Practical AI, engineered with careful prompting — research and writing tools, prompt engineering, and everyday utilities that just work. All free, all browser-based.
For students and researchers writing lab reports, essays, or a literature review — each tool works directly on your actual draft or data, not a generic template.
Tightens your lab reports and essays with discipline-appropriate scientific style.
Open →Map a research field fast — sources, themes and gaps organised for your review.
Open →The curiosity engine: upload a scientific image, receive a structured analysis.
Open →Your intellectual sparring partner — stress-tests a claim or idea before you commit it to paper.
Open →The Physics Solver Series: one problem solved by multiple methods, side by side.
Open →Compare fact-machine AI answers with mentor-style teaching — built for educators.
Open →19 hands-on tools grouped by what they actually teach — from planning a prompt before you write it, to seeing why AI fails where it fails, to building a public portfolio around what you can prove.
The 4D framework (Delegation, Description, Discernment, Diligence) turns "ask the AI" into a repeatable habit — decide upfront what it handles alone, what it doesn't, and what you're still accountable for.
Four short questions — Delegation, Description, Discernment, Diligence — turn into a complete, responsibly-scoped prompt. The same process behind my own AI-assisted blog writing.
Open →The live, twelve-question sibling of 4D Prompt Builder — your own Gemini key drives a Socratic interview that corrects vague or repeated answers, mirrors Description against Discernment, scans five cross-dimension gaps, and scores how solid each D actually is.
Open →Same four D's, different job — for students, not for finishing work. Configure a tutor role, pick study protocols, set your own honesty checkpoints, and get a reusable instruction block for any AI chat.
Open →Most bad AI output starts with a vague instruction, not a bad model. These tools scan your own prompt for the vague spots before you ever hit send.
Learn prompt engineering by doing — build, test and compare prompting frameworks.
Open →Paste a draft prompt and see, line by line, which instructions are checkable and which are just a hope — then fix the vague ones with a real AI-run prompt and build a refined version ready to copy anywhere.
Open →Turns your AI's own vague answer into 5 questions about what you left out — then folds your real answers back into a sharper prompt. Clarifies, never pads.
Open →Not tips and tricks — the actual mechanics. See where next-token prediction, similarity scoring, and long-context reasoning quietly fail, instead of just being told that they do.
Watch a Markov chain, an n-gram model, an RNN, and a transformer each guess the next word from the same sentence — and see where confident-sounding fabrication actually comes from.
Open →Nine ways to measure "how similar are these two sentences" — from cosine similarity to a bioinformatics sequence-alignment algorithm — run side by side so you can see exactly which one an embedding model actually uses, and why.
Open →Drag a session-length slider and watch a real blog paragraph quietly reword itself — specific, mid-paragraph facts break first, with no warning shown anywhere, illustrating the lost-in-the-middle effect directly.
Open →A real ~2,500-word test document with one true fact and two lookalike decoys — the full passage shown openly, a real copy-paste prompt to run yourself, and no invented score.
Open →Push faders toward the fine-tuning artifact you want to demonstrate, and get a prompt that biases an essay proportionally — see exactly what imbalance does to AI writing.
Open →The habit that actually matters once you're using AI for real work: a mechanical check before you believe what it just told you, calibrated to how much a mistake would actually cost.
Paste an AI answer and run it through five known compound-failure patterns every time — whether anything feels wrong or not, with a real verification prompt for anything flagged.
Open →Not a blog tool, and it can't make AI more accurate — self-report three properties, get a fourth scanned mechanically, and see exactly which verification habit your task's risk profile calls for.
Open →Have a document too long to trust an AI to summarize in one go? Split it into small overlapping windows, extract each one separately, then stitch the notes back into one summary — with a side-by-side comparison and a mechanical check for which facts actually survived.
Open →Maps a blog paragraph by paragraph before a single fact gets checked — what depends on what, what repeats itself, what's drifted from the point. Pairs with Authenticity Loop below.
Open →You pick which paragraphs to check. Each one gets verified against a real source and restated in your own words before you decide keep, correct, or flag. Nothing moves without you.
Open →A claim on a resume is cheap. These tools turn a claimed skill into checkable evidence, then diagnose why a post about it isn't getting read.
Turns a claim you'd make about yourself into actual evidence — a four-step wizard from raw skill, to a sharpened one-liner, to a sitemap-ready portfolio page, to verified proof it's real.
Open →Diagnoses the mechanical reasons a social post gets throttled — body links, hashtag spam, buried hooks — per platform, then hands you a cleaned version and a rewrite prompt.
Open →One topic, four dials — next-token style, knowledge depth, working memory, steerability. Build your own blog prompt and a shareability-optimized one, then get a real mechanical diff and a comparative-analysis prompt for both.
Open →Turns what you actually did with AI into a clear, honest disclosure statement — a short interview, a check for inflation on either side, then three ready-to-use versions calibrated to an assignment, a job application, published content, or internal work.
Open →Small, self-contained tools for anyone curious about a specific everyday AI question — no course or setup required.
Hear and practise tricky scientific pronunciation.
Open →Understand what separates an AI app from an AI agent.
Open →The intent-based search engine that thinks before it searches.
Open →For students who need more than a demo — a complete, multi-stage AI companion for figuring out what's next and proving you're ready for it.
A free four-stage tool: discover roles that genuinely fit you, prove it in a graded case study and mock interview, then get an honest final recommendation on where you actually stand a chance — evidence-based, not just your first guess.
Open →For small business owners with no legal team, no HR, no PR between them and the customer. Decide whether AI should even touch a task, describe it properly if it should, then run the result through a four-lens check before it goes out under your name.
Open →Pilot-test an AI support bot before it ever talks to a real customer. Ask it questions the way customers would, mark what it got right or wrong, and export a ready-to-use knowledge base, escalation rules, and system prompt for whatever you deploy.
Open →Turn your invoices, bank balance, and fixed obligations into a weekly, explainable vendor-payment recommendation — urgency-ranked, cash-flow checked, and always left for you to approve and send yourself.
Open →Turn what you've already decided about AI in your business into a one-page policy — what AI handles, what stays human, how you stay accountable — drafted with AI, edited by you, and stress-tested before you share it.
Open →For teachers, any subject. Design a course/unit blueprint, then generate lesson plans, weekly plans, study guides, slide decks, and assessments from it — all reviewed by you before they reach a student.
Open →