SEO & marketing · checked 2026-10-19

Can I vibecode Profound?

KINDA· weekend project

The loop at the centre is simple: send a fixed list of prompts to the AI engines every day, record whether and how your brand is named and which sources are cited, and chart the trend. Open-source trackers already do it, and an agent can build a personal one in a weekend on the engines' own APIs. The differences that matter are that API answers are not always what the chat apps show, that Google's AI Overviews and AI Mode have no official API, and that Profound's volume estimates, website analytics and AI Marketer rest on data and infrastructure you can't build. You get the tracking, not the market data.

variesprice

—you'd save a year

weekendbuild time

—people replaced it

Before you start

What you need on hand.

DifficultyIntermediate · Some system setup or keys.

TimeAn afternoon to build; a few weeks of daily runs before the trend means anything

Claude Code, Codex or CursorAny of them can run the prompt.
Node.js 22better-sqlite3 compiles a native module, so build tools must be installed.
API keys for the engines you trackEach provider bills separately; start with two engines and add more once the cost is clear.
A prompt listExport your Profound prompts and topics as CSV: Topic and Prompt are the required columns.
A machine that runs cronA laptop works if it's awake at the run time.

Before you buildProfound's 7-day Trial shows what the output looks like. For a tracker without a contract, try an open-source one such as Elmo before building.

1. Install one agent: Claude Code, Codex, Cursor.
2. Make an empty folder and open it in the agent.
3. Paste the prompt below, press enter, and answer its questions as it builds.

The prompt

Build me a personal AI-visibility tracker to replace Profound for one brand. Requirements:

- Node.js 22 + TypeScript: a daily CLI job and a Fastify dashboard (Eta pages, uPlot charts).
`npm start` serves http://localhost:4950; `npm run run-prompts` does one pass.
- Config in ./brand.yml (`yaml`, checked with `zod`): brand name and aliases, domain, up to 5
competitors with aliases, and 20 to 50 prompts, each with a topic tag and a country. Refuse to
run on an empty or duplicated prompt list.
- Engines behind one `Engine` interface, model names in .env: OpenAI (`openai`, Responses API with
the web search tool), Perplexity (Sonar, through the OpenAI-compatible base URL), Google Gemini
(`@google/genai`, Google Search grounding) and Anthropic (`@anthropic-ai/sdk`, web search tool).
An engine without a key is skipped, not an error.
- Run each prompt on each engine once a day (`node-cron`, 07:00 local; skip pairs already done
today; one retry). Store the full answer, the cited URLs and the model name.
- Analysis per answer: is the brand named (aliases, word boundaries), its position among the
brands named, which competitors appear, whether my domain is among the citations, and a
positive, neutral or negative label from one LLM call with a fixed rubric.
- Storage: better-sqlite3 in ./data/visibility.db. Keep every raw answer so the analysis can be
re-run when the rules change.
- Dashboard: visibility rate (answers naming the brand over all answers) per engine and per topic
over time, share of voice against competitors, a table of most-cited domains, and a page per
prompt with every answer for a chosen date.
- Budget guard: MAX_RUNS_PER_DAY in .env; print the estimated cost before the first run and stop
at the cap.
- Single user, one admin password in .env. No accounts, no telemetry, local-first: only the engine
APIs are called.
- Out of scope: Google AI Overviews and AI Mode (no official API), prompt-volume estimates,
website traffic attribution, content-writing agents.
- README: cost per run for each engine, a warning that API answers can differ from the chat apps,
and that answers vary between runs, so compare weekly averages and not single days.

Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.

Share on X

Max MiskowSponsoredAI agents, automations and MVPs, built for your company.Visit site ›

Then keep going

Next prompts, same folder.

The first prompt builds the core. Run these one at a time in the same folder to grow it into your own Profound.

Add a weekly summary to the AI-visibility tracker in this folder. Requirements:

- Every Monday at 08:00 send one email through nodemailer SMTP from .env.
- Compare the last 7 days with the 7 before: visibility rate per engine and topic, brands that
  appeared or disappeared, and domains that entered or left the citations.
- List the 5 prompts where visibility fell most, with a link to each prompt page.
- Use weekly averages only; say "not enough runs" when a prompt has fewer than 4 answers.
- Update the README.

Add a citation gap report to the AI-visibility tracker in this folder. Requirements:

- A page and a CSV export listing URLs cited in answers that name a competitor but not me,
  grouped by domain and topic, with how many answers cited each.
- Mark domains I already appear on, and domains that look like my own properties from brand.yml.
- Rank by citation count across the last 30 days and show the prompts that triggered them.
- Never fetch the cited pages; use only the stored data.
- Update the README.

Add prompt suggestions to the AI-visibility tracker in this folder. Requirements:

- Import the top 200 queries from Search Console (`googleapis`, service account JSON path in
  .env) for the last 90 days.
- Ask the configured LLM to rewrite each as a natural question a customer would put to an AI
  assistant, keep the topic tag, and drop duplicates by meaning.
- Show suggestions in a review page; only ones I accept are appended to ./brand.yml.
- Never edit brand.yml without a confirmation step.
- Update the README.

What you lose

Prompt Volumes: estimates of how often people ask a question, drawn from real prompts
Up to nine engines tracked the way a user sees them, including Google AI Mode, Copilot and DeepSeek
Tracking of AI-sourced traffic and attribution across your domains through Google Analytics
The AI Marketer and agents that draft content and stage changes in your CMS
ChatGPT Shopping tracking, SSO and a SOC 2 report for enterprise buyers

Why people still pay

Large brands pay for a measured view of how AI answers describe them across engines, regions and languages, backed by volume data and a team that turns findings into content changes. At that scale the fee sits next to an agency retainer, and the value is a number leadership trusts, which a home-built tracker with a few engines can't claim.

Moatproprietary data infrastructure scale integrations

Leaving Profound

Take your data, then cancel.

1 · Take your data out of Profound

1Enterprise lists CSV and JSON exports and an API; the Trial has no export or history.
2Prompts and topics can be exported as CSV or JSON from the Prompt Designer (the docs also describe uploading a CSV with Topic and Prompt as required columns and Tags, Regions and Language as optional ones).
3Citation data exports as CSV or JSON, and Prompt Volumes bulk keyword results export as CSV.
4We found no help article on exporting a full answer history: ask your account team how raw answers are delivered.

2 · Cancel Profound

1Enterprise is a sales contract: renewal and notice terms sit in your order form, and we found no public cancellation article.
2Agency Growth is bought online with consolidated billing; no cancel steps are published in what we read.
3Ask your account contact in writing for the notice period and the last day of access.

All-time history is an Enterprise feature and nothing we read says how long data is kept after a contract ends, so export before the last day.
Keep the prompt list: the CSV format above is a convenient shape for a tracker you run yourself.
Prompts refresh once a day on Eastern Time and new ones need 24 to 48 hours to gather data; expect a gap when you move to a new tool.

Checked 2026-10-19 against Profound's docs: 1, 2, 3

The escape hatch

Don't feel like building it? These are already free.

ElmoOpen sourceMIT, self-hostable AI visibility platform that tracks mentions and citations across ChatGPT, Claude, Perplexity, Gemini, Copilot, Grok and AI Overviews, with a live demo.★ 470 · active Oct 2026 · docker to self-host
WebVisit ↗GEO/AEO TrackerOpen sourceMIT, bring-your-own-keys dashboard for ChatGPT, Perplexity, Gemini, Copilot and Google AI Mode; its collection runs through Bright Data's paid scraper API.★ 299 · active Oct 2026 · one-click install
WebVisit ↗OneGlanceOpen sourceMIT tracker that reads ChatGPT, Perplexity, Gemini, Claude and Google AI Overview through their real product UIs rather than the APIs.★ 222 · active Oct 2026 · docker to self-host
WebVisit ↗

All 4 free alternatives to Profound ›

The numbers

Profound pricing

Plan	Monthly	Annual, per month	What you get
Trial (brands)	$0	$0	50 unique prompts run daily for 7 days, 3 engines (ChatGPT, Gemini, Google AI Overviews), 1 language and 1 region, limited AI Marketer credits, unlimited seats.
Enterprise (brands)	custom	—	Up to 9 engines, a tailored prompt plan, daily tracking, all-time history, CSV and JSON export, API, SSO, SOC 2, ChatGPT Shopping, dedicated support.
Agency Growth	$399	—	10 pitch workspaces a month for prospect audits; each full client workspace tracks 100 prompts daily with 400 agent credits a month; 5 agency seats; no API.
Agency Enterprise	custom	—	Tailored client and trial workspaces, a dedicated agency partner, premium Slack support, API.

Free tier: 7 days, 50 fixed prompts, 3 engines, no data history, no exports.
Billing: Brands have no self-serve paid plan: after the 7-day Trial the only route is an Enterprise quote through a demo, so terms sit in the contract. Agencies can buy Agency Growth online with consolidated billing. Older reviews quote $99 Starter and $399 Growth brand plans; the current page no longer lists them.
Hidden costs: AI Marketer and the agents run on credits that Enterprise sets with sales, and the pricing FAQ has entries on going over a credit allotment. Extra languages, regions and engines are part of the tailored Enterprise package. Prompt Volumes is sold through the sales team.
Source: www.tryprofound.com

Also on the list

Semrush$139/mo · SEO & marketingNOT REALLY

Ahrefs$29/mo · SEO & marketingNOT REALLY

AccuRanker$224/mo · SEO & marketingKINDA

Questions

Profound, briefly

Not answered here? Tell us and we'll add it.

Ask a question ›

A free Trial runs 50 prompts a day for 7 days on ChatGPT, Gemini and Google AI Overviews, with a fixed prompt set and no history, export or API.

Kinda. The core is a weekend build with the prompt on this page, but real gaps remain: Prompt Volumes: estimates of how often people ask a question, drawn from real prompts, Up to nine engines tracked the way a user sees them, including Google AI Mode, Copilot and DeepSeek.

Profound's pricing is usage-based or varies by plan.

Prompt Volumes: estimates of how often people ask a question, drawn from real prompts; Up to nine engines tracked the way a user sees them, including Google AI Mode, Copilot and DeepSeek; Tracking of AI-sourced traffic and attribution across your domains through Google Analytics; The AI Marketer and agents that draft content and stage changes in your CMS; ChatGPT Shopping tracking, SSO and a SOC 2 report for enterprise buyers. If any of those matter to you, keep paying.