Code & Career

Every cloud cost tool shows you the bill went up. None of them fix it. So you stare at a dashboard, see the number, and still have no idea why. AWS. GCP. Azure. OpenAI. Anthropic. 14+  providers. Seven-figure bills. And every tool just says: "You spent more this month." Great. Thanks. That's not actionable. Then I found 𝗻𝗮𝗯𝗹𝗲. Launched today. It's an agent that lives inside Claude and Cursor. Finds the waste across your cloud and AI spend, drafts the fix as a Terraform PR or ticket you approve, then watches your next bill to confirm the savings actually showed up. Here's how it works: ➟ Find the cause — points to the exact change that drove your bill up, down to the day it happened ➟ Fix it — writes the fix and waits for your go-ahead. Never changes anything on its own ➟ Prove it — checks your next bill to confirm the money was really saved Other tools claim a number. 𝗻𝗮𝗯𝗹𝗲 proves it on your actual bill. What makes it different: ✓ Pulls every connected provider, normalizes to FOCUS format, answers in plain English ✓ 14+ connectors — AWS, Azure, GCP, OpenAI, Anthropic, Stripe, Datadog, Snowflake, GitHub, and more ✓ Local-first, propose-only, your own key ✓ Never invents a number ✓ Built to handle seven-figure bills spread across 14 providers And for AI bills specifically: Most of an AI bill is input tokens billed at full price, plus calls sent to a frontier model a cheaper one would have answered. nable reads the split from your real usage and shows you the cheapest way to get the same output. Connect GitHub and it goes even deeper — you can track $ per PR, giving you real efficiency KPIs for your AI spend across engineering. Finally, a way to measure if your AI tools are actually worth it. Real example from an early user: → 89% of their Bedrock bill was input tokens, billed at full price with no caching → $10.7k/year = $896/mo in prompt-caching savings on the first scan Free for solo use. Team is $100/month flat for your whole team — not per seat. Local-first. Run it yourself, or let them host it. Credentials live in your OS keyring. No data lake to breach. 👉 Stop guessing. Start fixing. 📌 Install nable → https://getnable.com/ 𝗻𝗮𝗯𝗹𝗲 is live on ProductHunt right now: https://www.producthunt.com/products/nable

Most AI video tools give you clips. This one gives you a movie. The current AI video workflow is broken. You generate a clip. Then another. Then another. You stitch them together in editing software. You pray characters stay consistent. You hope the visual flow works. It's fragmented. Manual. Painful. Then I found OpenArt AI Director (https://tolt.link/openartdirectoranuj). Launches June 23rd. And honestly? It feels like the video equivalent of vibe coding. You're not prompting. You're vibe directing — shaping stories through conversation, scene by scene. Here's what makes it different: ➟ 5-minute cinematic videos in a single generation. Not seconds. Minutes. ➟ Conversational creation — develop stories, characters, environments, and style through chat. No prompt wrestling. ➟ Multimodal inputs — text, images, video, audio, music, and voice all work together. ➟ Scene-to-scene consistency — characters, voices, and visual styles stay coherent across every shot. ➟ Timeline-based shot editing for when you want precise control. ➟ 8+ languages with phoneme-level lip sync — English, Mandarin, Japanese, Spanish, French, and more. ➟ Remembers long-term projects and brand guidelines. Stay on-brand, on-mode. What this actually means: ✓ Short films, micro-dramas, and trailers — without a crew ✓ Music videos with narrative flow ✓ Brand ads that stay on-brand across every scene ✓ Explainer videos that actually teach ✓ UGC content with recurring characters I keep calling this vibe directing because that's what it actually is — you're describing what you want, reacting to what comes back, and shaping it scene by scene through conversation. 👉 You're not assembling clips anymore. You're directing. OpenArt AI Director is now live on Product Hunt. If you're excited about where AI filmmaking is heading, I'd appreciate you checking them out and supporting the launch: 👉 https://www.producthunt.com/products/openart

Your AI agent is doing things in production you don't know about. And that's a problem. You shipped it. It's "working." But what's it actually doing? Where is it failing? Which users are hitting the same bug over and over? Most teams are flying blind. Then I found Latitude V2 It's an open-source monitoring platform for AI agents. MIT licensed. You can run it yourself, read every line, make it yours. Here's the three-act flow: See what your agent actually does in production ➟ Clusters thousands of conversations into one clear picture ➟ Shows what people ask for, where they hesitate, escalate, or drop off ➟ Semantic search — type "users asking about a competitor" in plain English and the exact conversations come back Catch what's breaking before users do ➟ When your agent keeps failing the same way, Latitude collapses those moments into one signal: the problem, how often it fires, and why ➟ Monitors run against every new conversation — you hear about patterns before your users do ➟ Automatic issue detection, or set your own Fix it without leaving your editor ➟ The MCP server brings signals, traces, and searches straight into your coding agent ➟ Turn real failures into a dataset and verify the fix worked before you ship ➟ Automated evals generated from real examples, grounded in actual failure modes What this actually means: ✓ Understand what users actually ask for ✓ Spot tool failures, escalations, and churn before they become problems ✓ Get alerted on Slack or email when issues are detected ✓ Turn any issue into an eval that runs on every new trace ✓ Manage everything from your coding agent — no UI required OpenTelemetry compatible. Drop in the SDK or point your existing OTEL pipeline. Trusted by teams building agents at scale. Set up in less than 5 minutes. 👉 Stop flying blind. Start seeing what your agent really does. 📌 Star on GitHub → https://github.com/latitude-dev/latitude-llm 📌 Get started for free → https://latitude.so Latitude V2 is live on ProductHunt right now: https://github.com/latitude-dev/latitude-llm 🔄 Repost if you've ever debugged an AI agent by reading logs at 2am.

Finally, outbound that is not just another text sequence. Outvid helps teams send personalized videos, voice notes, emails, and follow-ups across multiple channels from one workflow. Here’s the thing about most outreach tools: They help you send more messages. But they don’t help you say something worth replying to. You still end up with generic cold emails, shallow personalization, disconnected channels, and replies scattered across different inboxes. Most “personalization” is just first-name merge fields. And prospects know it. Then I found Outvid. It’s an AI outbound platform that deep researches every prospect, writes personalized outreach, creates and films founder-led videos and voice notes, sends across multiple channels, follows up automatically, and brings replies into one unified inbox. Here’s how it works: ➟ Upload your prospect list ➟ Outvid researches each person and company to find what’s actually worth mentioning ➟ It writes the email, video script, voice note, and follow-up messages ➟ It records a unique video or voice note with your real face and voice ➟ It sends across email, LinkedIn, WhatsApp, and Instagram, follows up, and puts replies in one unified inbox No fake avatars. No shallow personalization. No switching between five tools. What you can do with it: ✓ Send personalized videos, voice notes, emails, and follow-ups ✓ Run outbound across email, LinkedIn, WhatsApp, and Instagram ✓ Turn one founder recording into unique videos and voice notes for every prospect ✓ Use deep prospect research instead of first-name merge fields ✓ Automate follow-ups without sounding generic ✓ Manage replies from one unified inbox ✓ Give every prospect a message based on their company, role, and context One recording → N personalized videos and voice notes. Each one names their company, their role, and the trigger that earned the outreach. Built for founders doing founder-led sales, sales teams trying to stand out, agencies running outbound for clients, and GTM teams that want research, personalization, sending, and replies in one place. Trusted by 5,000+ founders. 👉 No generic blast campaigns. Just replies that actually convert. 📌 Try Outvid → outvid.ai 🔄 Repost if you’ve ever sent a “personalized” email that was clearly a template.

dakbkf

Most "AI startup tools" give you a blank chat box and a blinking cursor. Like that's helpful. "You can build anything!" Cool. I don't know what to build. Or how to start. Or if the idea is even good. That blank box is where startup dreams go to die. Then I found Fonda. Launched today on Product Hunt. It's not another chatbot pretending to be helpful. It's a structured AI co-founder that actually runs a process with you. Here's what's different: ➟ Real business opportunities tailored to your skills, experience, goals ➟ A proven 14-step journey: Discover → Validate → Launch → Scale ➟ Remembers every decision, interview, and pivot across sessions ➟ Kills weak ideas before you waste months on them ➟ Ships your landing page, business case, and MVP plan once validated ➟ One clear next move. Every day. No guesswork. Not "here's some generic advice." Not "write a business plan" with zero scaffolding. An actual co-founder that: ✓ Scans real market signals (funding rounds, launches, demand shifts) ✓ Designs falsifiable tests — not vanity surveys ✓ Gives you a scored verdict: go, refine, or pivot ✓ Helps you find and talk to real customers ✓ Auto-drafts the boring stuff (landing page, legal checklist, roadmap) Grounded in peer-reviewed entrepreneurship research from Stanford, Harvard, INSEAD. Free to start. 👉 I've seen a lot of "AI for founders" tools. This one actually gets it. 📌 Meet your AI co-founder → https://fonda.co/ Fonda is live on ProductHunt right now: https://www.producthunt.com/products/fonda

I just calculated how much time I waste typing the same Slack messages. It was embarrassing. "Thanks, I'll take a look." "Great meeting, here's the recap." "Let me check with the team and get back to you." Every day. Same phrases. Same DMs. Same Gmail replies. I've been copy-pasting from a hidden Notes app like some kind of digital peasant. Then I found 𝗧𝗲𝘅𝘁𝗔𝗻𝘆𝗡𝗼𝘄. Small extension. Does one thing well. ↳ Store your reusable replies once ↳ Insert them into Slack DMs or Gmail compose with one click ↳ Templates can include subject lines for Gmail (huge time saver) No more retyping. No more hunting for that "perfect response" you wrote last week. Here's what actually surprised me: ➟ Google Sheets integration — import and update templates in bulk ➟ Works for sales outreach, support replies, marketing proposals, link building ➟ Central library so your whole team says the same thing the same way What this saves: ✓ Sales teams stop rewriting intro messages ✓ Support stops typing "thanks for reaching out" 50x a day ✓ Anyone using Slack or Gmail gets back minutes every hour It's not glamorous. But neither is typing the same sentence for the 400th time. 26 users today. But honestly? Everyone who uses Slack or Gmail needs this. 👉 Free extension. Try it once and you'll see. 📌 Stop retyping yourself → https://textanynow.com/plugin

Copy-pasting between 12 AI chats isn't "workflow." It's chaos with extra steps. You know the ritual. Open ChatGPT → copy response → paste into Claude → copy again → paste into Gemini → lose the original thread → scroll endlessly → give up. Every new chat box starts from zero. Context collapses. Your best thinking disappears into a black hole of disconnected tabs. This is how "power users" work today. And it's broken. Slashspace (Previously RabbitHoles AI) fixes the actual problem: ↳ Not another chat window. ↳ A canvas where AI lives spatially — not chronologically. Here's what changes: ➟ Every agent, every response, every source sits on one infinite canvas ➟ Nodes branch, connect, and reference each other (no copy-paste) ➟ Your email, Slack, calendar, PDFs, YouTube videos all drag onto the canvas ➟ Agents see the full context — not just the last 3 messages ➟ Canvas saves locally as files. You own your work. Built with 1,600 power users over 1.5 years. This isn't a weekend MVP. It's the most mature canvas AI on the market. What this unlocks: ✓ Research? Deep agents that synthesize across dozens of sources ✓ Coding? Cursor Agent runs directly on the canvas ✓ Writing? Spatial layout > endless scrolling ✓ Founder work? Plan, draft, edit, review — all in one view No more "which tab had that answer from yesterday?" 👉 Your context stays intact. Your focus stays locked. 📌 Stop losing your best work to disconnected chats → https://www.slashspace.ai/ Slashspace (Previously RabbitHoles AI) is live on ProductHunt right now: https://www.producthunt.com/products/slashspace-ai

"You forgot the button. Again." Every developer knows this moment. You paste a screenshot into ChatGPT. "Fix the UI here." The AI writes perfect code. For the wrong button. Because it couldn't tell which one you meant. My friend just built the fix. It's called SlimSnap. Launched on Product Hunt today. Mac app that does one thing well: ↳ Annotate a screenshot → highlight the actual button ↳ SlimSnap turns it into structured JSON ↳ AI parses it element by element — no guessing Works with ChatGPT, Claude, Cursor, Lovable. Anywhere you paste screenshots. Free. Signed. Open MIT schema. No "sign up for a trial." No "talk to sales." Just download and paste. Here's what my friend actually needs right now: Honest takes. Critique. Broken edge cases. Upvotes are nice. But a comment that says "this broke when I tried X" is worth 10x more. 👉 Check it out → https://www.producthunt.com/products/slimsnap If you paste screenshots into AI for UI work — give it 2 minutes. Then tell them what sucks. That's the real help.

I ran a 7-figure Amazon FBA business. And for years, inventory planning felt like educated guessing. ↳ Order too late → stockout → lose sales and ranking ↳ Order too early → cash trapped in inventory ↳ Trust spreadsheets → spend hours analyzing data but still second-guess every decision I tried inventory tools. Most gave me dashboards, reports, and charts. But they didn't answer the questions that actually matter: • What should I reorder? • When should I reorder? • How much should I buy? That's why NovaMind AI was built. Created by a 7-figure Amazon seller, NovaMind helps FBA businesses make smarter inventory decisions using AI-powered forecasting and inventory intelligence. What makes it different? ➟ Forecasts future demand using real sales data ➟ Identifies stockout risks before they happen ➟ Prioritizes inventory based on business impact ➟ Helps sellers optimize cash flow and inventory investment ➟ Built specifically for Amazon FBA sellers No more guessing. No more spreadsheet chaos. No more costly inventory mistakes. Amazon SP-API approved. Built by sellers, for sellers. 👉 Your inventory should work for you. Not the other way around. 📌 See exactly what to reorder — before it's too late → https://inventory.ainovamind.com/ NovaMind AI is live on Product Hunt right now: https://www.producthunt.com/products/novamind-ai-2 If you sell on Amazon, I'd love to hear how you currently manage inventory planning.