I've built operator23.com, you build AI agents from plain english across 900 platforms. E.g. you can just prompt "I want an agent that everytime I get an email from a potential customer, please check if the customer is in the close stage in hubspot, and if the customer is in the close stage, notify me on slack and draft an email" Though many people do not know what they should automate, and products like n8n and zapier requires their customer to have an workflow visualize mindset, that they can write out the exact workflow and also know their own caveats and inefficiencies. Therefore I'm going towards, an agent that understand the customer company, users role, workflow, and platforms they use to give examples on agents it could deploy for them. Question: How do I design an onboarding that activates users who don't know what to automate

Llama 4 Scout

Vote: Yes

The question revolves around designing an onboarding process for users who are unsure about what they can automate. The user's platform, operator23.com, and similar services like n8n and Zapier require users to have a clear understanding of their workflows and what they want to automate. The goal is to activate users who lack this understanding. An onboarding process that activates users who don't know what to automate should ideally be interactive, guiding, and capable of providing examples based on the user's company, role, workflow, and platforms they use. This approach can help users discover potential automation opportunities. Therefore, the answer to whether you should design an onboarding that activates users who don't know what to automate is affirmative.

Qwen 3 235B

Vote: Yes

Yes is the appropriate vote because the question is asking for a design strategy to solve a real and common user onboarding challenge — activating users who don't know what to automate. The core issue is user awareness, not product capability. Therefore, the onboarding must educate and inspire users by surfacing personalized, actionable automation ideas based on their role, tools, and workflows. An effective onboarding should include a guided setup that asks about the user's job function, tools in use, and common tasks. This data fuels AI-generated, relevant agent suggestions tailored to their context. Including interactive walkthroughs, example automations by persona (e.g., sales manager, support lead), and one-click deployable agents reduces friction and builds early value. Finally, embedding a discovery mode where the AI observes (with permission) workflow patterns over time can proactively recommend automations, turning onboarding into an evolving, intelligent process.

claude-haiku-4-5

Vote: Yes

The question is asking for design guidance on solving a real problem: users who don't know what to automate. This is a legitimate product design challenge worthy of exploration. Your insight is sound—most automation tools assume users already understand their workflow inefficiencies. Your proposed solution (an agent that analyzes company context, user role, and existing platforms to suggest automations) directly addresses this gap and is implementable. An effective onboarding for this use case should include: (1) guided discovery questions about the user's role and pain points, (2) integration with their existing tools to analyze actual data flows, (3) AI-generated personalized automation suggestions with clear ROI explanations, and (4) one-click deployment of recommended agents. This approach reduces cognitive load compared to traditional workflow builders, making your product more accessible to non-technical users—a genuine competitive advantage.

Sonar Pro

Vote: Yes

The question asks how to design an onboarding process that activates users who don't know what to automate, which is a valid product design challenge for operator23.com. Option A ('Yes') affirms that such a design is possible and worthwhile, aligning with the user's goal of using an agent to understand company context and suggest automations. Search results support this directly: sources describe structured onboarding with personalization, example utterances, and AI-driven suggestions to guide users [1][2][4][5]. For instance, providing sample prompts and agent examples reduces friction for novices, mirroring HR-style orientation or agentic workflows that educate before deployment [1][6]. Competing tools like Zapier fail here due to requiring workflow visualization, but results emphasize AI's role in discovery via conversational interfaces and personalization engines [3][5], making activation feasible. Voting 'No' ignores these proven strategies and the product's core value prop.

Magistral Small

Vote: Yes

The proposed onboarding design is crucial because many users don't know what to automate. By understanding the user's company, role, workflow, and platforms, the system can provide tailored examples and suggestions. This approach lowers the barrier to entry and activates users by showing immediate, relevant automation possibilities. Unlike tools like n8n and Zapier, which require users to visualize workflows upfront, this method guides users, making it more accessible and engaging.

Grok 4.20

Vote: Yes

Yes, an effective onboarding is possible and essential. Start by prompting users for basic info like their job role, company type, and key platforms they use. Leverage the AI to instantly generate 3-5 tailored agent examples based on common inefficiencies for that role, pulling from a knowledge base of proven automations. This shows immediate value without requiring workflow visualization skills, unlike Zapier. Include one-click deployment for a suggested agent to drive activation.

GPT-5.4

Vote: Yes

A. The underlying question is not actually yes/no in substance, but option A is the better fit because the user is clearly asking for guidance on how to design onboarding, and an affirmative vote best maps to endorsing that direction. For users who do not know what to automate, onboarding should not begin with a blank prompt. It should begin with structured discovery: role, team, tools, goals, recurring tasks, bottlenecks, and approval constraints. The best activation pattern is "tell us about your work, we will suggest agents." Ask a few high-signal questions, connect their stack, then generate 3-5 recommended automations ranked by impact, ease, and confidence. Make the suggestions concrete and editable. For each proposed agent, show trigger, actions, apps involved, business outcome, and estimated time saved. Let the user launch a template immediately or refine it in plain English. A strong onboarding flow would be: choose role/use case, connect core tools, answer a short diagnostic, receive recommended agents, preview with real data, and activate one low-risk quick win first. Also use progressive disclosure. Do not ask them to map their whole workflow upfront. Start with one department-level job to be done like lead follow-up, support triage, meeting prep, CRM hygiene, or invoice chasing. Finally, create opinionated starter packs by persona: founder, sales rep, recruiter, customer success, ops, etc. People react better to examples that look like their day-to-day work than to abstract automation capabilities. So I vote A because the right onboarding is recommendation-led, persona-aware, and quick-win oriented rather than blank-canvas workflow design.

Claude Opus 4.6

Vote: Yes

This question is asking for design advice on building an onboarding flow that activates users who don't know what to automate. While the answer options (Yes/No) don't naturally map to this open-ended question, I'll vote A (Yes) as it aligns with affirming the approach described. The founder has correctly identified a key insight: most users don't know what they should automate. The proposed solution — an agent that understands the user's company, role, workflow, and platforms to suggest automations — is a strong approach to solving the cold-start problem. A good onboarding design here would involve: (1) a guided interview-style flow asking about role, tools used, and daily pain points; (2) presenting curated, role-specific automation templates based on those answers; (3) letting users activate a pre-built agent with one click to experience immediate value; and (4) progressively revealing more complex automation possibilities as the user gains confidence. Voting A because the direction is sound and the question implicitly asks whether this discovery-first onboarding approach is viable — it absolutely is.

GLM 5

Vote: Yes

The question asks 'How do I design an onboarding...' which is a strategic question, not a yes/no proposition. However, voting 'Yes' affirms that designing an onboarding specifically for users who don't know what to automate is the right approach—and it absolutely is. Your insight is correct: the biggest barrier for n8n/Zapier-style tools is that users need workflow visualization skills AND self-awareness of their own inefficiencies. That's a high bar. By flipping the model—where your agent proactively learns about the user's role, tools, and processes, then suggests automation opportunities—you remove that cognitive burden. An effective onboarding for this segment would: (1) Ask about their role and tools upfront, (2) Run a discovery interview or questionnaire about repetitive tasks, (3) Immediately show 2-3 concrete agent suggestions relevant to their stack, and (4) Let them deploy one with a single click. Show value before asking them to think like an automation engineer.

Gemini 3.1 Pro

Vote: Yes

While the prompt asks an open-ended question but only provides binary 'Yes/No' options, I am voting 'A' (Yes) to affirm that you absolutely can and should design an onboarding experience that bridges the imagination gap for your users. To activate users who don't know what to automate, shift the onboarding focus from 'creation' to 'discovery.' Instead of greeting them with a blank text box, start by asking them to connect the tools they already use (e.g., Gmail, Slack, HubSpot) and select their job role (e.g., Sales, Marketing, HR). Once you have this context, use your AI to proactively generate a personalized menu of 'Ready-to-Deploy Agents' tailored specifically to their tech stack and role. For example, if they connect HubSpot and Slack and say they are in Sales, immediately present them with the exact use case you described in your pitch as a one-click template. Finally, implement a conversational audit. Instead of asking 'What do you want to automate?', have your AI ask 'What task took up most of your time today?' or 'What is the most annoying part of your morning routine?' This lowers the cognitive load, allowing your AI to parse their plain English pain points and translate them into ready-made automation workflows.

Gemini 2.5 Pro

Vote: Yes

The fundamental flaw in the provided options is that they are binary 'Yes'/'No' answers to a 'How' question. Assuming the question is implicitly asking whether the proposed direction is a good one, then the answer is an emphatic 'Yes'. Your proposed onboarding strategy directly addresses the core user problem: the 'blank canvas' dilemma. Most users struggle not with the *how* of automation, but with the *what*. Traditional tools like Zapier place the burden of imagination entirely on the user, leading to high drop-off rates for those who lack a clear, pre-existing workflow to automate. By creating an agent that proactively understands a user's context (role, company, tech stack), you shift the model from a passive tool to an active consultant. This allows you to serve hyper-relevant, personalized suggestions that act as powerful 'Aha!' moments. Instead of asking 'What do you want to build?', you're saying 'Based on your role as a Sales Manager using HubSpot and Slack, here are three high-impact automations you could deploy right now.' This approach dramatically shortens the time-to-value. A user can go from sign-up to a deployed, valuable agent in minutes, without needing to visualize complex workflows. This not only solves their immediate problem but also educates them on the art of the possible, inspiring them to explore further and become truly activated, long-term users.

MiniMax M2.5

Vote: Yes

Your approach of building an agent that understands the customer's company, role, workflow, and platforms is the right direction. Users who don't know what to automate need contextual examples, not a blank canvas. For onboarding design: First, ask discovery questions about their role, industry, and main tools (CRM, email, Slack, etc.). Then show 3-5 pre-built agent templates specifically relevant to their context - a sales rep sees different suggestions than an accountant. Let them activate one with one click, then iterate from there. The key insight is that your product solves the 'blank canvas' problem that Zapier/n8n have. Your onboarding should lean into this by being recommendation-first rather than creation-first.

Sonar Reasoning Pro

Vote: Yes

Designing an onboarding for users who don't know what to automate is essential, not optional. Your product has identified a genuine friction point—competitors like n8n and Zapier require users to visualize workflows before they even know what problems automation could solve. The search results strongly support this approach. AI onboarding research emphasizes that effective activation requires teaching users what "good" looks like and aligning their actions with the value proposition[1]. One key principle: start with the problem, not the shiny solution—define concrete outcomes like reducing manual work by 60%[3]. Your proposed solution directly addresses this gap. By having your agent understand the customer's company, role, workflows, and platforms to proactively suggest automation examples, you're doing exactly what the research recommends: providing context, role-specific guidance, and concrete use cases upfront[1]. This is similar to how agentic workflows personalize experiences through templates and example-driven interactions[4]. This onboarding strategy isn't just good—it's your competitive advantage. It removes the mental barrier that stops users from adopting products like Zapier, making your platform accessible to users who want to automate but don't yet know how.

DeepSeek V3.1

Vote: Yes

This approach is highly effective because it directly addresses the core barrier: users' lack of awareness about automation opportunities. By designing an onboarding process where an AI agent understands the customer's context—such as their company, role, and platforms—it can proactively suggest tailored automation examples that resonate with their specific needs. This personalized guidance reduces the cognitive load on users and demonstrates immediate value, sparking ideas they might not have considered. It contrasts with tools like Zapier that require users to already visualize workflows, which can be intimidating and lead to drop-off. Ultimately, this method lowers the activation threshold by making automation accessible and relevant, increasing the likelihood that users will engage and deploy agents.

Kimi K2.5

Vote: Yes

Your diagnosis of the activation problem is correct: Zapier and n8n fail because they require users to already possess 'workflow visualization' capability and domain knowledge of their own inefficiencies. The solution is to design onboarding as an 'intelligent discovery' phase rather than a 'build' phase. Have your agent conduct a brief audit of the user's role, industry, and existing SaaS stack, then use that context to auto-generate 3-5 specific, deployable agent suggestions (e.g., 'I see you're a sales manager using HubSpot and Slack—shall I create an agent that alerts you when prospects go cold?'). This shifts the cognitive load from creation (hard) to curation/approval (easy), delivering immediate, personalized value without requiring the user to know what's automatable beforehand. It leverages the LLM's ability to recognize cross-platform patterns across similar roles, solving the cold-start problem that plagues template-based competitors.

GPT-5

Vote: Yes

Start with a 60–90 second intake: role, top goal (e.g., faster lead follow‑up), and connect tools (OAuth). Auto-detect stack and pull light metadata (pipelines, inboxes, channels) to personalize immediately. Run passive discovery (opt‑in, read‑only first): scan last 14–30 days of email/CRM/support/calendar for repetitive patterns (new lead arrives, deal stage changes, missed replies). Quantify event volumes to surface high‑impact candidates. Show 3–5 personalized “starter agents” as cards: Trigger → Actions, sample output using their data, expected weekly time saved, and risk level. Make “Preview on my data” and “Enable” one-click. Use a JTBD wizard: “When X happens in Y, do Z so I can W.” Pre-fill X/Y from their stack and recent events; let them edit steps via natural language or toggles. Always offer a plain-English prompt box. Safe simulation by default: run on the last 10 records, show step-by-step outputs, errors, and guardrails. Start agents in Draft/Human‑review mode with an easy approve/undo flow. Deliver an instant aha: auto-create one no-risk agent (e.g., daily Slack digest of new leads or unresolved tickets) and deliver the first result during onboarding. Add ROI meter and social proof: estimate time saved from their own event volumes; show what similar roles/companies automate; provide a role/stack-specific template gallery. Progressive permissions and trust: request read-only first; escalate when enabling writes. Provide logs, audit trails, versioning, and a one-click rollback. Lifecycle nudges: after connecting tools, send 3 personalized suggestions via in-app and Slack. Trigger contextual prompts when you detect repetitive manual actions (e.g., frequent CRM updates without follow-up). Design to activation metrics: time-to-value < 5 minutes; ≥2 connectors; first agent enabled + first successful run; 7‑day retained runs. A/B test chat-first vs gallery-first onboarding and template ordering by observed impact.

Final Summary

The AI Roundtable faced a classic linguistic trap—a 'how-to' question paired with a binary 'Yes/No' choice—and collectively decided to ignore the constraint in favor of a unanimous 'Yes.' The models, led by heavyweights like GPT-5 and Claude Opus 4.6, agreed that the current industry standard (n8n and Zapier) fails precisely because it demands a 'workflow visualization' mindset that most users lack. The consensus was clear: the only way to activate the uninitiated is to pivot from a passive 'creation' tool to an active 'discovery' consultant.

No agreement after 1 rounds of discussion. Final vote: Yes: 16

Strongest Arguments

• Yes: The most compelling argument, synthesized by GPT-5 and Gemini 2.5 Pro, is that onboarding should be an 'intelligent discovery' phase rather than a 'build' phase. By using a 60-second intake of a user's role and SaaS stack, the AI can proactively generate 'ready-to-deploy' agents, shifting the user's job from difficult creation to simple curation.