct competitors as an EU-sovereign, self-hostable AI-agent + embodied-avatar website widget. Deep-dives on Chatbase, Tidio (+Lyro), Voiceflow, Botpress, Chatbot.com (LiveChat/Text), Intercom (SMB tier/Fin), Crisp; shorter notes on Landbot, ManyChat, Chatsimple.

Method: Web research against official product/pricing/docs pages, verified to current 2026 state, with third-party cross-checks. Sources cited inline. Honest and critical: marketing claims are flagged where they diverge from real capability.

TL;DR for OpenAlice strategists (read this first):

• Embodied avatar = uncontested whitespace. ZERO platforms in this set ship a 3D/animated/talking-head avatar. Not one. This is our clean moat.
• Voice is the live battleground for 2026. Chatbase, Intercom/Fin, and Chatsimple shipped voice/phone agents; Tidio, Crisp, Landbot, ManyChat have none. Voiceflow has the deepest telephony heritage.
• EU data residency is rarer than expected and a real wedge. Crisp (NL/DE-hosted, French co.) is the EU-sovereignty benchmark. Tidio claims EEA storage. Intercom offers EU/AU regions (gated to higher annual plans). Most others are US-centric with no residency commitment.
• Self-hosting is effectively dead across the field. Botpress markets open-source, but the runnable self-host server (v12, AGPL) is sunset — new builds are Cloud-only. Nobody else even pretends. Our self-host story is differentiated by default.
• Pricing is universally opaque. "No per-resolution fees," "free AI," and headline base prices routinely hide model-weighted credits, per-conversation add-ons, and steep mid-market cliffs.

1. Chatbase (chatbase.co)

Positioning + scale. "The leading AI customer-service platform" — train an AI support agent on your data, give it actions, deploy across chat/voice/social (chatbase.co↗). AI-first, developer-flavored. 100% bootstrapped, ~$10M ARR, ~26 employees, 10,000+ paying customers across 80 countries (wboc press release↗; productled.com↗). Zero outside funding (no VC pressure, no war chest). Named customers: IHG, Miele, National Grid, Noon. Built on Supabase (supabase.com↗).

Pricing — model: credit-based (per-message), flat monthly, NOT per-resolution, NOT per-seat (pricing page↗):

Pricing critique (marketing vs reality): Chatbase loudly advertises "no per-resolution fees" vs Intercom Fin — true, BUT credit cost varies 5× by model: standard models = 1 credit/response; GPT-5.2/Gemini Pro = 2, Claude Sonnet = 3, Grok 4 = 4, Claude Opus = 5 credits/response (chatarmin.com↗). Premium-model usage silently shrinks your budget 5×. Add-ons: extra credits ~$12–$40/1k, remove branding $1,188/yr, extra agent $300/agent/yr. Knowledge base is storage-capped (400KB–40MB) — chokes large KBs.

Feature inventory:

1. LLM — Multi-model, user-selectable: GPT-5.2, Claude Sonnet/Opus, Gemini, DeepSeek, Meta, Mistral, Grok. Broadest model menu in the set.
2. Knowledge/RAG — docs/PDF/URL/text ingest, auto-retrain, "tickets as source"; storage-capped.
3. Omnichannel — Web, WhatsApp, Instagram, Messenger, Slack, email (blog↗).
4. Voice — YES. Voice AI Agents (May 2026): inbound phone via Twilio, real-time, 95+ languages, inherits chat agent's full action library (globenewswire↗). Standard+.
5. Avatar/embodiment — NONE.
6. Persona/brand — custom personas/tone, brand-voice; white-label ($1,188/yr or Enterprise).
7. Lead/CRM/commerce — lead capture + qualification; Shopify, Stripe, HubSpot, Salesforce "AI Actions." Critics call CRM depth shallow.
8. Analytics — advanced reporting, agent-performance insights, source suggestions (Pro).
9. Human handoff — smart escalation to live chat/helpdesk tickets (Standard+).
10. A/B testing — NOT documented (absent).
11. Integrations/API — public API + client libraries; Stripe, Zendesk, Salesforce, HubSpot, Shopify, Slack, WhatsApp, Calendly, Twilio, WordPress, Zapier + Make. API-first strength.
12. Security — SOC 2 Type II + GDPR, no-train-on-data, Trust Center. HIPAA = Enterprise only. No public ISO 27001. No advertised EU data residency (Supabase, likely US-centric).
13. Proactive/triggers — outbound campaigns (Standard+), voice outbound.
14. Multilingual — 80+ languages chat (95+ voice), auto-detect + translation.
15. Deployment — JS embed, no-code builder, WordPress, API components.
16. Packaging — flat monthly + model-weighted credits; cheap entry, premium-model + add-ons inflate TCO; single agent until Pro.

Self-host? EU residency? No self-host (Supabase cloud). No advertised EU residency. GDPR on paper only.

Top 3 differentiators: (1) Voice AI phone agents sharing KB+actions; (2) broadest user-selectable multi-LLM choice; (3) API-first + flat "no per-resolution" pricing. Top 3 gaps: (1) storage-capped KB; (2) credit model is a hidden 5× lever; (3) no A/B, no avatar, no EU residency, no self-host, shallow CRM.

2. Tidio (tidio.com) — incl. Lyro AI

Positioning + scale. Live-chat / helpdesk suite for SMBs with Lyro ("the first conversational AI for SMBs") bolted on (tidio.com/ai-agent↗). Helpdesk first, AI second. ~300,000 businesses worldwide, ~3% of Shopify merchants (techfundingnews↗). ~$48M revenue (2024), ~174 employees, $25M Series B (PeakSpan) (getlatka↗). Far larger install base than Chatbase, lower revenue-per-customer (freemium long tail).

Pricing — model: three parallel quotas (per-billable-conversation + per-Lyro-AI-conversation + per-Flows-visitor); seats flat (pricing page↗):

The $59 → $749 cliff is real — a 12× jump with nothing in between (chatarmin↗). Lyro AI is a separate add-on, billed per AI conversation: from $39/mo for 50 conversations, scaling to ~$79–$149/mo for 500–1,000 chats; ~$0.58–$0.78/conversation (fritz.ai↗). Free Lyro quota = 50 conversations LIFETIME, not per month — a deliberately misleading "free AI" framing (help.tidio.com↗). Flows is yet another add-on (from $29/mo for 2,000 visitors). Real AI deployment = base + Lyro + Flows ≈ $100–$250/mo.

Feature inventory:

1. LLM — Lyro on Anthropic Claude + Tidio in-house models. Single-vendor (Claude), not user-selectable. Native MCP support. Claims 67% resolution.
2. Knowledge/RAG — auto-scrapes KB/docs; website scan caps at ~60 pages (eesel.ai↗).
3. Omnichannel — Strong unified inbox: web, Messenger, Instagram, WhatsApp, email. Core strength. No Slack for end-users.
4. Voice — NONE. Lyro is text/inbound only. Clear loss vs Chatbase.
5. Avatar/embodiment — NONE.
6. Persona/brand — widget colors, on-brand tone, custom handoff. White-label is enterprise/agency only.
7. Lead/CRM/commerce — Strong e-commerce tilt: native Shopify/WooCommerce/WordPress (order/catalog access, product recs); Klaviyo, HubSpot, Salesforce, Pipedrive, Mailchimp. Best-in-class for Shopify stores.
8. Analytics — real-time live visitor tracking (online/pages/location), resolution rates, CSAT.
9. Human handoff — Yes — full shared inbox / live-chat heritage. Stronger than Chatbase.
10. A/B testing — YES. Built-in experiments on bot scripts/widget prompts.
11. Integrations/API — Shopify, WooCommerce, Messenger, Instagram, WhatsApp, Zendesk, HubSpot, Klaviyo, GA4; Zapier (9,000+ apps). Public API thinner than Chatbase.
12. Security — SOC 2 Type II (A-LIGN), GDPR + CCPA + EU-US DPF + EU AI Pact (security↗). EU data residency — data on EEA servers (advantage over Chatbase). No public ISO 27001, no HIPAA.
13. Proactive/triggers — Strong: time-on-page, exit-intent, location triggers; proactive live-chat via Flows.
14. Multilingual — Lyro: 45+ languages, auto-detect.
15. Deployment — JS snippet, native WordPress plugin (one of the most-installed), Shopify App Store app, drag-drop Flows builder.
16. Packaging — cheap base, three stacked quotas + $59→$749 cliff make true cost opaque/steep.

Self-host? EU residency? No self-host. YES, EU data residency (EEA servers) — a real advantage.

Top 3 differentiators: (1) mature live-chat + shared-inbox heritage + unified omnichannel; (2) e-commerce depth (best Shopify integration); (3) EU residency + SOC 2 + GDPR/DPF/AI-Pact + built-in A/B + proactive triggers. Top 3 gaps: (1) no voice, no avatar; (2) confusing stacked pricing + $59→$749 cliff + "lifetime-50 free AI" trick; (3) single-LLM (Claude only) + 60-page KB scan cap.

3. Voiceflow (voiceflow.com)

Positioning + scale. "Enterprise conversational AI / voice AI platform — collaboratively build AI agents." Voice/telephony heritage. ~250,000+ users/teams (Amazon, BMW, US Bank among Fortune 500); $38.5M total funding (latest $15M led by OpenView, ~$105M valuation; ARR figures vary by tracker) (voiceflow raise↗; BetaKit↗; Tracxn↗). Closed source.

Pricing — model: per-editor seat + usage credits (hybrid). Public pricing page now pushes "book a demo" for businesses (voiceflow.com/pricing↗); documented self-serve tiers (costbench↗; ringly.io↗):

Credit model: every interaction (chat, LLM call, voice minute, KB query) burns credits; self-serve converts credits→USD ÷200; API-heavy turns 5–10+ credits. Annual discount cut 20%→10%. Real 5-editor/50k-msg teams report ~$450–500/mo vs headline $150 (credits table↗).

Feature inventory:

1. LLM — Model-agnostic: OpenAI (GPT-4/4o-mini/3.5), Claude, Google, Groq; multi-LLM routing + fallbacks (Business).
2. Knowledge/RAG — built-in vector KB; PDF/text/URL-scrape; source caps tiered (~50 Free → 3,000 Pro → 10,000 Business).
3. Omnichannel — Web widget, Twilio SMS; WhatsApp/Messenger/Instagram via connectors (VoiceGlow/FlowBridge); Alexa/Google Assistant. Native messenger breadth thinner — some channels need third-party glue.
4. Voice — Core strength (heritage). Phone agents via Twilio, IVR, STT (Deepgram/AssemblyAI), TTS (ElevenLabs/Google), Krisp noise-cancel, answering-machine detection, ~1.4s latency. Strongest voice/telephony in the set.
5. Avatar/embodiment — NONE.
6. Persona/brand — system-prompt/persona config; white-label widget (Business/Enterprise).
7. Lead/CRM/commerce — Capture step; Salesforce, HubSpot, Zendesk, Shopify, Google Sheets, Airtable.
8. Analytics — transcripts + dashboards, rated "basic vs enterprise tools."
9. Human handoff — Weak/absent natively — "lacks built-in live-agent handoff," needs custom API/Zendesk. No native shared inbox.
10. A/B testing — Yes, via Logic step + experimental Test Platform + agent-to-agent testing (docs↗).
11. Integrations/API/SDK — Functions (reusable JS) + API blocks + webhooks + Zapier; public dialog-manager API; function/tool calling.
12. Security — ISO/IEC 27001:2022, SOC 2 (Type 1; Type 2 in progress), GDPR. HIPAA not advertised.
13. Proactive/triggers — Trigger step; proactive web-widget messaging (needs custom code).
14. Multilingual — ~19 languages (narrowest of the big builders).
15. Deployment — JS snippet + visual drag-drop flow builder; no-code with pro-code escape hatches. Cloud-only.
16. Packaging — per-seat + credits; cheap headline, expensive at scale; voice/credit costs spike.

Self-host? EU residency? No — cloud-only, closed-source. Enterprise offers "private cloud hosting" + custom residency contracting, but no customer self-host/on-prem.

Top 3 differentiators: (1) best-in-class voice/telephony stack; (2) strong collaborative visual builder + version history; (3) ISO 27001 + SOC 2 + Fortune-500 logos. Top 3 gaps: (1) no native human handoff / shared inbox; (2) cloud-only, no self-host, no residency control below Enterprise; (3) credit-pricing opacity (real cost ≫ headline), pricing now hidden behind sales.

4. Botpress (botpress.com)

Positioning + scale. "Enterprise-grade AI agent platform — the open-source hub to build & deploy LLM agents." ~3,000 paying customers, ~1M free users, 2,000+ devs joining daily (Kia, EA, Husqvarna); $25M Series B (June 2025) at $120M post-money (Framework VP, Deloitte + HubSpot Ventures), ~$45M total (BetaKit↗; Series B blog↗). GitHub `botpress/botpress` = 14,748 stars, MIT, active — but this repo is now the SDK + CLI + integrations + plugins, NOT a runnable server (github.com/botpress/botpress↗).

Pricing — model: flat base + pay-as-you-go AI Spend (usage-based, NO LLM markup). Updated May 2026; trackers conflict (botpress.com/pricing↗; pricing update↗; lindy↗):

AI Spend = the differentiator: LLM inference, embeddings, web search billed as variable "AI Spend" — passed through at provider cost with NO markup, bundled proportionally into your message allotment. Configurable spend caps. You pay roughly what you'd pay calling the LLM directly.

Feature inventory:

1. LLM — Model-agnostic, link your own provider via Hub; GPT-4/Claude/others; custom inference router.
2. Knowledge/RAG — strong: URLs, PDFs, docs, CSV/JSON, databases, site scrape; answers with citations; managed vector DB (100MB→5GB+).
3. Omnichannel — Broad/native: Web, WhatsApp, Messenger, Instagram, Telegram, Slack, MS Teams + more via Hub. Wider native coverage than Voiceflow.
4. Voice — supported (Enterprise-gated); secondary vs Voiceflow's telephony depth.
5. Avatar/embodiment — NONE.
6. Persona/brand — system-prompt/persona; white-label webchat + branding removal (Plus+).
7. Lead/CRM/commerce — HubSpot, Zendesk, Stripe, Linear + webchat forms; Shopify via Hub/Zapier.
8. Analytics — conversation analytics, transcripts, sentiment insights (Plus+), team analytics (Team+).
9. Human handoff — Yes — native (Plus+), team routing (Team+). Stronger HITL than Voiceflow.
10. A/B testing — Not advertised (absent).
11. Integrations/API/SDK — Botpress Hub marketplace + public API, TypeScript SDK, CLI, function/tool calling, webhooks, Zapier. Strongest developer/extensibility story in the set (the MIT repo IS this tooling).
12. Security — SOC 2, GDPR, encryption, RBAC, audit logs; data-residency at Enterprise. HIPAA/ISO 27001 not prominently advertised.
13. Proactive/triggers — event/webhook-driven proactive flows.
14. Multilingual — 100+ languages, auto-detect + integrated webchat translation (broadest in set).
15. Deployment — JS snippet + visual Studio builder + heavy pro-code (SDK/CLI). True no-code→pro-code spectrum.
16. Packaging — flat base + no-markup AI-spend (transparent variable cost); cheapest entry ($0 PAYG), steep base jumps.

Self-host? EU residency? — THE CRITICAL NUANCE. Botpress markets "open source" and the MIT botpress/botpress repo is active, but it contains the SDK/CLI/API-client/integrations — NOT a deployable server. The actual self-hostable runtime is legacy `botpress/v12` (separate repo, AGPL-3.0, 153 stars, last pushed April 2025), which Botpress has officially sunset — "no longer available for purchase, download, or new deployments; all new users should use Botpress Cloud" (v12 repo↗; v12 docs↗). In practice, new builds are Cloud-only. Genuine self-hosting means running frozen, unsupported v12 missing all post-2025 features. EU residency = Cloud Enterprise commitment, not customer-controlled on-prem. This is the single most important caveat: anyone choosing Botpress *because* it's "open-source/self-hostable" is choosing a deprecated path.

Top 3 differentiators: (1) no-markup pay-as-you-go AI-Spend transparency; (2) broadest native omnichannel + 100+ languages + native human handoff; (3) developer extensibility (Hub marketplace, MIT SDK/CLI, public API, $0 entry). Top 3 gaps: (1) "open-source/self-host" is largely a 2026 myth (real self-host = sunset AGPL v12); (2) steep base cliffs ($89→$495→$1,495) + separate AI/channel costs; (3) no A/B, no avatar, voice weaker than Voiceflow, HIPAA/ISO 27001 unclear.

5. Chatbot.com (ChatBot by Text / LiveChat)

Positioning + scale. "Help, convert, and sell with a data-driven AI chatbot" — SMB website chatbot, RAG-grounded. Parent: Text (formerly LiveChat Inc.), Polish public co. on the Warsaw Stock Exchange (LiveChat, HelpDesk, KnowledgeBase, OpenWidget). Now branded "ChatBot by Text" with a migration banner (chatbot.com↗). Scale is Text-wide: "Over 35,000 brands"; LiveChat/Text ~33K customers, ~$80.6M revenue (2024) (getlatka↗).

Pricing — model: per-seat monthly subscription with included AI resolutions (pricing↗):

Add-ons: extra resolutions $49.50/50-pack, extra API calls $29.50/100K. Pricing-model caveat: legacy standalone ChatBot still listed (Shopify) at $50–$499/mo per-valid-chat (~$1/100 extra chats) (Shopify↗) — the new unified Text pricing shifted from valid-chats to AI-resolutions.

Feature inventory:

1. LLM — Deliberately opaque. "Powered by Text AI"; "we handle AI providers like Anthropic, Google, OpenAI for you" — no public model disclosure (chatbot.com/ai↗). Newer LLM AI Agent + legacy deterministic Visual Builder (Stories) coexist.
2. Knowledge/RAG — Strong: website scan, files (PDF/DOCX/PPTX/TXT/CSV ≤50MB), Articles, KnowledgeBase scan; anti-hallucination grounding.
3. Omnichannel — Web, Messenger, Slack, LiveChat, Shopify, WordPress, Zendesk/Freshdesk; Wix/Squarespace/Webflow/BigCommerce. ABSENT: no native WhatsApp, Instagram, or SMS — a real 2026 gap.
4. Voice — NONE. Text-only.
5. Avatar/embodiment — NONE.
6. Persona/brand — custom role/tone/boundaries; widget styling; white-label only on Enterprise.
7. Lead/CRM/commerce — in-conversation lead capture; Shopify commerce (product search/recs, order lookup, product cards). No native CRM (webhook/Zapier out).
8. Analytics — Reports/Analytics + Reports API; chat/ticket summaries + topics (Growth). Moderate depth.
9. Human handoff — context-preserving transfer; native LiveChat escalation; ticketing. Shared-inbox layer lives in LiveChat/Text, not ChatBot standalone.
10. A/B testing — Yes — A/B Test action splits traffic across up to 5 flow variants (help↗).
11. Integrations/API — public REST API (api.chatbot.com), Chat Widget JS API, webhooks, Zapier. Curated first-party catalog, no large app store.
12. Security — Thin public posture. No /security page; GDPR consent tooling. SOC 2 Type 2 + SSO gated to Enterprise. ISO 27001, EU residency NOT stated.
13. Proactive/triggers — behavior/context-triggered greetings; metered campaigns (3 Essential / 30 Growth).
14. Multilingual — "most popular languages" + auto-detect, no published count; some features English-only.
15. Deployment — JS widget / GTM; native plugins (WordPress, Shopify, Wix, Squarespace, Webflow, BigCommerce); Messenger, Slack, LiveChat embed.
16. Packaging — per-seat ($19/$79/user annual) + AI-resolution metered + resolution add-on packs.

Self-host? EU residency? No — cloud-only (Text-hosted). No EU-residency option surfaced.

Top 3 differentiators: (1) built-in multi-variant A/B testing (rare at this tier); (2) strong grounded RAG with multi-format ingest; (3) mature Shopify commerce + clean dual AI-Agent/Visual-Builder paradigm. Top 3 gaps: (1) no native WhatsApp/Instagram/SMS; (2) no voice, no avatar, hidden LLM; (3) weak public security disclosure (no /security page, SOC 2 gated, no ISO/EU residency stated).

6. Intercom (intercom.com / fin.ai) — SMB tier + Fin

Positioning + scale. "The complete AI-first customer service platform" — the category-defining shared Inbox, now centered on the Fin autonomous AI agent. Intercom rebranded to "Fin" (May 2026) (SalesforceBen↗), and Salesforce signed to acquire it for ~$3.6B (June 15, 2026) citing Fin's ~30,000-company base and proprietary "Apex" model (Salesforce PR↗). Fin reports 40M+ resolved conversations, ~67% resolution rate (fin.ai/channels↗).

Pricing — two-axis: per-seat plans + per-outcome AI charges. Genuinely expensive once stacked.

Fin AI agent: $0.99 per outcome (once per conversation) — Resolution / Procedure-handoff / Disqualification $0.99; Qualification $9.99 (fin.ai/pricing↗). Standalone (Zendesk/Salesforce/HubSpot): 50-outcome/mo minimum (~$49.50). Add-ons: Copilot $29/agent/mo, Pro analytics $99/mo, Proactive Support Plus $99/mo; SMS/WhatsApp/phone/email PAYG. The real bill = seats + outcomes + add-ons + channels — it stacks fast.

Feature inventory:

1. LLM — Blended/evolving: OpenAI → Claude (Fin 2 "powered by Anthropic Claude") → now anchored on proprietary "Apex" ("world's first specialized customer-service LLM"). Bespoke RAG engine (retrieval/reranker/escalation component models); providers kept proprietary (Fin AI Engine↗).
2. Knowledge/RAG — Strong: Help Center, Snippets, past conversations, PDFs/docs, public URLs/sites, synced Zendesk articles, live data via Data Connectors, Content Targeting per segment.
3. Omnichannel — Excellent: web Messenger, email, voice/phone, WhatsApp, Instagram, Messenger, SMS, Slack, Discord, API — same AI/knowledge on every channel with context across handoffs (fin.ai/channels↗).
4. Voice — Yes — full product. Fin Voice 2 (Apex Flash): answers calls, resolves from KB, hands off with context; two-way audio, no IVR; inbound + proactive outbound; Twilio/Genesys/RingCentral/Five9/8x8/Aircall; ~28–30 languages (fin.ai/voice↗).
5. Avatar/embodiment — NONE. Name + static profile image only.
6. Persona/brand — custom name/avatar per brand; 5 tone presets; answer length; natural-language Fin Guidance per channel. No true white-label except via Fin Agent API.
7. Lead/CRM/commerce — Fin for Sales (CRM lead creation in Salesforce/HubSpot/Attio) + Fin for Ecommerce (deep Shopify catalog/order sync).
8. Analytics — Excellent/AI-native: automation rate, resolution rate, CX Score (rates every conversation without surveys), Optimize, Topics Explorer, Trends.
9. Human handoff/inbox — Core strength. Omnichannel shared Inbox; Fin + humans share one customer record; Copilot per-agent AI assist ($29/agent/mo, +31% conversations closed claim).
10. A/B testing — Yes — native A/B of outbound message variations (help↗).
11. Integrations/API — Large App Store (~565 apps per tracker); REST API v2.15; webhooks; iOS/Android/React-Native/Cordova SDKs; new Fin API platform (enterprise from ~$250k/yr).
12. Security — Excellent: SOC 2 Type II, ISO 27001:2022, ISO 27018, ISO 27701, ISO/IEC 42001:2023 (AI), GDPR, HIPAA, CCPA, HDS. Data residency: US, EU (Dublin), Australia (Sydney) — gated to Advanced/Expert annual; cross-region migration unsupported (Regional Data Hosting↗).
13. Proactive/triggers — Excellent: outbound in-product messages, push, banners, Product Tours, checklists, tooltips, Series (omnichannel journey builder), proactive voice.
14. Multilingual — 45+ languages with auto-detection (text); ~28–30 voice; language locked once per conversation.
15. Deployment — Web JS snippet (or NPM @intercom/messenger-js-sdk); native mobile SDKs; headless Fin Agent API.
16. Packaging — per-seat ($29/$85/$132) + $0.99/outcome + stacked add-ons + PAYG channels.

Self-host? EU residency? Cloud-only (AWS), no self-host. EU + AU data residency confirmed (gated to Advanced/Expert annual).

Top 3 differentiators: (1) Fin Voice 2 (true inbound+outbound phone AI); (2) best-in-class shared Inbox + Copilot HITL with one shared customer record; (3) proprietary Apex LLM + deep AI-native analytics + enterprise compliance (ISO 42001 + HIPAA + EU/AU residency). Top 3 gaps: (1) expensive + complex stacked pricing; (2) no avatar, cloud-only; (3) strategic uncertainty mid-Salesforce-acquisition.

7. Crisp (crisp.chat) — the EU-sovereignty benchmark

Positioning + scale. "The AI Customer Support Platform for Every Business" — all-in-one omnichannel shared-inbox + AI chatbot + KB + CRM. Crisp IM SAS, built and HQ'd in France 🇫🇷, branded "Made in Europe" (crisp.chat↗). Homepage: "Trusted by 10,000 companies"; broader install base ~400k+ (free-tier-inflated). EU-sovereign angle is Crisp's strongest strategic asset — a genuinely EU-hosted, French-incorporated platform positioned directly against US-hosted Intercom/Zendesk.

Pricing — flat per-workspace, NOT per-seat (the famous model) (pricing↗):

Pricing critique: "flat pricing" headline is undercut by metered AI — each tier ships a small credit bundle; heavy AI is consumption-priced beyond it. Essentials' ~450-convo allotment is "barely enough for light automation," pushing AI-serious buyers to $295 Plus (featurebase↗). Flat-on-conversations but metered-on-AI. (Old "Basic/Pro/Unlimited" tier names are outdated.)

Feature inventory:

1. LLM (MagicReply) — agent-assist: suggested answers, rephrase/expand, tone matching, configurable confidence (says "I don't know" + hands off). Model opaque — originally OpenAI, now proprietary in-house "Mirage" model; no longer names GPT/Claude (crisp.chat/ai↗).
2. Knowledge/RAG — strong: website crawl/scrape, KB auto-reindex, Answers Snippets (Q&A ≤1000 chars), conversation-history ingest, file upload (PDF/CSV).
3. Omnichannel — Web, email, WhatsApp, Messenger, Instagram DM, SMS, Line, Viber (+ Telegram in integrations). Omnichannel inbox unlocks at Essentials.
4. Voice — NONE. (LiveTranslate is text, not voice.)
5. Avatar/embodiment — NONE.
6. Persona/brand — chatbot personality via prompts; widget theming; white-label gated to Plus ($295).
7. Lead/CRM/commerce — built-in Support CRM (50K/200K profiles), contact forms, segmentation; Shopify + WordPress/Stripe in marketplace.
8. Analytics — from Essentials; "advanced analytics" on Plus.
9. Human handoff/inbox — ⭐ Core strength — inbox-first. Centralized omnichannel shared inbox is the spine; AI fallback/escalation + routing + sub-inboxes (2 Essentials / 5 Plus). Crisp's strongest category.
10. A/B testing — Not surfaced (treat as absent).
11. Integrations/API — strong: 100+ integrations, public Crisp Marketplace (third parties build/publish/sell plugins), REST API + Chat SDK + webhooks.
12. Security — EU residency (CRITICAL). ⭐⭐ Crisp's flagship advantage. "All Crisp data is held on servers hosted in the EU." Messaging data → Netherlands; plugin data → Germany. Explicitly "does not plan to store data outside the EU." GDPR-compliant (DPO, 24h breach notice, Privacy-by-Design). SOC2 referenced. Caveat: AI relay servers ("Mirage") sit in US/UK/Singapore for latency but store no data except connection logs (GDPR status↗). No self-host documented.
13. Proactive/triggers — chat triggers (Mini+); Campaigns module for proactive/marketing; unlimited automations (Plus).
14. Multilingual — LiveTranslate 50+ languages (some pages say 100+); chatbox UI localized in 60+ (realtime translation↗).
15. Deployment — JS snippet; mobile SDKs / Chat SDK; iOS/Android agent apps.
16. Packaging — flat per-workspace ($0/$45/$95/$295/custom) + seat bundles + metered AI credits layered on top.

Self-host? EU residency? No self-host (SaaS only). YES, EU residency genuinely — primary data NL+DE, EU-only commitment, French co. Only asterisk: AI relays (no data storage) touch US/UK/SG.

Top 3 differentiators: (1) EU data sovereignty + French/Made-in-Europe brand (strongest sovereign positioning in set); (2) flat per-workspace pricing (volume-independent); (3) inbox-first omnichannel + open marketplace/API/SDK. Top 3 gaps: (1) no voice, no avatar; (2) opaque metered AI (undisclosed model, small credit bundles, no A/B); (3) no self-host + AI relay leaves EU (asterisk for strictest buyers).

8. Short Notes — Landbot, ManyChat, Chatsimple

Landbot (landbot.io) — No-code visual drag-drop conversational flow builder for web + WhatsApp + Messenger, aimed at marketing teams/agencies (landbot.io↗).

• Pricing: Starter €40/mo (~$46; 500 chats/100 AI chats/2 seats), Pro €100/mo (2,500 chats/300 AI), WhatsApp €80–€200, Business €400+ (pricing↗).
• AI: hybrid — GPT-powered LLM nodes inside structured flows; AI chats metered + expensive (~€0.10 each beyond bundle).
• Differentiator: best-in-class visual flow editor for non-coders.
• No avatar, no voice. EU angle: Spanish company (Hello Umi S.L., Barcelona) — EU-incorporated but some sub-processors outside EEA; weaker residency story than Crisp.

ManyChat (manychat.com) — Instagram / WhatsApp / Messenger marketing & DM-automation leader (comment-to-DM funnels, broadcasts) (manychat.com↗).

• Pricing: scales by monthly active contacts — Free (1,000), Pro from $29/mo (2,500 contacts), Business $69+, Elite (enterprise). AI gated to Pro/Business; AI Step add-on ~$29/mo (AI-Pro ≈ $58/mo) + Meta WhatsApp fees (pricing↗).
• Channels: Instagram, TikTok, Messenger, WhatsApp, Telegram, SMS, Email.
• AI: "AI Steps" / AI Agent learns business + tone, auto-replies to DMs/comments.
• Differentiator: dominant for Instagram/social DM marketing automation.
• No avatar, no voice. US company — not an EU-residency play.

Chatsimple (chatsimple.ai → "Expertise AI") — AI sales agent for websites; qualifies inbound visitors in real time, deanonymizes traffic (name + company), books meetings (chatsimple.ai↗).

• Pricing: Free tier + paid from ~$500/mo (premium B2B-positioned) (saasworthy↗).
• Channels: Web, WhatsApp, Messenger, Instagram; 175+ languages.
• AI/differentiator: sales-focused agent — lead qualification, visitor identification, native CRM sync (HubSpot/Salesforce/Zoho/Calendly).
• Has VOICE (chat + voice agent — the only one of these three with voice). No avatar. SOC 2 Type II + GDPR but US-oriented, no EU-residency commitment.

9. Comparison Table — Platforms × 16 Categories

Legend: ✅ strong/native · 🟡 partial/gated/add-on · ❌ absent · — n/a

The single most striking row: Category 5 (Avatar/embodiment) is ❌ across all ten platforms. No competitor in the SMB widget category ships any embodied/animated avatar.

10. Where OpenAlice Wins / Must Catch Up

Where we WIN (defend and amplify)

1. Embodied 3D avatar — uncontested whitespace. Not a single competitor (0/10) ships an animated/talking-head avatar. The entire category is a text bubble (a few add phone voice). Our VRM-based embodied widget is a categorical differentiator, not an incremental one. This is the moat — lead every comparison with it.

1. Genuine EU self-host + sovereignty. Self-hosting is effectively dead across the field: Botpress's "open-source" self-host is a sunset AGPL v12 dead-end; everyone else is cloud-only. EU residency is rarer than expected — only Crisp (NL/DE), Tidio (EEA claim), and Intercom (EU/AU, gated) offer it, and none offer customer-controlled on-prem. OpenAlice self-host + EU-sovereign is a clean wedge against US-hosted incumbents AND against Crisp (whose AI relays leave the EU and who has no self-host escape hatch).

1. Honest analytics + A/B + transparent model/pricing. The field is full of opaque pricing tricks: Tidio's "lifetime-50 free AI," Chatbase's hidden 5×-by-model credits, Voiceflow hiding numbers behind sales, "flat" pricing that is actually metered-on-AI (Crisp). Model transparency is a shared weakness (Crisp/Chatbot.com hide which LLM; Intercom proprietary). A named-model, predictable-cost, honest-resolution-reporting story differentiates — and built-in A/B testing is missing from Chatbase, Botpress, Crisp, and Chatsimple.

Where we LAG (must catch up)

1. Omnichannel breadth + conversation continuity. Botpress (WA/IG/Telegram/Slack/Teams native), Intercom (every channel + one shared record), Tidio and Crisp (unified inboxes) all have deeper native channel coverage with context carried across handoffs. Our roadmap's live-inbox → WhatsApp → Shopify sequence is the right direction — but this is the biggest functional gap to close, and the most table-stakes.

1. Language count + auto-detect. Botpress 100+, Chatsimple 175+, Chatbase 80+ languages with auto-detection. We need a credible multilingual story (count + auto-detect) to compete outside English-first markets — currently a quantitative deficit.

1. Integration marketplace + ecosystem depth. Intercom (~565 apps), Botpress (Hub + MIT SDK/CLI), Crisp (100+ marketplace with third-party publishing), Tidio (Zapier 9,000+). A widget with few native integrations + no third-party marketplace looks thin to SMB buyers who live in Shopify/HubSpot/Zapier. Prioritize Shopify + HubSpot + Zapier natively, then a public API/SDK story.

Secondary catch-up watch-items

• Voice is escalating fast (Chatbase, Intercom/Fin, Voiceflow, Chatsimple all shipped phone/voice agents in 2026). An embodied voice avatar would fuse our moat (avatar) with the field's hottest 2026 feature — high-leverage combination.
• Human handoff / shared inbox is core for Tidio/Intercom/Crisp/Botpress; a credible HITL inbox is needed to be taken seriously as a support tool, not just a marketing widget.
• Compliance certifications (SOC 2, ISO 27001, ISO 42001) are increasingly table-stakes for mid-market — our EU-sovereign pitch needs the certs to back it.

One-line strategic frame

OpenAlice's wedge: "the only embodied AI widget that is also genuinely EU-sovereign and self-hostable, with honest pricing." Lead with the avatar (categorical, 0/10 competitors), back it with EU self-host (a wedge even against Crisp), and out-honest the field on pricing/analytics/A/B — while urgently closing omnichannel, language-count, and integration-marketplace gaps to reach table-stakes.

Sources cited inline throughout. Pricing verified against official pages June 2026; figures vary where vendors are mid-repricing (Botpress May-2026 update, Voiceflow credit model, Intercom/Fin rebrand + Salesforce acquisition) — treat exact dollar figures as ± and re-confirm on vendor pricing pages before quoting externally.