Marka Sesi Kılavuzu
Tüm pazarlama içeriklerinde tutarlı marka tonu sağlar.
HAZIR VE KOPYALANABİLİR
Prompt Kütüphanesi
305+ test edilmiş prompt şablonu; yazıdan koda, görsel üretiminden kariyere kadar. Değişkenleri ({böyle}) kendinize göre doldurup doğrudan kullanın.
Best way to use screenshots with ChatGPT & Claude
I paste a lot of screenshots into ChatGPT / Claude. Half the time it latches onto the wrong button, log line, or panel, then I have to explain too much via prompts. To save my time and get work done faster with AI & Screenshots, I made Lookhere to just make screenshot self-explanatory so I don't have to waste my energy in writing prompts. It’s a fast, distraction-free tool desi
From Loops to Graphs: The shift in Agent architecture
Saw this breakdown on X discussing Andrej Karpathy's perspective on agent architecture: "Two Autonomous Agent loops made Karpathy's loop 1000x better with Graph Engineering." The core idea is that moving from a single sequential execution loop to a connected graph topology gives agents vastly better context and task orchestration. From an engineering standpoint, this makes comp
How we get AI to write like a human after thousands of pieces of content (rules, tone scales, exemplars)
I run an agency and we've written thousands of pieces of content with AI at this point. The default writing style is kind of obnoxious with every model, and honestly it seems to have gotten worse as the models got smarter, not better. I think it's because the labs are all optimizing for software engineering now, so the models write like they're optimizing code. Shorthand, missi
I distilled Google's official Gemini agent guidelines into a battle-tested agentic workflow system prompt
If you have built autonomous agents or multi-step tool-calling workflows with LLMs, you have likely run into the standard failure modes that break production agents: Premature Action Bias: The model fires off tool calls or answers the user before mapping out prerequisites and the logical order of operations. Fragile Error Handling: When an API call fails or returns unexpected d
I found a prompt that turns 80 pages of course material into 20 pages of comprehensive notes - Norra NotebookLM
So long story short notebookLM has been a game changer and has helped me get good grades and save tons of time and while I was exploring this is the best workflow that helped me huge chunks of data into small comprehensive course Phase 1: Preparation and Segmentation Split the Module (Crucial Step): I upload my module (usually 300–400 pages) to ILovePDF and split it into indivi
i started putting wrong examples in my prompts. it works better than the right ones.
so i noticed something weird last week was building a prompt to classify support tickets. bug report vs feature request. standard few-shot, gave it 3 clean examples of each. worked fine on my test data then threw a real ticket at it and it got it wrong. "the export button is too slow, we need this fixed" - it called that a feature request. which, fair, it kind of is. but the cu
Most people tell AI what to do. Very few people show it what "good" looks like.
One of the easiest ways to improve AI outputs isn't writing longer prompts. It's giving examples. Instead of this: «Write a product description.» Try this: «Write a product description following this structure: - A short opening hook - Three benefit-focused bullet points - A professional but friendly tone - End with a clear call to action» Notice what's different. You're no l
Mythbusting ChatGPT "Secret Slash Commands" + Free Image Preset Keyword List
Lately, I've seen tons of posts hyping up "secret codes" for ChatGPT image generation that use slash commands (as if they're special prompt triggers). I tested them out both with and without the / prefix—and the results are identical. The slash prefix does nothing; they are just standard descriptive keywords being passed to the model. To save you the trouble of hunting them dow
Stop describing your tone to an AI writing tool. Make it reverse-engineer it instead.
Every "write in my voice" attempt I made used to be me listing adjectives. Casual but professional, confident not arrogant, and so on. The output always came back sounding like a brand guideline wrote it, because adjectives are not a style, they are a vibe. What actually works is making the tool extract the rules from your own samples first, then write against those rules. Two
The cheapest prompt optimisation I found this year was deleting the navigation from context entirely
Spent a while trying to trim prompts for an agent that pulls data off a few sites daily. Tightened instructions, cut examples, and compressed the system prompt. Marginal gains at best. Then I actually looked at the token breakdown, and the prompt was never the problem. Navigation was DOM dumps, screenshots, and the model reasoning its way to a button it had already located in t
We were optimizing output tokens to save money. Turns out 95% of our bill was input.
Analyzed a day of token logs across an autonomous coding agent setup running on internal codebases. The raw count: 769M input tokens vs 7.4M output tokens (~104:1). Because long agent runs re-read session history (files, AST diffs, test outputs) every turn, input costs accounted for ~95% of total spend. Optimizing output length turns out to be looking at the wrong variable. Thr
Duygulu TTS Marka Sesi Rehberi
Ses sentezinde tutarlı ve doğal marka tonunu uygulanabilir kurallara dönüştürür.
E-posta Ton Düzeltme
Yazılmış bir e-postanın tonunu ve netliğini iyileştirir.
E-posta Tonunu Yeniden Yazma
Var olan bir e-postayı istenen tonda ve uzunlukta yeniden yazar.