The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
While building InvoiceKit, I stopped reviewing business features as a list of screens and used five questions: Who owns the record? What can change? Where is it summarized? Which state leaves the app? How is persistence verified? For an invoice, those questions connect owner-scoped Firebase access, revisions and statuses, dashboard and detail views, and PDF/CSV sharing. If any answer is vague, the estimate still contains hidden work.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Trending
retrofuturism
Share your favorite designs from the past exploring the distant future. Or your own take on the aesthetic 🚀
halftone
Dots. Texture. Grain. Let's see your designs using the halftone technique.
Midjourney
Creatives are showing off what they're making with Midjourney.
colorgrade
Video editors sound off: How do you enhance the mood of your video in color grading?
Related posts
Your AI agent just did 9 things: pulled 412 orders, matched them to Stripe, caught a duplicate £248 charge, drafted a ledger entry, opened a ticket. Whole run cost $0.011.
Now the real design question — what does the user actually SEE?
A — The transcript. A paragraph of text and a collapsed "Used 3 tools".
B — The run canvas. Nine colour-coded rows: what ran, what it cost, what got flagged, and one step that stops and waits for you before money moves.
Same agent. Same run. Two completely different products.
Vote A or B — and tell me why. I build the B version for a living.
0 voted
0%
1 voted
100%
1 vote
Closed
Wow! So beautiful and professional work!👌
Your churn number is four problems wearing one coat.
An expired card needs retry logic. A week-three exit needs a better activation path. That's one percentage on your dashboard pointing at two unrelated repairs.
Project title: AI Voice & Text Personal Assistant for Slack
Project description:
Built a conversational AI assistant in n8n that lives inside Slack and handles both voice and text messages, giving users a personal assistant they can talk to naturally, without switching apps or typing long commands.
The problem: The client wanted a way for their team to quickly get answers or delegate tasks through Slack, including by voice, without building a separate app or bot from scratch, and needed the assistant to hold context across a conversation rather than treating every message as a one-off.
What I built:
Message filtering: Every incoming Slack message is automatically checked to filter out bot-generated messages, so the assistant only responds to real human input.
Voice and text handling: The system detects whether the incoming message is a voice note or plain text. Voice messages are automatically downloaded and transcribed into text using an audio transcription model, while text messages pass straight through, so both input types end up in the same format for processing.
Conversational AI agent with memory: The transcribed or typed message is sent to a personal AI agent (built on Groq) that generates a response. The agent uses a memory module to retain context across the conversation, so it can handle follow-up questions naturally rather than starting fresh every time.
Instant reply in Slack: Once the agent generates a response, it's posted straight back into the Slack channel, so the interaction feels like a live conversation.
Outcome: The client's team now has a Slack-native assistant that understands both typed and spoken messages, remembers conversation context, and responds instantly, no separate app, no manual transcription, no lost context between messages.
This is good