Fluidworks - SaaS onboarding agent that talks , clicks, and guides users

@gsvineeth 

Thank you.

You can tune the agent’s voice based on a video or audio sample.

If you want the onboarding to sound like your founder, we can make that happen. Just share a few recordings of their onboarding or demo calls.

@tonytom_ 

Absolutely. Fluidworks maps every screen, UI element, and decision point during training, so the agent can guide users through multi-step flows with conditional branches. It tracks the user’s current state in real time, chooses the correct next action or explanation, and can backtrack or jump ahead if the user deviates. Complex paths such as role-based onboarding or feature gating are handled automatically once they’re captured in the training videos and sandbox walkthroughs.

@evgenii_zaitsev1 

We are anti thesis of whatever exists in market - dynamic and personal .

@evgenii_zaitsev1 

Thank you! We see three core differentiators:

1. Multimodal interaction

Fluidworks combines voice, cursor control, and on-screen highlights in the same flow. Traditional tours show tooltips; chatbots talk but can’t act. We do both, so users hear the explanation and watch the UI execute the steps in real time.

2. In-session context engine

Because the agent runs inside each user’s live session, it reads role, plan, page state, and even CDP or CRM attributes. That lets it branch instantly. An admin sees provisioning steps while an end user skips straight to daily tasks, with no extra hard-coding.

3. Fast, low-lift training

You give us a sandbox login and five short walkthrough videos. Our mapping layer crawls every screen, tags each element, and builds a graph of possible paths. No month-long playbook authoring. Most customers are live in under ten days.

The result is guidance that feels like a helpful teammate, not a scripted tour. And it keeps getting smarter with every user interaction. Happy to dive deeper if you’d like!

@supa_l Really appreciate that! That’s exactly the goal. Make onboarding feel like a helpful teammate sitting beside the user, without adding extra work for the product or support teams. Activation up, tickets down, happier users all around. Thanks for the support!

@suhasmotwani 

Thanks man 🙏

@iamarnob6543 

Thank you - Please tell us how do you feel about it ?

@malithmcrdev 

Did you able to try it ?

@madalina_barbu 

Thank you - Do create an agent and take it for a spin

@jonas_wallengren 

Thanks for the kind words.

Really appreciate you looking under the hood.

You’re spot on. While the headline feature is real-time voice guidance, we built the agent to degrade gracefully. A user can flip to Silent Mode at any point (or start there by default).

In that state:

• The agent’s narration switches to a lightweight chat overlay, so every step and prompt still appears as text

• Cursor automation and in-app highlights continue unchanged, keeping the “show-me-where-to-click” experience intact

• Users can toggle voice back on with one click or a keyboard shortcut if their context changes

On the admin side you can decide whether new sessions launch in voice, silent, or “follow system mute,” so it adapts to libraries, flights, or offices that frown on audio.

Happy to walk you through the toggle flow on a quick call or spin up a sandbox so you can try it firsthand. Let me know what works for you.

@dessignnet 

Please try it out and give us your feedback

@oleifrking 

Damn - do you happened to know the name of company ?

@patpijanowski 

Thanks! That’s exactly what we built for.

How deep can it branch?

Very deep , think state-machine rather than linear script.

During training we capture every screen, element, and branch point, then store that as a graph. At runtime the agent:

1. Reads context in real time (role, plan, feature flags, current URL, DOM state).

2. Chooses a path through the graph based on that context—so Sales, Admin, or End-User roles each get their own flow without extra hard-coding.

3. Re-plans on the fly if the user skips, backtracks, or changes settings mid-flow.

Out of the box it supports multi-step decision trees, gated features, and role-based variations as long as those paths exist in the sandbox or video samples you give us. Most customers cover 90 % of their branching logic with five short walkthrough videos per role; we handle the rest through iterative updates.

@tom_shapland 

Thanks for the kind words! We agree: every product should welcome new users with a helpful guide, not a static tour. If any specific apps come to mind, let us know; we’re always looking for the next user we can delight.

@success_saint 

Thank you - Any feedbacks for us ?

@mohamed_gamal_mgm_ 

Technically yes , but we plan to support other languages on a case-by-case basis .

Let’s talk if you are interested. https://cal.com/fluid/hn-chat

@beyza_gu 

“Ah, the ancient language of Keyboard Cat! 🐈⌨️

Translating… ‘Your product is awesome, keep it up.’

Glad we cracked the code
thanks for the support!”

@rohini4 

Great question! 🙌

• Real-time AI voice: Speech is generated on the fly with low-latency TTS, adapting to every click and branch.

• Product-team controls: Choose a voice from our library, tweak tone or speed, and drop in custom script snippets for key moments.

• Custom voices (optional): Provide a short studio recording and we can clone a branded voice for you.

Nothing is hard-wired or pre-recorded; everything is dynamic and fully customizable.

@dzaitzow 

Think of it this way:

• Arcade → a slick GIF demo: great for marketing pages, but it’s still a “watch me click” replay.

• Intercom tours → scripted tooltips and checklists: decent for basics, but you have to keep every step up-to-date by hand.

• Fluidworks → a live voice co-pilot: it clicks inside the app, answers “how do I…?” in real time, personalizes the path mid-session, and auto-remaps when the UI changes—so onboarding feels like a helpful teammate, not a slide deck.