CtrlK
Lindy is a no-code AI agent platform that lets businesses and individuals build autonomous agents to automate workflows across sales, customer support, operations, and scheduling. Founded in 2023 by Flo Crivello and based in San Francisco, the platform allows users to connect agents to over 2,500 third-party applications through a visual workflow builder without writing code. Lindy is used by more than 400,000 professionals and reached $5.1 million in annual revenue by the end of 2024 with a team of roughly 37 employees.
The platform sits at an interesting point in the automation market: more capable than traditional tools like Zapier for tasks requiring judgment, but more accessible than developer-oriented platforms like n8n. Users build agents by defining triggers (an incoming email, a new CRM entry, a scheduled time), attaching action nodes, and connecting to their apps. Once deployed, agents run without supervision.
Flo Crivello grew up in France and moved to the United States after reading Atlas Shrugged, arriving with limited funds, no professional network, and imperfect English. He eventually joined Uber as a product manager in 2015, working there before leaving to start his own company.
In 2020, Crivello founded Teamflow, a spatial video conferencing product designed to replicate the feel of a shared physical office for remote teams. Teamflow raised approximately $52 million across multiple rounds, including an $11 million Series A led by Battery Ventures in November 2021 and a $35 million Series B led by Coatue Management in July 2021, with Battery Ventures and Menlo Ventures also participating. The product grew sharply during the COVID-19 pandemic as companies scrambled for remote collaboration tools.
The problem was that Teamflow's growth was tied directly to pandemic-driven remote work. As employees returned to offices starting in 2022, growth slowed significantly. By mid-2022, Crivello was running a 50-person company with a product whose core market was contracting.
The pivot began with a practical question from Teamflow's sales team: could AI automatically update Salesforce after sales calls? Crivello started exploring the possibility and, as he later described it, "kept climbing the ladder of abstraction" until he recognized he was actually building a platform for AI agents broadly, not just a Salesforce integration.
OpenAI had released the GPT-3 API in mid-2022, well before ChatGPT became a mainstream product, and Crivello saw an opportunity. On January 1, 2023, he made the decision to pivot Teamflow entirely. He reduced the team from 50 people to around 15, which involved letting go of the entire go-to-market organization as well as engineers whose work would not carry forward. It was one of the harder decisions of the pivot: Crivello described having to part with people who had been with the company through the Teamflow years.
In March 2023, Crivello posted a demo video on Twitter showing Lindy's early vision: an AI assistant that could handle calendar management, email drafting, and other routine professional tasks. The video generated 70,000 waitlist signups. The response validated the concept even before the product was working properly.
The name "Lindy" comes from the Lindy effect, a concept in probability and complexity theory suggesting that the future life expectancy of non-perishable things increases with their age. For ideas and technology, something that has survived a long time has already proven its durability, and is more likely to continue surviving. The name reflects Crivello's interest in building products that persist rather than chasing trends.
Lindy's first version was, by Crivello's own account, badly broken. Agents occasionally sent emails with text that literally stated "the user wants me to send an email," rather than sending an actual message. Despite these problems, the product found early customers interested in the concept.
The initial positioning framed Lindy as an "AI employee." This framing proved too abstract for the product it actually was at that stage. Crivello repositioned Lindy as the "Zapier of AI" with the tagline "if Zapier and ChatGPT had a baby," which grounded the product in something familiar. The repositioning made it easier for potential customers to understand what they were buying.
By the time Lindy reached $100,000 in monthly recurring revenue, Crivello concluded the original architecture could not scale to meet what customers actually needed. He spent five to six months rebuilding the product from scratch, launching Lindy 2.0 in November 2024. The rebuild moved away from a fully LLM-driven approach to a constrained visual workflow editor where language models handle specific nodes within a structured flow, rather than determining the entire execution path.
In May 2024, after the repositioning and while Lindy 2.0 was in development, Lindy hit product-market fit. Crivello described the period as overwhelming: more customers than the team could handle, servers breaking under load.
Lindy's current funding situation reflects the pivot from Teamflow. The $52 million raised by Teamflow in 2021 carried over when the company changed direction, meaning Lindy began operations in early 2023 with substantial runway already in place. The company had raised a seed round of $3.9 million, a Series A of $11 million, and a Series B of $35 million, all under the Teamflow entity before becoming Lindy. Battery Ventures has remained a consistent investor across rounds, with Neeraj Agrawal and Brandon Gleklen involved on the Battery side.
As of 2024, Lindy reported $5.1 million in annual revenue with a 37-person team, which translates to roughly $138,000 in revenue per employee. The company has not announced a separate external funding round under the Lindy name.
The decision to pivot rather than shut down Teamflow and start fresh allowed Crivello to carry existing investor relationships and financial runway into the new venture. This meant Lindy could operate without pressure to raise immediately during its early, pre-revenue period in 2023. It also meant the company avoided the dilution of a new seed round at a moment when Lindy had not yet proven product-market fit.
Lindy's core interface is a visual agent builder where users construct automation workflows using a flow-based editor. Workflows begin with a trigger and branch into sequences of actions. Each action can call an external application, run a conditional check, invoke a language model, or call another Lindy agent.
Agents are built by typing natural language instructions in prompt fields within each node, rather than writing code. The platform generates structured outputs from each node that subsequent nodes can reference, building up context as the workflow progresses. Crivello describes this as an "append-only ledger" model where each step adds to a shared context log.
The philosophy behind Lindy 2.0's architecture reflects a shift in thinking about AI reliability. Rather than asking a single language model to reason through an entire complex task, the platform keeps language model involvement constrained to specific decision points while traditional conditional logic handles routing and flow control. This approach reduces the frequency of unpredictable LLM behavior in production workflows.
Lindy ships with over 100 pre-built agent templates covering common business scenarios. Templates eliminate the need to build workflows from scratch for standard use cases.
Available templates include:
| Category | Template examples |
|---|---|
| Sales | Lead generator, lead qualifier, CRM updater, meeting scheduler, lead outreacher, meeting coach |
| Customer support | Website support chatbot, ticket dispatcher, FAQ generator, ticket triage with Slack escalation |
| Operations | Email triage and drafting, meeting summarizer, action item extractor, follow-up sender |
| Recruiting | Candidate screening, interview scheduler, outreach sequences |
| Voice/phone | Inbound call handler, outbound calling campaign, appointment scheduler |
| Productivity | Meeting prep researcher, calendar coordinator, document summarizer |
Users can start from a template and modify it, or build an agent from scratch using the flow editor.
Lindy agents activate based on triggers that define when an agent should start running. Available trigger types include:
Multiple triggers can be combined to build more sophisticated routing logic.
Lindy agents can invoke other Lindy agents as part of a workflow. This allows users to build hierarchical systems where a general-purpose agent delegates to specialized agents for specific tasks. A primary assistant might handle email triage and then call a separate recruiting agent when a candidate email arrives, or call a support agent for customer inquiries.
Crivello uses this architecture himself: he maintains separate agents for personal scheduling, meeting recording, email processing, customer support handling, and podcast transcription, with a top-level assistant routing tasks to the right specialized agent.
Lindy agents can store and retrieve persistent memory, which lets them maintain context across conversations and over time. An agent helping with sales outreach can remember that a particular contact was previously unresponsive to cold emails, or that a prospect expressed interest in a specific feature. Memory entries can be created automatically or manually, and agents can be instructed to prioritize certain types of memory over others.
The platform uses an append-only ledger structure within a workflow run, building context as each node executes. Across runs, agents access stored memory to inform their responses. However, Crivello has noted that excessive memory accumulation causes problems: agents that have stored thousands of minor facts start applying irrelevant memories in the wrong contexts. He actively prunes agent memory to keep the most relevant information. The scheduling agent that books his meetings, for instance, requires detailed memory about his preferences (buffer times, preferred days, video call versus in-person settings) but would be confused by memory from the email-drafting agent.
All paid plans include a knowledge base where users can upload documents, paste text, or connect data sources. Agents can query this knowledge base during workflow execution to retrieve relevant information. Support agents typically load product documentation, FAQ content, and policy documents into the knowledge base so they can answer questions accurately without relying solely on the language model's training data. Knowledge base capacity starts at 1 million characters on the free tier.
In 2025, Lindy introduced Gaia, its AI phone agent system built on Deepgram Flux, a low-latency speech-to-text model. Gaia is designed to handle inbound and outbound phone calls autonomously, with sub-second response times that Lindy claims beat competing voice agent products by more than 500 milliseconds.
Gaia supports calls in over 30 languages and handles tasks including:
The phone agent integrates with the rest of Lindy's workflow system, so a call handled by Gaia can trigger follow-up emails, CRM updates, or escalation to a human agent via Slack. Calls are priced at $0.19 per minute using GPT-4o, with other model options available at different price points.
Gaia distinguishes itself through integration depth: rather than treating phone calls as isolated conversations, the agent can look up customer records mid-call, schedule appointments in calendar systems, and log outcomes directly to connected CRMs during or after the call.
Lindy's Pro and higher plans include a computer use feature that allows agents to control a cloud browser, enabling automation of web-based tasks that do not expose APIs. This extends agent capabilities to applications that lack direct integrations.
Lindy's integration ecosystem has grown substantially since launch, largely through a partnership with Pipedream.
Before adopting Pipedream Connect, Lindy had built approximately 250 integrations over two years at a development cost exceeding $1 million. The main bottleneck was the OAuth registration process for each new service: registering an OAuth client for a new application, waiting for approval from the third-party service, and then building and maintaining production-ready API actions. Missing integrations cost deals because, as Crivello noted, "even just a one-week delay kills the deal 90% of the time."
The Pipedream Connect partnership gave Lindy access to over 2,500 integrations instantly, along with pre-approved OAuth clients and production-ready API actions. Combined with an Apify partnership for web scraping, the platform now connects to over 6,500 apps and data sources.
Core native integrations include:
| Category | Applications |
|---|---|
| Email and calendar | Gmail, Google Calendar, Outlook, Microsoft 365 |
| Communication | Slack, Microsoft Teams, Zoom, Twilio, iMessage/SMS |
| CRM | HubSpot, Salesforce, Pipedrive, Apollo.io |
| Productivity | Notion, Airtable, Google Sheets, Google Drive, Google Docs |
| Project management | Linear, Asana, Jira |
| Finance | QuickBooks |
| Support | Zendesk, Intercom |
| Database | Neon, Supabase |
Beyond direct integrations, agents can use HTTP request nodes to connect to any service with a REST API.
Lindy uses a tiered subscription model with a credit-based usage system. All plans include a seven-day free trial with no credit card required.
| Plan | Monthly price | Key inclusions |
|---|---|---|
| Free | $0 | 400 credits/month, core agent builder, 1M character knowledge base |
| Plus | $49.99 | Increased credits, 2 connected inboxes, all integrations |
| Pro | $99.99 | 3x usage versus Plus, 3 connected inboxes, computer use (browser automation) |
| Max | $199.99 | 7x usage versus Plus, 5 connected inboxes, high-volume workloads |
| Enterprise | Custom | SSO, SCIM provisioning, HIPAA compliance, audit logs, dedicated support, signed BAA |
Voice calls via Gaia are billed separately at $0.19 per minute (using GPT-4o). Credit consumption varies by action type: simple operations cost little per execution, while actions involving large language model calls consume more credits. This variable cost structure means actual monthly bills can differ significantly from the base plan price, depending on workflow complexity and volume.
The free tier's 400 credits per month are sufficient for light experimentation but limited for operational workflows, since most useful automations require premium actions that consume credits faster than the free allocation.
Lindy operates in a market that includes both traditional automation tools (Zapier, Make) and newer AI-oriented platforms (Gumloop, n8n). Each platform has different strengths depending on the use case.
| Feature | Lindy | n8n | Gumloop | Zapier |
|---|---|---|---|---|
| Primary audience | Business users building AI agents | Developers and technical teams | Data/content pipeline builders | Non-technical users, simple automations |
| No-code interface | Yes, visual flow editor | Yes, but technical | Yes, canvas-based | Yes |
| Self-hosting | No | Yes | No | No |
| AI agent depth | Core feature | Plugin-based | Core feature | Limited |
| Voice/phone agents | Yes (Gaia) | No | No | No |
| Native integrations | 250+ native, 2,500+ via Pipedream | 400+ | 100+ | 7,000+ |
| Pricing model | Credit-based tiers | Node-based / self-hosted free | Credit-based | Task-based tiers |
| Starting paid price | $49.99/month | $20/month (cloud) | $97/month | $19.99/month |
| Open source | No | Partially (fair-code) | No | No |
| Computer use | Yes (Pro+) | No | No | No |
Lindy's clearest differentiation is its phone agent capability and the depth of its AI agent architecture, particularly the multi-agent coordination features. n8n is the stronger choice for teams that need self-hosting, custom code execution, or developer-grade control over workflows. Gumloop is better suited for content pipelines, web scraping, and data-heavy workflows. Zapier covers the broadest app surface with the most straightforward setup for simple trigger-action automations.
Sales teams use Lindy to automate the administrative tasks that surround the selling process. A common workflow is lead qualification: when a prospect fills out a form, an agent researches the company and contact across web sources, scores the lead, updates the CRM record, routes it to the correct sales rep, and sends a personalized outreach email, all without human intervention.
Other sales use cases include automated follow-up sequences after meetings, CRM updates logged from email activity and call notes, calendar coordination to reduce scheduling back-and-forth, and real-time meeting coaching where an agent listens to sales calls and surfaces relevant product information or objection-handling guidance.
The Sauna Place, a Lindy customer, reported saving 15 to 20 hours per week after deploying sales automation agents.
Support teams deploy Lindy agents to handle routine tickets before they reach human agents. A support agent can read an incoming ticket, check a connected knowledge base, generate a reply, and send it, escalating to a Slack channel only when the question falls outside the agent's knowledge or confidence threshold.
Truemed, a payment processing company for HSA and FSA health purchases, used Lindy to automate customer support at scale. The company ran over 6,000 emails through Lindy's support agent and automated 36% of all support tickets. The cost per resolved ticket dropped from $1.00 to $0.33, a 67% reduction. The same implementation included AI agents for handling HSA/FSA eligibility verification questions and generating regulatory-compliant responses for health payment inquiries.
Operations teams use Lindy to eliminate routine coordination tasks. Meeting-related automation is particularly common: agents join video calls, record them, generate summaries with decisions and action items, and distribute notes to relevant team members or CRM records immediately after the call ends.
Email management is another common deployment: an agent reads incoming email, categorizes it, drafts replies in the user's voice for approval, schedules follow-up reminders, and handles routine requests autonomously while forwarding complex issues to the user.
Lindy's iMessage and SMS delegation feature is distinctive. Users can text a message to their Lindy agent on their phone, such as "schedule a call with Sarah next week" or "send the proposal to the client," and the agent handles the task without requiring the user to open a computer or a dedicated app. This makes the platform accessible in situations where users are away from their desks.
Recruiting teams use Lindy to screen candidates and coordinate scheduling. When a candidate submits an application, an agent reviews it against the job requirements, sends initial screening questions, schedules interviews with available team members, and updates the applicant tracking system, reducing the administrative workload on recruiters.
Lindy has been reviewed by a range of users across media outlets and direct-experience write-ups, and the feedback pattern is fairly consistent.
Positive reception focuses on the platform's accessibility for non-technical users, the quality of the pre-built templates, and the breadth of integrations following the Pipedream partnership. Reviewers note that getting a functional agent running takes minutes rather than hours when using a template. The meeting summarizer and email triage workflows receive consistent praise as genuinely useful out of the box.
The YouTuber MattVidPro published a review of Lindy 2.0 shortly after its quiet release in late 2024, calling it the best AI agent platform he had tested. The video was not a paid promotion; the attention it generated permanently inflected Lindy's growth trajectory, according to Crivello.
Critical feedback centers on a few recurring themes. The credit-based pricing model creates unpredictability, particularly for complex workflows where credit consumption is hard to estimate in advance. Several reviewers noted that the free plan's 400 monthly credits are insufficient for any meaningful operational workflow since almost all practical automations require premium actions. One reviewer described it as "credit anxiety" that discouraged experimentation.
Reviews of Lindy's performance on complex workflows are more mixed. Lindy works well for well-defined, template-adjacent tasks but can produce inconsistent results when workflows branch significantly or when agents need to handle edge cases outside their training distribution. A user attempting to recreate a sophisticated content processing system reported frequent errors that burned through credits without completing the task.
Call transcription accuracy in Gaia has been flagged as an occasional issue. Voice-native platforms focused exclusively on phone automation may offer better performance for high-volume, latency-sensitive calling applications than Lindy, which treats voice as one channel within a broader workflow system.
Some European users have raised questions about data handling documentation relative to what is available from platforms like n8n or Make, which have more explicit GDPR guidance in their public documentation. Lindy holds SOC 2 Type II, GDPR, HIPAA, and PIPEDA certifications, but the reviewers noted the documentation is less detailed than what regulated industries typically want to see before deployment.
Lindy has a 400,000+ user base and a growing community forum, suggesting strong adoption despite the criticisms. Its trajectory from a broken early product to a platform handling tens of millions of automated tasks speaks to significant engineering progress over a short period.
Several limitations are worth noting for teams evaluating Lindy:
Credit cost unpredictability. Because credits are consumed per action and the rate varies by action type, it is difficult to forecast monthly costs for a complex workflow before deploying it. Users on the free and Plus tiers can exhaust their monthly allocation mid-month, halting all agents until the next billing cycle.
No self-hosting. Lindy is a cloud-only SaaS product. Organizations with data residency requirements that prohibit sending data to a US-based cloud service cannot use Lindy in its current form. n8n's self-hosted option addresses this concern for those teams.
Voice at the generalist level. Gaia handles phone calls capably for most business scenarios, but dedicated voice platforms built specifically for high-volume, sub-400ms-latency calling (such as Vapi or Bland AI) may outperform Lindy in those specialized applications.
Complex workflow reliability. Like all LLM-based systems, Lindy agents can produce unexpected outputs when confronted with edge cases. The Lindy 2.0 architecture reduces this risk compared to the original version by constraining LLM involvement, but users should expect to iterate on agent prompts and test workflows under realistic conditions before using them in production.
No code export. Unlike n8n, Lindy does not allow users to export workflows as code or self-host the runtime. All workflow execution happens within Lindy's infrastructure.
Customer support response times. Several Trustpilot and independent reviews noted that Lindy's support team can be slow to respond, with some users reporting waiting multiple days for responses to technical issues. This is a common growing pain for startups scaling quickly, but it affects users who encounter problems with production workflows.
Lindy holds SOC 2 Type II certification, meaning its security controls have been independently audited. The platform is also certified under GDPR (applicable to users in the European Economic Area), HIPAA (for healthcare-related workflows in the United States), and PIPEDA (for Canadian users).
All data is encrypted with AES-256 at rest and in transit. Lindy states explicitly that user data is not sold to third parties and is not used to train any language models. Enterprise accounts can request a signed Business Associate Agreement for HIPAA-covered workflows, which is a contractual requirement under US healthcare law before handling protected health information.
The Enterprise plan also includes SSO (Single Sign-On) and SCIM provisioning for centralized identity management, and audit logs for tracking which agents took which actions. These controls are primarily relevant for security teams at larger organizations that need a full audit trail of automated agent behavior.