ResearchAudio

ResearchAudio / distribution kit

Put the working audit beside the claim.

Add a browser-based AI checklist or calculator to a technical article, documentation page, course, or resource library. Readers can inspect the protocol or change every assumption without leaving your site.

Read the complete publisher guide →

One field, then copy

Name the site carrying the widget.

This creates a readable source label for aggregate attribution. It is added to the link only; nothing is submitted or stored.

No plugin. No account. No API key.

Embed a free AI calculator in three steps.

Every widget is a hosted iframe with a tiny resize listener. Paste the generated code into a custom HTML block and the working tool adapts to its content.

  1. 01

    Name your site

    Replace “partner-site” above with a short publication or domain name so visits remain attributable.

  2. 02

    Copy one widget

    The iframe and resize listener copy together, so results stay visible without guessing a fixed height.

  3. 03

    Paste and publish

    Use a custom HTML block and keep the visible ResearchAudio credit. If a platform strips the resize listener, the iframe retains its conservative fallback height.

Works with
  • WordPress
  • Webflow
  • Ghost
  • Beehiiv
  • Static HTML
  • Auto-resizes after interaction
Evaluation / protocol audit

AI benchmark audit

Twelve checks for versions, scaffolds, budgets, grading, task distribution, uncertainty, artifacts, cost, and reproducibility.

Open full checklist →
Copy-ready iframe

Agents / authority boundary

AI agent security

Twelve controls for goal integrity, least privilege, approvals, isolation, tool provenance, secrets, audit trails, shutdown, and recovery.

Open full checklist →
Copy-ready iframe

Tokens / monthly budget

LLM API cost

Requests, input and output tokens, caching, retries, and attached per-request charges.

Open full calculator →
Copy-ready iframe

Inference / hardware fit

LLM GPU memory

Model weights, KV cache, runtime headroom, usable VRAM, and minimum GPU count.

Open full calculator →
Copy-ready iframe

Local inference / hardware fit

Local LLM GPU compatibility

Model size, precision, GPU capacity, usable memory, runtime reserve, and minimum device count.

Open full checker →
Copy-ready iframe

Local inference / model discovery

What LLM can I run?

Start with available VRAM and find the largest listed model-size floor at INT4, INT8, or FP16.

Open full finder →
Copy-ready iframe

Model repository / source lookup

Hugging Face VRAM + badge

Paste a public repository, calculate three source-backed weight floors, and copy a tracked passive model-card badge.

Open full lookup →
Copy-ready iframe

Context / concurrency

LLM KV cache

Exact cache memory per token, sequence, context window, concurrency level, and GQA savings.

Open full calculator →
Copy-ready iframe

Voice / resolved outcome

AI voice agent cost

Platform, telephony, STT, TTS, LLM, failed resolutions, and human handoffs.

Open full calculator →
Copy-ready iframe

Automation / business case

AI agent ROI

Failures, human review, recurring cost, implementation cost, savings, and payback.

Open full calculator →
Copy-ready iframe

Agents / loop reliability

AI agent loop diagnostic

Ten checks for termination, validation, recovery, budgets, escalation, and traces.

Open full diagnostic →
Copy-ready iframe

Economics / successful task

AI cost per successful task

Retries, failures, human review, API spend, and the loaded cost of a successful outcome.

Open full calculator →
Copy-ready iframe

Automation / task fit

AI task fit diagnostic

Test whether a task has a target, judge, error signal, bounds, cadence, and clear ownership.

Open full diagnostic →
Copy-ready iframe

Codex / project configuration

Codex CLI config.toml generator

Build project_doc_fallback_filenames, project_doc_max_bytes, and AGENTS.md-aware Codex settings.

Open full generator →
Copy-ready iframe

Codex / automation

Codex exec command builder

Compose codex exec --json commands with sandboxing, schemas, ephemeral mode, and Git checks.

Open full builder →
Copy-ready iframe

Tokens / cache economics

Prompt caching cost

Compare cache reads, writes, hit rate, total savings, and break-even usage.

Open full calculator →
Copy-ready iframe

Voice / latency architecture

Voice AI architecture latency

Model endpointing, transcription, fast and slow branches, TTS, and first-audio playout.

Open full calculator →
Copy-ready iframe

Live specimen

This is the actual widget.

ResearchAudio does not receive checklist answers or calculator inputs. Private inputs stay in the visitor’s browser; the Hugging Face lookup sends only the entered public model ID directly to Hugging Face’s public API. The frame includes a visible ResearchAudio credit, an attributable “open full tool” link, and the existing free-subscription path.

Published the widget?

Request a verified publisher listing.

Share the live page after the ResearchAudio credit is visible. We verify the implementation before adding it to publisher distribution materials, giving useful integrations another discovery path.

Submit the live embed for review →

Public page URLs only. Do not include private analytics, subscriber data, or credentials.

Publisher FAQ

Before you paste the code.

How do I embed an AI calculator on my website?

Enter your site name, choose a ResearchAudio tool, copy its auto-resizing embed code, and paste it into a custom HTML block on your website.

Which website platforms support the AI tool embeds?

Any platform that accepts iframe or custom HTML blocks can use the widgets, including WordPress, Webflow, Ghost, Beehiiv, and static HTML sites.

Do embedded ResearchAudio tools collect visitor inputs?

ResearchAudio does not receive checklist answers or calculator inputs. Private inputs stay in the visitor’s browser. The Hugging Face lookup sends only the entered public model ID directly to Hugging Face’s public API.

Can I embed the AI tools for free?

Yes. All seventeen widgets are free to embed in public articles, documentation, tutorials, courses, and resource pages as long as the visible ResearchAudio credit remains.

Use conditions

Free to embed. Keep the credit visible.

Use the widgets in articles, documentation, tutorials, courses, and public resource pages. Each tool remains hosted by ResearchAudio, so improvements appear without replacing the code.

Do not remove the ResearchAudio credit, place the tool behind a paywall, claim it as your own, or use it to collect sensitive inputs. These are planning instruments, not certifications, provider quotes, or financial advice.

For builders who show their work

Get the evidence behind the next AI claim.

ResearchAudio traces sources, costs, limitations, and adoption questions for engineers and product builders.