Skip to content
Z Zendikt
France edition · 10 products ranked · Verified 2026-05-17

Top 10 AI Agent Platforms in France for 2026

Independent French AI agent platform ranking, EUR pricing, EU AI Act timeline, CNIL and ANSSI guidance, and honest coverage of Mistral AI, Dust.

France verdict (TL;DR)

Verified 2026-05-17

France is the European epicentre of sovereign AI, and that shapes the agent platform market decisively. Mistral AI (Paris) is the strategic anchor: Mistral Agents (the developer API) and Le Chat (the enterprise assistant) mean French enterprise buyers now have a credible French-built alternative to OpenAI-based agent platforms. Dust (Paris, Sequoia-backed) is the leading French AI workflow and agent platform for enterprise teams. LangSmith and LangGraph remain dominant in French tech startups (Aircall, Spendesk, Doctolib-tier). The EU AI Act (prohibited practices effective February 2026; GPAI rules from August 2026) and CNIL guidance on AI are the regulatory frame. French government and CAC 40 companies have publicly stated preferences for Mistral-based agent platforms for sensitive workloads.

Picks for France

  • French tech startups and scale-ups building agents (sovereignty-neutral workloads): langsmith LangSmith is the observability and eval standard at French tech companies (Aircall, Spendesk, Mirakl, Contentsquare-tier). EU data residency (Frankfurt/Dublin) satisfies CNIL and EU AI Act data governance requirements. GDPR DPA available.
  • French enterprise and CAC 40 building agents on sovereign AI (Mistral-first): langgraph LangGraph supports Mistral models natively through LangChain integrations, making it the natural orchestration layer for French sovereignty-conscious enterprises that want to swap OpenAI for Mistral API without re-engineering their agent architecture.
  • French SMB and mid-market no-code automation: n8n n8n is self-hostable on French or EU-resident infrastructure (OVHcloud, Scaleway, AWS eu-west-3 Paris), which is a decisive factor for French companies prioritizing CNIL compliance and data sovereignty. Strong French community and reseller ecosystem.
  • French non-technical teams automating workflows: make Make has an established French user base and French-language community. EU-resident data processing available. Tiers starting at €9/month are accessible for French SMB.
  • French enterprise teams wanting a sovereign AI workflow platform: copilot-studio Microsoft has a strong CAC 40 footprint and a Paris data center. Copilot Studio with Azure France Central data residency is the default for French enterprises on M365 that need EU-sovereign agent workflows without building on US-only infra.
Market context

How the ai agent platforms market looks in France

France occupies a structurally distinct position in European AI agent platform adoption: the sovereignty argument is not rhetorical, it is commercially real. Mistral AI (founded 2023, Paris, Series B $640M at approximately $6B valuation as of 2024) has become the reference French AI company and its Mistral Agents developer API is a credible alternative to the OpenAI Assistants API for French developers building agentic systems. The French government (DINUM, Direction interministérielle du numérique) has publicly prioritized Mistral-based AI for sensitive government workloads. Several CAC 40 companies (Carrefour, Orange, BNP Paribas, Safran) have announced Mistral-based agent or assistant pilots.

Dust (Paris, founded 2022, Sequoia-backed) is the most important French-built AI agent platform for enterprise knowledge work. Dust positions as an AI workflow and agent platform for enterprise teams, with data connectors (Notion, Confluence, Google Drive, Salesforce), custom agent configurations, and an emphasis on team-level knowledge management. Dust is used by French scaleups and internationally (Qonto, Pennylane, Gorgias among disclosed customers). It is the closest French challenger to what Relevance AI or Lindy do in the US mid-market.

Linkup AI (Paris) and Light On (Paris) are adjacent infrastructure plays that embed in agent stacks: Linkup provides real-time web retrieval for agents; Light On (now Arcee AI France) built production-scale LLM adapters. Neither is a general agent platform but both appear in French-built agent architectures.

The EU AI Act timeline is now the dominant compliance variable for French enterprise agent buyers: prohibited AI practices (social scoring, real-time biometric surveillance) are banned from February 2026; GPAI transparency requirements apply from August 2026; high-risk AI system requirements (employment, credit, healthcare) are phased from 2026-2027. CNIL issued specific guidance on RGPD-compliant AI in 2024, including requirements for automated decision-making transparency and data minimization in AI systems.

Compliance & local rules

EU AI Act: prohibited practices effective February 2026; GPAI model rules from August 2026; high-risk system requirements phased 2026-2027. RGPD (French equivalent: CNIL enforcement) requires data protection impact assessments for high-risk AI processing, transparency in automated decisions, and data minimization. CNIL AI guidance (2024) specifies that AI systems processing personal data must document purpose limitation and legal basis; this directly applies to AI agents accessing employee or customer data. ANSSI (Agence nationale de la securite des systemes d'information) published AI security guidance in 2024; French government procurement of agent platforms requires ANSSI-compatible security practices. SecNumCloud certification (French sovereign cloud certification) is required for Restricted-level government data; only OVHcloud and Outscale hold SecNumCloud; agent platforms must run on these for the most sensitive French government workloads. ISO 27001 certification expected in French enterprise procurement for agent platform vendors.

At a glance

Quick comparison, ranked for France

Product Best for Starts at 10-emp/mo* Pricing G2 Geo
1 LangSmith
Engineering teams building production agents
$0 + $0/emp $0 4.5 Global; strongest in US, EU, UK
9 LangGraph
Engineering teams building production agents
$0 + $0/emp $0 4.5 Global
5 n8n
Engineering-led automation teams
$0 + $0/emp $0 4.7 Global; strongest in EU, US
10 Make (formerly Integromat)
Mid-market non-engineering teams
$0 + $0/emp $0 4.7 Global; strongest in EU, US
4 CrewAI
Engineering teams building multi-agent systems
$0 + $0/emp $0 4.6 Global
7 Microsoft Copilot Studio
Microsoft-anchored enterprises
$0 + $0/emp $0 4.3 Global; strongest in US, EU, UK; worldwide
3 Relevance AI
Mid-market with mix of ops and engineering
$0 + $0/emp $0 4.6 Global; strongest in US, AU, UK
6 Salesforce Agentforce
Salesforce-anchored enterprises
$2 $2 4.4 Global; strongest in US, EU, UK
2 Lindy
SMB ops teams
$0 + $0/emp $0 4.7 Global; strongest in US
8 ServiceNow AI Agents
ServiceNow-anchored enterprises
Quote - 4.4 Global; strongest in US, EU, UK

*10-employee monthly cost = base fee + (per-employee × 10) using the lowest published tier. For opaque-pricing vendors, no value is shown.

Verified local pricing

What buyers in France actually pay

Median annual deal size by employee band, in EUR. Crowdsourced from anonymized buyer disclosures.

Product Employee band Median annual (EUR) Sample Notes
LangSmith Developer (free) €0 67 Free tier; dominant usage tier at French startups
LangSmith Plus (small team) €1,700 28 Approx €1,560/seat/year; EUR-billed via card
n8n Self-hosted (free; OVHcloud or Scaleway) €0 98 Self-hosted on French-resident infra; zero license cost; dominant French SMB pattern
Make (formerly Integromat) Core €108 134 €9/month; EU-billed; widely used French SMB tier
Microsoft Copilot Studio M365 bundle (CAC 40 and French enterprise) €0 54 Bundled in M365 Copilot; Azure France Central data residency; standalone ~€180/tenant/month
Local challengers

France-built or France-strong vendors worth knowing

Not yet ranked in our global top 10, but credible options for France buyers and worth a shortlist.

Mistral AI (Mistral Agents)

Visit ↗

Paris-based foundation model company (founded 2023, Series B $640M, ~$6B valuation). Mistral Agents is the developer API for building agentic systems on top of Mistral models (Mistral Large, Mistral Small, Codestral). The French sovereignty anchor. Le Chat is the enterprise assistant product. Used by French government and CAC 40 for sensitive AI workloads.

Dust

Visit ↗

Paris-based AI workflow and agent platform for enterprise teams (founded 2022, Sequoia-backed). Data connectors to Notion, Confluence, Google Drive, Salesforce. Custom agent configs. French scaleup default (Qonto, Pennylane, Gorgias). The closest French-built answer to Relevance AI or Lindy for enterprise knowledge teams.

Linkup AI

Visit ↗

Paris-based retrieval API for AI agents providing real-time web search grounding. Embeds in LangGraph, CrewAI, and Mistral Agents workflows as the web-retrieval tool. Used by French and European agent builders needing grounded, up-to-date knowledge.

Excluded for France

Global picks that don't fit here

  • ServiceNow AI Agents
    ServiceNow has moderate French enterprise presence but is not in the top tier for French SMB or mid-market. Dust, n8n, and Mistral-native agent platforms are better fits for most French enterprise agent needs outside dedicated ServiceNow ITSM shops.
The France ranking

All 10, ranked for France

Same intelligence as the global ranking, vendor trust, review patterns, verified pricing, compliance, reordered for the France market.

#1

LangSmith

Industry-standard agent observability and evaluation platform.

Founded 2022 · San Francisco, CA · private · 5–10,000 employees
G2 4.5 (380)
Capterra 4.6
From $0 + $0 /mo + /employee
● Transparent pricing
Visit LangSmith

LangSmith is the observability and evaluation platform from LangChain (the company behind LangChain and LangGraph), founded 2022, last valued $1.25B+ (2024). The product covers agent tracing, evaluation, prompt management, and debugging, purpose-built for production agent deployments. Strengths: native LangChain/LangGraph integration, deepest agent tracing in category, mature evaluation framework (LangSmith Evals), and fastest product velocity. Best fit for engineering teams shipping agents to production. Trade-offs: pricing scales with traces ($0-$199+/seat plus per-trace usage), framework-agnostic mode less polished than native LangChain integration, and competition from purpose-built challengers (Braintrust, Helicone) is intensifying.

Best for

Engineering teams (5-500 engineers) building production agents using LangChain/LangGraph or any framework, prioritizing observability and evaluation depth.

Worst for

Non-engineering teams (Lindy/Relevance AI better fit), Salesforce-anchored enterprises (Agentforce native better), or buyers wanting fully managed agent platform without engineering effort.

Strengths

  • Industry-standard agent observability
  • Native LangChain/LangGraph integration
  • Deepest agent tracing in category
  • Mature LangSmith Evals framework
  • Fastest product velocity
  • Fits production agent deployments

Weaknesses

  • Pricing scales with traces (per-usage costs)
  • Framework-agnostic mode less polished
  • Competition intensifying from Braintrust/Helicone
  • Per-seat scaling adds up at higher tiers
  • Some enterprise governance features still maturing

Pricing tiers

public
  • Developer
    5K traces/month free
    $0+$0 /mo +/emp
  • Plus
    Per seat; 10K traces/seat
    $39 /mo
  • Enterprise
    Custom; SSO, SOC 2, dedicated support
    Quote
Watch for
  • · Per-trace overage costs scale fast
  • · Higher-volume orgs hit per-seat ceilings quickly
  • · Custom evaluators consume LLM API budget

Key features

  • +Agent tracing and debugging
  • +LangSmith Evals (eval framework)
  • +Prompt management and versioning
  • +Production monitoring
  • +Native LangChain/LangGraph integration
  • +Framework-agnostic SDK
  • +Annotation queues for human feedback
80+ integrations
LangChainLangGraphOpenAIAnthropicPineconeWeaviate
Geography
Global; strongest in US, EU, UK
#9

LangGraph

Open-source agent orchestration framework from LangChain.

Founded 2024 · San Francisco, CA · private · 5–500 employees
G2 4.5 (180)
Capterra 4.5
From $0 + $0 /mo + /employee
● Transparent pricing
Visit LangGraph

LangGraph is the agent orchestration framework from LangChain, launched 2024. The product is a graph-based framework for building stateful, multi-actor agents, designed for production-grade agent systems where reliability and observability matter. Strengths: graph-based stateful agent architecture, native LangSmith integration, LangGraph Cloud for managed deployment, and strong fit for production agent systems. Best fit for engineering teams building production agents with non-trivial state and control flow. Trade-offs: framework-only requires engineering effort, learning curve steeper than CrewAI, and observability requires LangSmith (not bundled).

Best for

Engineering teams (5-200 engineers) building production agent systems with non-trivial state, control flow, and observability requirements.

Worst for

Non-engineering teams (Lindy/Relevance AI better fit), pure single-agent use cases (LangChain or OpenAI Agents simpler), or buyers wanting fully managed without engineering.

Strengths

  • Graph-based stateful agent architecture
  • Native LangSmith integration
  • LangGraph Cloud for managed deployment
  • Built for production agent systems
  • Open-source (MIT)
  • Time-travel debugging

Weaknesses

  • Framework-only requires engineering effort
  • Learning curve steeper than CrewAI
  • Observability requires LangSmith
  • Documentation gaps in advanced features
  • Less mature than LangChain

Pricing tiers

public
  • Open Source
    Self-hosted; bring your own LLM
    $0+$0 /mo +/emp
  • LangGraph Cloud
    Managed deployment; bundled with LangSmith
    Quote
  • Enterprise
    Custom; SSO + governance
    Quote
Watch for
  • · LLM API costs separate
  • · LangSmith for observability
  • · LangGraph Cloud usage scales

Key features

  • +Graph-based agent orchestration
  • +Stateful multi-actor agents
  • +Native LangSmith integration
  • +LangGraph Cloud (managed deployment)
  • +Time-travel debugging
  • +Human-in-the-loop primitives
  • +Streaming + interruption support
80+ integrations
LangChainLangSmithOpenAIAnthropicPineconePostgres
Geography
Global
#5

n8n

Open-source workflow automation with AI nodes, self-hostable.

Founded 2019 · Berlin, Germany · private · 10–10,000 employees
G2 4.7 (480)
Capterra 4.7
From $0 + $0 /mo + /employee
● Transparent pricing
Visit n8n

n8n is the open-source workflow automation platform with AI nodes, founded 2019 in Berlin. The product is a fair-code-licensed automation tool (similar to Zapier/Make but self-hostable) that added agent capabilities through AI nodes 2024-2025. Strengths: self-hostable open-source (huge for regulated industries), 400+ integrations, AI agent nodes for adding LLMs into workflows, and engineering-friendly. Best fit for engineering-led automation teams wanting self-hosted control. Trade-offs: not a pure agent platform (it's workflow automation with AI added), no-code experience less polished than Lindy, and SaaS pricing at scale comparable to commercial alternatives.

Best for

Engineering-led automation teams (10-1,000 employees) wanting self-hosted control, large integration footprint, and AI nodes within workflow automation.

Worst for

Pure no-code SMBs (Lindy better fit), pure agent engineering (CrewAI/LangGraph better), or buyers wanting fully managed without DevOps.

Strengths

  • Self-hostable open-source (fair-code license)
  • 400+ integrations
  • AI agent nodes for LLM workflows
  • Engineering-friendly (custom nodes)
  • Right call for regulated industries
  • Active community

Weaknesses

  • Not a pure agent platform (workflow + AI)
  • No-code UX less polished than Lindy
  • SaaS pricing at scale comparable to commercial
  • Documentation gaps for AI features
  • Customer support on community tier only

Pricing tiers

public
  • Self-hosted Community
    Open-source; unlimited
    $0+$0 /mo +/emp
  • Cloud Starter
    5K executions/month
    $24 /mo
  • Cloud Pro
    10K executions/month
    $60 /mo
  • Enterprise
    Custom; self-hosted + SSO
    Quote
Watch for
  • · Self-hosted infra costs
  • · Per-execution overage on Cloud
  • · AI LLM API costs separate

Key features

  • +Workflow automation
  • +AI agent nodes
  • +400+ integrations
  • +Custom nodes (JavaScript)
  • +Self-hostable open-source
  • +Cloud SaaS option
  • +Fair-code license
400+ integrations
SalesforceHubSpotSlackOpenAIAnthropicPostgreSQLAWS
Geography
Global; strongest in EU, US
#10

Make (formerly Integromat)

Mid-market no-code workflow automation with AI agent nodes.

Founded 2012 · Prague, Czech Republic · private · 10–10,000 employees
G2 4.7 (480)
Capterra 4.8
From $0 + $0 /mo + /employee
● Transparent pricing
Visit Make (formerly Integromat)

Make (formerly Integromat) is the mid-market no-code workflow automation platform, founded 2012, acquired by Celonis in 2020 and rebranded to Make in 2022. The product is similar to Zapier but with more visual flexibility and better mid-market pricing, and added AI agent capabilities through OpenAI/Anthropic nodes 2023-2024. Strengths: visual scenario builder, 1,500+ integrations, AI nodes for adding LLMs into workflows, and competitive pricing. Best fit for mid-market non-engineering teams extending existing automation. Trade-offs: not a pure agent platform, AI capability ceiling lower than dedicated agent tools, and post-Celonis direction has been measured.

Best for

Mid-market non-engineering teams (50-1,000 employees) extending existing workflow automation with AI nodes, primarily Zapier alternatives at higher complexity.

Worst for

Pure agent platform needs (LangSmith/CrewAI/Lindy better), engineering-led automation (n8n better), or buyers wanting deepest AI integration.

Strengths

  • Visual scenario builder
  • 1,500+ integrations
  • AI nodes (OpenAI, Anthropic)
  • Competitive mid-market pricing
  • Established 12+ year track record
  • Best for non-engineers

Weaknesses

  • Not a pure agent platform
  • AI capability ceiling lower than dedicated tools
  • Post-Celonis product velocity measured
  • Some complexity in advanced scenarios
  • Support response times vary

Pricing tiers

public
  • Free
    1,000 ops/month
    $0+$0 /mo +/emp
  • Core
    10K ops/month
    $9 /mo
  • Pro
    10K ops/month + advanced
    $16 /mo
  • Teams
    Per user; multi-user
    $29 /mo
  • Enterprise
    Custom; advanced features
    Quote
Watch for
  • · Per-operation overage costs
  • · AI node usage scales with LLM API costs
  • · Some advanced features gated

Key features

  • +Visual scenario builder
  • +1,500+ integrations
  • +AI nodes (OpenAI, Anthropic, etc.)
  • +Custom apps (developer SDK)
  • +Webhooks and APIs
  • +Mobile apps
  • +Schedule and event-based triggers
1500+ integrations
SalesforceHubSpotSlackOpenAIAnthropicGoogle Workspace
Geography
Global; strongest in EU, US
#4

CrewAI

Open-source multi-agent orchestration framework.

Founded 2023 · San Francisco, CA · private · 5–500 employees
G2 4.6 (180)
Capterra 4.5
From $0 + $0 /mo + /employee
● Transparent pricing
Visit CrewAI

CrewAI is the open-source multi-agent orchestration framework, founded 2023 by João Moura. The product is a Python framework for building agent crews, multiple agents that collaborate on complex tasks with role-based responsibilities. Strengths: clean role-based agent abstractions (researcher, writer, reviewer), open-source with active community, strong fit for engineering teams building multi-agent systems, and CrewAI Plus for managed deployment. Trade-offs: framework-only approach requires engineering effort, Python-only, and observability requires LangSmith or similar (not bundled).

Best for

Engineering teams (5-100 engineers) building multi-agent systems with role-based responsibilities, comfortable with Python framework adoption.

Worst for

Non-engineering teams (Lindy/Relevance AI better fit), single-agent use cases (LangChain/LangGraph more flexible), or non-Python shops.

Strengths

  • Open-source (Apache 2.0)
  • Clean role-based agent abstractions
  • Made for multi-agent systems
  • Active community
  • CrewAI Plus for managed deployment
  • Founder-led; strong product velocity

Weaknesses

  • Framework-only requires engineering effort
  • Python-only
  • Observability requires separate tool (LangSmith)
  • Documentation gaps in advanced features
  • Smaller community than LangChain

Pricing tiers

public
  • Open Source
    Self-hosted; bring your own LLM
    $0+$0 /mo +/emp
  • CrewAI Plus
    Per workspace; managed deployment
    $99 /mo
  • Enterprise
    Custom; SSO + governance
    Quote
Watch for
  • · LLM API costs separate
  • · Production deployment infra costs

Key features

  • +Multi-agent orchestration
  • +Role-based agent abstractions
  • +Tool use
  • +Agent collaboration patterns
  • +Open-source framework
  • +CrewAI Studio for visual building
  • +CrewAI Plus for managed deployment
50+ integrations
OpenAIAnthropicLangChainOllamaGitHub
Geography
Global
#7

Microsoft Copilot Studio

Native Microsoft 365 / Power Platform agent builder.

Founded 2023 · Redmond, WA · public · 500–500,000+ employees
G2 4.3 (1,280)
Capterra 4.4
From $0 + $0 /mo + /employee
● Transparent pricing
Visit Microsoft Copilot Studio

Microsoft Copilot Studio is Microsoft's low-code agent builder, founded 2023 (rebranded from Power Virtual Agents). The product covers building AI agents for Microsoft 365, Teams, and Power Platform integration. Strengths: bundled-or-near-bundled with Microsoft 365 enterprise contracts, native Microsoft Graph integration (Teams, SharePoint, Outlook), and strong fit for Microsoft-anchored enterprises. Trade-offs: outside Microsoft ecosystem the product is meaningfully weaker, agent capability ceiling lower than dev SDKs (Copilot Studio is great for predictable Microsoft workflows; less great for novel multi-step tasks), and pricing creates surprise costs (per-message billing).

Best for

Microsoft 365 enterprise customers (1,000+ employees) extending Microsoft 365 with low-code agents, particularly Teams, Outlook, and SharePoint workflows.

Worst for

Non-Microsoft shops (LangSmith/Lindy better), engineering-led agent building (LangSmith/CrewAI better), or buyers needing deep multi-agent orchestration.

Strengths

  • Bundled or near-bundled with Microsoft 365 enterprise
  • Native Microsoft Graph integration
  • Fits Microsoft-anchored enterprises
  • Power Platform integration
  • Public company financial transparency
  • FedRAMP High authorized

Weaknesses

  • Outside Microsoft ecosystem meaningfully weaker
  • Agent capability ceiling lower than dev SDKs
  • Per-message pricing creates surprise costs
  • Customer support varies by region
  • Innovation pace below Salesforce Agentforce

Pricing tiers

public
  • Pay-as-you-go
    Per-message billing; ~$0.01/message
    $0+$0 /mo +/emp
  • Copilot Studio Pro
    Per tenant; 25K messages/month
    $200 /mo
  • M365 Bundled
    Bundled with M365 E5 (some features)
    Quote
Watch for
  • · Per-message costs scale with usage
  • · Premium features in higher tiers
  • · M365 license required

Key features

  • +Low-code agent builder
  • +Native Microsoft Graph integration
  • +Power Platform integration
  • +Teams + Outlook + SharePoint workflows
  • +Voice + chat channels
  • +Pre-built templates
  • +Mobile apps
300+ integrations
Microsoft 365Microsoft TeamsMicrosoft GraphSharePointOutlookPower Platform
Geography
Global; strongest in US, EU, UK; worldwide
#3

Relevance AI

Mid-market low-code agent platform with strong pre-built skills.

Founded 2020 · San Francisco, CA · private · 50–500 employees
G2 4.6 (240)
Capterra 4.6
From $0 + $0 /mo + /employee
● Transparent pricing
Visit Relevance AI

Relevance AI is the mid-market low-code AI agent platform, founded 2020. The product covers agent building, deployment, and management with a low-code interface that engineering teams and ops teams both use. Strengths: strong pre-built skills library, mid-market sweet spot ($199-$1,499/month), and tight CRM integrations. Best fit for mid-market companies (50-500 employees) wanting low-code agents with engineering escape hatches. Trade-offs: not the right fit for pure no-code SMB use (Lindy better) or pure dev-SDK engineering (LangSmith + framework better), and product velocity has been mixed in 2024-2025.

Best for

Mid-market companies (50-500 employees) wanting low-code agent platform with engineering escape hatches and strong CRM integrations.

Worst for

Pure no-code SMBs (Lindy better fit), pure engineering teams (LangSmith + framework better), or enterprise needing deep governance (Agentforce/ServiceNow better).

Strengths

  • Strong pre-built skills library
  • Mid-market sweet spot ($199-$1,499/month)
  • Low-code with engineering escape hatches
  • Tight CRM integrations (Salesforce, HubSpot)
  • Multi-agent orchestration
  • Founder-led

Weaknesses

  • Not the right fit for pure no-code SMB
  • Product velocity mixed 2024-2025
  • Support is hit-or-miss
  • Smaller community than LangSmith
  • Some integration gaps vs Lindy

Pricing tiers

public
  • Free
    100 credits/month
    $0+$0 /mo +/emp
  • Pro
    10K credits/month
    $199 /mo
  • Team
    50K credits/month
    $599 /mo
  • Business
    200K credits/month
    $1499 /mo
  • Enterprise
    Custom; SSO + governance
    Quote
Watch for
  • · Credit overage costs
  • · Some advanced integrations gated
  • · Per-seat pricing at higher tiers

Key features

  • +Low-code agent builder
  • +Multi-agent orchestration
  • +Pre-built skills library
  • +Custom skills (Python/JavaScript)
  • +CRM integrations native
  • +Workflow triggers
  • +Agent monitoring
80+ integrations
SalesforceHubSpotSlackNotionGoogle WorkspaceOpenAIAnthropic
Geography
Global; strongest in US, AU, UK
#6

Salesforce Agentforce

Native Salesforce agent platform extending CRM with AI agents.

Founded 2024 · San Francisco, CA · public · 1,000–500,000+ employees
G2 4.4 (540)
Capterra 4.5
From $2 /mo
● Transparent pricing
Visit Salesforce Agentforce

Salesforce Agentforce is Salesforce's native AI agent platform, launched 2024 (originally branded "Einstein Agentforce"). The product extends Salesforce CRM with AI agents for sales, service, marketing, and commerce. Strengths: native Salesforce data and metadata access (no data movement required), default for Salesforce-anchored enterprises, Atlas Reasoning Engine for agent planning, and aggressive roadmap. Trade-offs: outside Salesforce ecosystem the product is irrelevant, pricing meaningful (per-conversation $2 + per-Agentforce-seat tiers), and many capabilities still maturing as of 2026.

Best for

Salesforce-anchored enterprises (1,000-50,000 employees) extending CRM with AI agents, particularly customer service, sales, and marketing agents on Salesforce data.

Worst for

Non-Salesforce shops (LangSmith/CrewAI/Lindy better), engineering-led agent building (LangSmith/CrewAI better), or value-driven mid-market.

Strengths

  • Native Salesforce data access
  • Default for Salesforce-anchored enterprises
  • Atlas Reasoning Engine
  • Strong customer service agent use case
  • Public company financial transparency
  • Massive Salesforce sales motion

Weaknesses

  • Outside Salesforce ecosystem irrelevant
  • Pricing meaningful (per-conversation $2 + seats)
  • Many capabilities still maturing
  • Implementation heavy for non-trivial agents
  • Support depends on tier

Pricing tiers

public
  • Agentforce
    Per conversation; consumption-based
    $2 /mo
  • Agentforce Studio (build)
    Per seat; agent builder access
    Quote
  • Enterprise Bundle
    Custom; bundled with Salesforce Sales/Service Cloud
    Quote
Watch for
  • · Per-conversation costs scale with usage
  • · Salesforce platform license required
  • · Implementation services

Key features

  • +Native Salesforce data integration
  • +Agentforce Studio (low-code agent builder)
  • +Atlas Reasoning Engine
  • +Pre-built agent templates (sales, service, etc.)
  • +Salesforce Data Cloud integration
  • +Voice + chat channels
  • +Mobile apps
200+ integrations
Salesforce Sales CloudSalesforce Service CloudSalesforce Marketing CloudSalesforce Data CloudSlackOpenAI
Geography
Global; strongest in US, EU, UK
#2

Lindy

No-code AI agent builder for SMB ops automation.

Founded 2022 · San Francisco, CA · private · 10–500 employees
G2 4.7 (280)
Capterra 4.7
From $0 + $0 /mo + /employee
● Transparent pricing
Visit Lindy

Lindy is the no-code AI agent platform for SMB ops automation, founded 2022 by Flo Crivello. The product lets non-engineers build agents that handle scheduling, inbox management, lead routing, and recurring ops tasks. Strengths: no-code agent building (no SDKs required), strong fit for SMB ops automation, modern UX, and aggressive product velocity. Best fit for SMBs (10-200 employees) wanting agent-driven workflow automation without engineering effort. Trade-offs: agent capability ceiling lower than dev SDKs (Lindy is great for predictable workflows; less great for novel multi-step tasks), pricing per-task can scale unpredictably, and enterprise governance still maturing.

Best for

SMB ops teams (10-200 employees) wanting no-code agent automation for predictable workflows (scheduling, inbox, lead routing, recurring tasks).

Worst for

Engineering teams building novel agents (LangSmith + dev SDKs better), enterprise needing deep governance (Salesforce/ServiceNow agents better), or pure observability use case.

Strengths

  • No-code agent builder (no SDKs required)
  • Works for SMB ops automation
  • Modern UX
  • Aggressive product velocity
  • Pre-built skills for common workflows
  • Founder-led; strong VC backing

Weaknesses

  • Agent capability ceiling lower than dev SDKs
  • Pricing per-task scales unpredictably
  • Enterprise governance still maturing
  • Support response times vary
  • Less suitable for novel multi-step tasks

Pricing tiers

public
  • Free
    400 tasks/month
    $0+$0 /mo +/emp
  • Pro
    5,000 tasks/month
    $49 /mo
  • Business
    30,000 tasks/month
    $199 /mo
  • Enterprise
    Custom; SSO + governance
    Quote
Watch for
  • · Per-task overage costs
  • · Some advanced integrations gated to higher tiers
  • · Per-seat pricing on Business+

Key features

  • +No-code agent builder
  • +Pre-built skills (scheduling, email, lead routing)
  • +Calendar integration
  • +Email integration (Gmail, Outlook)
  • +CRM integration (Salesforce, HubSpot)
  • +Agent triggers (scheduled, event-based)
  • +Mobile apps
60+ integrations
GmailOutlookSalesforceHubSpotSlackZoomCalendly
Geography
Global; strongest in US
#8

ServiceNow AI Agents

Native ServiceNow agent platform for enterprise ITSM workflows.

Founded 2024 · Santa Clara, CA · public · 5,000–500,000+ employees
G2 4.4 (380)
Capterra 4.5
Custom quote
○ Sales call required
Visit ServiceNow AI Agents

ServiceNow AI Agents (part of the Now Assist family) is ServiceNow's native AI agent platform. The product extends ServiceNow ITSM, ITOM, HRSD, and CSM with AI agents for incident routing, ticket triage, and workflow automation. Strengths: native ServiceNow data and workflow access, default for ServiceNow-anchored enterprises (10,000+ employees), and tight integration with the ServiceNow data graph. Trade-offs: outside ServiceNow ecosystem the product is irrelevant, pricing meaningful (Now Assist licenses run $20K-$1M+/year), and capability still catching up to Salesforce Agentforce in non-ITSM use cases.

Best for

ServiceNow-anchored enterprises (5,000-100,000 employees) extending ITSM, ITOM, HRSD with AI agents, particularly incident routing, ticket triage, and approval workflows.

Worst for

Non-ServiceNow shops (LangSmith/CrewAI better), customer-facing AI agents (Salesforce Agentforce better), or mid-market wanting lower TCO (overpriced).

Strengths

  • Native ServiceNow data integration
  • Default for ServiceNow-anchored enterprises
  • Strong ITSM agent use case (incident, ticket triage)
  • Tight ServiceNow data graph access
  • Public company financial transparency
  • Mature enterprise compliance

Weaknesses

  • Outside ServiceNow ecosystem irrelevant
  • Pricing meaningful (Now Assist $20K-$1M+/year)
  • Capability catching up to Agentforce in non-ITSM
  • Implementation heavy
  • Support inconsistency reported

Pricing tiers

opaque
  • Now Assist (Pro Plus)
    ~$20K-$200K/year typical
    Quote
  • Now Assist (Enterprise)
    $200K-$1M+/year for large enterprises
    Quote
Watch for
  • · Bundled with ServiceNow platform license
  • · Per-employee scaling
  • · Implementation services

Key features

  • +Native ServiceNow data integration
  • +AI Agent Studio
  • +Pre-built ITSM, ITOM, HRSD agents
  • +Now LLM (proprietary)
  • +Multi-channel (email, Teams, Slack)
  • +Mobile apps
200+ integrations
ServiceNow ITSMServiceNow ITOMServiceNow HRSDMicrosoft TeamsSlackOpenAI
Geography
Global; strongest in US, EU, UK

Frequently asked questions

The questions buyers actually ask before they sign.

Should French enterprise buyers prioritize Mistral-based agent platforms over LangSmith/LangGraph?
For sensitive workloads involving French citizen data, government-classified information, or CAC 40 proprietary IP, the sovereignty argument for Mistral is commercially valid: Mistral is a French company, subject to French and EU law, with data processed in European infrastructure. LangSmith and LangGraph (LangChain, Inc., San Francisco) process data on US infrastructure by default; EU data residency is available but the parent company is US-incorporated. For most French tech startups running non-sensitive SaaS workloads, LangSmith and LangGraph remain the technically superior choice because of ecosystem depth, tooling maturity, and developer community. The practical answer: use LangGraph as the orchestration layer but configure it to call Mistral API instead of OpenAI, satisfying sovereignty at the model level while keeping engineering tooling advantages.
What is the EU AI Act impact on agent platforms in France from 2026?
From February 2026: prohibited AI practices are banned (real-time remote biometric surveillance in public, social scoring, subliminal manipulation). No agent platform in this ranking does any of these; procurement impact is minimal for standard enterprise use cases. From August 2026: GPAI transparency rules require model providers (Mistral, OpenAI, Anthropic) to publish summaries of training data and comply with copyright. From 2026-2027: high-risk AI system requirements (Annex III: employment, education, credit, healthcare) require conformity assessments, technical documentation, and registration in the EU database. Agent platforms used in hiring, credit scoring, or medical triage in France will need this documentation. Most agent platform vendors (LangSmith, Agentforce, Copilot Studio) are preparing GPAI documentation; verify per vendor before signing high-risk contracts.
LangSmith vs CrewAI vs Lindy, which is right for me?
LangSmith is observability, use it alongside any agent framework when shipping to production. CrewAI is a multi-agent framework, use when you're building agent crews from code in Python. Lindy is a no-code agent builder, use when non-engineers need to build agents for predictable workflows. Most serious AI engineering teams use LangSmith + a framework (LangGraph, CrewAI, OpenAI Agents SDK). Most ops teams use Lindy or Relevance AI. The categories are complementary, not competitive.
How does this differ from AI Coding Assistants?
AI coding assistants (Top 10 AI Coding Assistants) help engineers write code, Cursor, Copilot, Claude Code. AI agent platforms (this ranking) are infrastructure for building production agent systems. Some overlap: Claude Code uses MCP servers and the Claude Agent SDK to build agents. But the buyer journey is different, coding assistants serve developers writing code; agent platforms serve teams shipping autonomous AI to production.
How much should I budget for AI agent platforms?
Solo developer: $0-$100/mo (LangSmith Developer free, CrewAI/n8n/LangGraph open-source). Small engineering team (5-25 engineers): $200-$2,000/mo (LangSmith Plus, CrewAI Plus, Relevance AI Pro). Mid-market (25-200 employees): $2K-$20K/mo (LangSmith Enterprise, Relevance AI Business, Lindy Business). Enterprise (1,000+ employees): $50K-$1M+/year (Salesforce Agentforce, ServiceNow AI Agents, LangSmith Enterprise). LLM API costs are separate everywhere.
Do I need observability for agents?
Yes. Building agents without observability is operationally negligent in production. Agents make non-deterministic decisions; without traces and evals you cannot debug failures or prevent regressions. LangSmith is the default; Braintrust and Helicone are credible alternatives. Even open-source agent frameworks (CrewAI, LangGraph) work with paid observability tools. Budget for observability separately from the agent framework.
When should I use a dev SDK vs a no-code platform?
Dev SDK (LangGraph, CrewAI, OpenAI Agents SDK, Anthropic Agent SDK) when: (1) you have engineering capacity, (2) the workflow is novel or complex, (3) you need fine-grained control. No-code platform (Lindy, Relevance AI, Make, n8n) when: (1) the workflow is predictable, (2) ops teams need to own the agent, (3) speed-to-deploy matters more than capability ceiling.
How does Salesforce Agentforce compare to LangChain?
Different categories. Agentforce is a Salesforce-native enterprise agent platform, designed for Salesforce-anchored enterprises extending CRM with AI. LangChain/LangGraph is a developer framework, designed for engineering teams building production agents from code. Agentforce is opinionated and Salesforce-centric; LangChain is unopinionated and infrastructure-agnostic. For Salesforce shops, Agentforce is the natural choice; for non-Salesforce engineering teams, LangChain.
Can I evaluate agent platforms via free trial?
Open-source / free-tier: LangSmith Developer (free 5K traces), CrewAI (free), n8n (free self-hosted), LangGraph (free open-source), Lindy Free, Relevance AI Free, Make Free, Microsoft Copilot Studio (free PAYG). Paid trials: Lindy Pro/Business, Relevance AI Pro, Make Pro/Teams, n8n Cloud. Demo only: Salesforce Agentforce, ServiceNow AI Agents.
How does this overlap with AI SDR and chatbots?
AI SDR (Top 10 AI SDR Software) is purpose-built finished products, autonomous SDRs replacing human SDR work. AI agent platforms (this ranking) are infrastructure for building agents. Some AI SDRs are built on agent platforms (e.g. several AI SDRs use LangChain under the hood). AI Chatbots (forthcoming ranking) are similar, finished products vs platform tooling.

Final word

Looking at a different market? See the global AI Agent Platforms ranking, or pick another country at the top of this page.

Last updated 2026-05-17. Local pricing reverified quarterly. Found something inaccurate? Tell us.