Free calculators and generators for AI, automation, websites and search
Twelve tools that run entirely in your browser: revenue models for missed calls and response time, cost models for AI receptionists, websites, agencies and consultants, a scored build-versus-buy decision, an AI visibility audit, and generators for schema, llms.txt and voice agent scripts. Every result is shareable by URL.
The twelve calculators, and what each one answers
Missed-call revenue calculator — We are missing calls and I want a number I can put in front of my partner. About 1 min.
Lead response time calculator — I want to know whether responding faster is worth changing how we work. About 2 min.
AI receptionist cost calculator — I want to know what a voice agent really costs before anyone quotes me. About 2 min.
Automation ROI calculator — Someone spends hours a week moving data between systems and I need to justify fixing it. About 2 min.
Website cost estimator — I need to budget a website and every quote I get is a different number. About 2 min.
AI agency cost estimator — I am about to receive agency proposals and I have no idea what normal looks like. About 2 min.
AI consultant rate benchmark — I need to know whether the rate I have been quoted is normal for the tier I am buying. About 1 min.
Build vs buy vs agency decision tool — We know what we need. We do not know whether to build it, buy it, or pay someone. About 3 min.
AI visibility self-audit — We rank on Google but never appear in the AI answer above it. About 5 min.
JSON-LD schema generator — I need valid structured data on my site and I do not want to hand-write JSON-LD. About 3 min.
llms.txt generator — Everyone says I need an llms.txt and nobody will tell me whether it does anything. About 2 min.
Voice agent script builder — I am about to configure a voice agent and I do not know what it should say. About 5 min.
Which one to open first
We are missing calls and I need a number for the conversation. → missed-call-revenue. It turns a vague sense of leakage into a monthly figure with the arithmetic published underneath it.
We reply to leads slowly and I want to know what that costs. → response-time-calculator. It prices the gap between two response times using your own conversion numbers rather than a statistic nobody can verify.
Someone is about to quote me for a voice agent. → ai-receptionist-cost. Knowing the five component costs turns a per-minute quote into something you can actually evaluate.
A person spends hours a week moving data between systems. → automation-roi. Payback in months is the number that decides this, and it needs a loaded hourly cost rather than a wage.
I need to budget a website and every quote is different. → website-cost-estimator. Seven build types with published bands, plus the maintenance most budgets forget.
I am about to receive agency proposals. → ai-agency-cost. Four different things trade under one label, and the bands for each are published.
I have been quoted a consultant rate and cannot tell if it is normal. → ai-consultant-rates. Six tiers from freelancer to MBB, with hourly, daily and retainer bands side by side.
We cannot decide whether to build this, buy it, or hire someone. → build-vs-buy. Seven weighted questions, all weights published, and it tells you which two answers decided your result.
We rank on Google but never appear in the AI answer above it. → ai-visibility-check. Six weighted dimensions and a prioritised fix list, with no email gate on the report.
My site has no structured data and I do not want to hand-write JSON-LD. → schema-generator. One connected graph with real @id cross-references beats three disconnected blocks.
Everyone says I need an llms.txt and nobody will say whether it works. → llms-txt-generator. It generates one and tells you plainly that 97% of them got no AI-bot traffic at all.
I am configuring a voice agent and do not know what it should say. → agent-script-builder. Greeting, qualification, booking and guardrails, assembled into a script and a system prompt.
Smaller utilities on the same shelf
Missed-call calculator (embeddable) — The three-input embeddable widget. Copy the iframe and drop it on your own blog. (30 sec)
AI website audit — Paste a URL and get specific AI-agent opportunities for that site. Server-backed, so it takes a moment. (30 sec)
Personalised agent demo — Name your business and industry, then talk to a live agent that knows who you are. (30 sec)
Voice script generator — Model-written dialogue script for your niche. The structured builder is the client-side alternative. (1 min)
AI vendor comparison matrix — Vendors against criteria, sortable and filterable. (3 min)
Production prompt library — Copy-paste-ready prompts for sales, ops, marketing and support. (browse)
Prompt critic — Paste a prompt, get it scored and rewritten. (30 sec)
LLM model selector — Four questions to the right model for a task. (45 sec)
Voice latency simulator — Hear what response latency does to a phone conversation. (30 sec)
Automation scenario generator — Industry and use case in, candidate automation scenarios out. (1 min)
Agent cost estimator — The original call-volume cost estimator. The receptionist cost model supersedes it. (1 min)
Lead qualification ROI — Hours and dollars reclaimed from qualifying leads by hand. (2 min)
How are these tools built, and where do the numbers go?
Every calculation runs in your browser. There is no server round-trip, no account, no email gate in front of a result, and nothing you type is transmitted anywhere. You can confirm that by opening your network tab while you use one. That is a design decision with a cost: it means the tools cannot save your work, cannot look anything up about your business, and cannot be smarter than the inputs you give them. In exchange, they are fast, they work offline once loaded, and there is no version of using them that turns into a sales call you did not ask for. Results are shared by URL. Every input is written into the query string as you type, so copying the address bar and sending it reproduces your exact model on someone else's screen. That is deliberately the only sharing mechanism, because it requires nothing from either of you.
Why does every tool publish its methodology?
Because an output nobody can reproduce is an assertion, and assertions are worth very little in a decision that involves spending money. Each tool publishes the literal formulas it evaluates, every assumption baked into it with the default it ships, and a list of things it cannot tell you. That last section is the one that matters most. A calculator that does not state its limits is inviting you to use it for something it cannot do. It also lets you find the assumption you disagree with, which is the fastest route to a useful conversation. Most disagreements about a calculator result are actually disagreements about one input or one assumption, and those are settleable in a sentence once they are visible.
Where do the benchmark numbers come from?
The cost tools publish observed 2026 market bands: per-minute component costs for voice agents, website build and maintenance ranges, automation build and retainer ranges, agency engagement bands by segment, and a consulting rate table from freelancer to MBB. Every one of them is presented as a band with a label saying it is a market observation. None is a TrainYourAgent quote, none is a promise, and none is a claim about what your project will cost. Where a tool shows a TrainYourAgent price, it reads it from the published service pages rather than restating it, so there is exactly one place in this codebase where our prices are written down. Where a number could not be verified, no number is published. That is why the response-time calculator ships without a conversion multiplier and asks for your own figures instead, and why the missed-call model exposes its recapture rate as a scenario you choose rather than a rate we assert.
Why is there no email gate on any of these?
Because a gated calculator is a lead form wearing a costume, and it converts a useful thing into an annoying one. The results, the breakdowns, the fix lists and the generated files all render immediately and completely. If you want to email yourself a copy there is an optional field on some of them, and the result works identically whether or not you use it. The commercial logic is straightforward. A visitor who runs a tool, gets a number, and acts on it without ever contacting us is a good outcome. A share link that lands in front of five people at a company we have never spoken to is a better outcome than a captured address attached to a report nobody opened. And if the number turns out to be large enough to act on, every service page has its price published on it and is one click away.