You've probably seen Jev everywhere this week. It's the latest AI model that's been the hype of the web. In short: it says yes or no, or sorts your data. Jev is not a chatbot. It doesn't write. It only decides. You give it text + a question. It gives back: - Noul → yes or no (0 to 1) - Choice → pick one option (billing / sales / support…) - Score → a level on a scale What text can you give it? - Skool: posts, comments, DMs, member intros - Ads: ad copy, headlines, comments on your ads - Funnels: your about page, landing page, sales page - Inbox: emails, support tickets, form answers So I built Bruv, a CLI tool that lets AI agents (codex, Claude Code, pi etc..) use Jev. Run it locally for free, or plug in paid models. CLI tool = a program you use by typing commands in a terminal (example: bruv setup). For Skoolers, what does it do? It turns daily Skool activity and your community data into clear actions: who needs help, which posts matter, which members may leave, and what to do next. This is just my way of shipping fast and make tools that will make your community grow hopefully! This is my first public project. The code isn't perfect. bruv, tool link : https://github.com/hichem300/bruv/

Do you guys have any tips for increasing close or booking rate? it may be interesting to hear them! If you reply to leads within 5 minutes of them reaching out — your close rate almost quadruples. Doesn’t matter if it’s a DM, cold email, Upwork message, or ad lead. Just. Reply. Fast. Easiest way? Pipe leads into Slack. Get pinged. Respond instantly. That's it. You can use Instantly's feature to connect your Slack to it or use a webhook with N8N or Make, it may seem silly, but big upside. Most people sleep on it.

most ai icebreakers suck. they’re generic, robotic, and get ignored. but if you do it right, ai personalization can 2x your cold email replies —turning 1 booked call/day into 2 and even more… without sending more emails. here’s how to do it 👇 why it works - decision-makers get 8–20 cold emails/day. yours needs to stand out. - personalization is an amplifier: 3% reply rate → 6% reply rate. - but ai should only write your icebreaker, not the whole email. step-by-step 1. scrape lead data from linkedin using apify or EXA AI 2. send that data to ai with a good prompt (rules, examples, clear tone) 3. ai writes 1–2 sentences (the icebreaker) 4. you write the rest of the email yourself 5. send & watch replies go up prompt tips - tell ai who it is (“you’re an sdr writing warm, specific compliments”) (using the system prompt) - feed it rich inputs (job title, niche, achievements, signals) - set rules (no full names, warm tone, under 2 sentences) - use placeholders for specific signals + metrics you want referenced here's the prompt if you wanna copy it : We just scraped a linkedin profile. Your task is to take their summaries and turn them into catchy, personalized openers for a cold email campaign to imply that the rest of the campaign is personalized. You'll return your icebreakers in the following JSON format: { "icebreaker": "Cool move building out/That shift into {thing} stood out/Smart play using {superSpecificThing}, especially the part about {specificSignal}.\n\nNoticed {anotherThing} came up too, so figured I’d reach out with something that might be helpful." } Rules: - Keep the tone warm, simple, and friendly, like you're talking to someone at a coffee shop. Think 3rd or 5th grade reading level. - No full names in the icebreaker. That reads too robotic. - Make sure to use the above format when constructing your icebreakers. We wrote it this way on purpose. - Mix up the first sentence starter using warm, friendly phrases pick what fits best based on what you're referencing.

Over the last 3 years, I struggled hard with prompting, I thought it was some secret trick that others used. Truth is, communicating clearly is the key. Human communication is really oonga boonga.. Here's the framework I use that’s making my AI workflows and also talking to AI generally 10x better: - Use a System Prompt to set the AI's role, think of it as the identity - Follow it with a User Prompt that clearly defines the task. - Then alternate User ↔️ Assistant examples to “train” it (one-shot or few-shot prompting). For example: - System: “You’re an AI sales expert.” - User: “Write a cold email opener, based on a lead information.” - User: "[INSERT LEAD INFORMATION] - Assistant: “Love your work on X—especially with Y.” This way, your AI gives you consistent, scalable output.

openai just dropped free models. you can now stop paying for chatgpt. they're called gpt-oss. 2 versions: 20b and 120b (that’s billion with a b). and yes, they can think, reason, and talk just like gpt. crazy. you can plug them into your automations in n8n and never pay api costs again. 2 ways to do it 👇 method 1 – huggingface + curl - go to huggingface.com - search “gpt-oss” models - grab the curl command (it’s under “deploy → inference API”) - paste it into n8n using the http request node - drop your huggingface token in - done. you’re live. method 2 – groq (free llm provider) - sign up at groq.com - get an api key (starts with sk-) - open n8n, use the ai agent node - choose groq as provider - paste your key - choose gpt-oss or another model (i like llama3 too) - set temperature to 0.7 (sweet spot for smart + creative) bonus tip: you can make a public ai chat with this setup. free chatbot. no fees. no limits. hope this helps!