Questions, answered without the marketing voice

You describe your product once. GrowthPilot builds a structured profile of it, researches the competitors you name, audits the SEO of a website URL you give it, proposes a prioritised growth strategy, and drafts the content to execute that strategy — briefs, posts, and landing copy. It is a research-and-drafting copilot: it produces a plan and a first draft, not a finished, published campaign.

It cannot buy or manage ads, post to your social accounts, send email campaigns, or measure real performance — there are no ad-platform or analytics integrations today. It cannot verify claims about your product that you did not give it, it cannot guarantee a ranking or a conversion rate, and it does not replace a marketer's judgement about what is true, legal, and on-brand for your business.

Founders, solo marketers, and small teams who need the output of an agency retainer without the retainer — people who know their product well but do not have time to do competitor teardowns, keyword research, and content production by hand every week.

A short written description of what you sell and who buys it. Everything else is optional and makes the output better: your website URL for the SEO audit, two to five competitor URLs for the competitive analysis, and your brand colours and tone for the content. You do not need to connect an account or install a tracking script.

No. Auto-publishing is on the roadmap, not in the product. Today you review the drafts in the dashboard and copy them out. We think that is the right default for a beta anyway — nothing generated by a model should reach your audience without a human reading it first.

It is a beta and we run it like one. Features change, jobs occasionally fail and need re-running, and there is no uptime guarantee or support SLA. If you need a system of record for your marketing, this is not that yet.

Groq is the primary provider, running open-weight Llama models — every generation is tried there first. Google Gemini is configured as the fallback and only sees a request when Groq fails or is unavailable. There is no embeddings provider and no vector database involved: your product profile is not indexed, it is sent as part of the prompt. Your prompts — which include the product descriptions, URLs, and competitor notes you supply — are sent to whichever of those providers handles the request, and are subject to that provider's terms.

We do not train any model on your content, and we do not fine-tune on customer data. We cannot make that promise on behalf of the third-party providers beyond what their own API terms say, which is why we name them rather than hiding them behind the phrase "AI-powered". If a provider's terms matter to your compliance posture, read Groq's and Google's directly before you paste anything sensitive into the product.

Yes, it can be wrong, and sometimes it will be confidently wrong. Language models invent plausible specifics — statistics, feature claims, competitor pricing, quotes. Treat every generated draft as a first draft from a fast junior who has not fact-checked anything. Verify any factual or comparative claim before you publish it, especially claims about a competitor.

From public web pages only — the URLs you supply, fetched the same way a browser or a search-engine crawler fetches them. There is no scraping behind logins, no purchased data broker feeds, and no access to a competitor's private analytics. Pages are summarised into positioning and messaging notes; source text is never copied into your generated content verbatim.

No. Give it what you would be comfortable putting on a public landing page. Unreleased roadmaps, customer lists, pricing you have not announced, and anything covered by an NDA do not belong in a prompt that leaves our infrastructure for a third-party model.

In a managed Postgres database hosted by Supabase, with row-level security so a project's rows are only readable by the account that owns them. Generated content, product profiles, competitor summaries, and audit results all live there. See the privacy policy for retention and deletion.

It means the product costs nothing today, has no card on file, and no checkout exists — there is literally nothing to pay for yet. In exchange you are using unfinished software with usage caps, and we ask for feedback when something is wrong. It does not mean free forever, and it does not mean a permanent free tier has been promised.

A daily cap on AI generation requests, enforced per account, so a single account cannot exhaust the shared provider quota. If you hit the cap, generation pauses until the next day; nothing is lost. The exact number moves as we tune capacity — the pricing page always shows the current one. The beta is also designed around a single project per account: nothing in the product is built for more than one, but there is no check stopping you from creating a second, and we would rather say that than describe a limit we do not enforce.

Not automatically, and not without notice. There is no payment method stored, so no charge is technically possible today. When paid plans launch we will announce pricing ahead of time and you will have to actively choose a plan — silence will never be treated as consent to bill.

No, and the pricing page does not pretend otherwise. Those tiers list the direction we are building in; the prices are not set and there is no way to purchase them. If one of them is what you need, tell us through the contact form and it will shape what we build next.

Yes. Everything generated is visible in the dashboard and can be copied out as text. A structured bulk export is a fair thing to want and is on the list; if you need it, say so and it moves up.

You do. We claim no ownership of your inputs or of the drafts produced from them, and we do not resell them. Be aware that generated text is not automatically eligible for copyright protection in every jurisdiction, and that a model can produce something similar for another user from a similar prompt — so review anything you intend to rely on commercially.

Your account email and name, whatever you type into the product (product descriptions, website URLs, competitor URLs), the content generated for you, and operational logs of AI usage. If you use the contact form we store the name, email, and message you send, plus a salted hash of your IP address for abuse prevention — never the raw address.

Yes. Email us through the contact form and we will delete the account and its project data. Deleting a project cascades to its profiles, competitors, audits, strategies, and generated content. Cached AI responses and aggregate usage counters are retained without account identifiers.

No advertising or third-party analytics cookies. The only cookies set are the Supabase authentication cookies that keep you signed in, plus a local preference for light or dark mode. The marketing pages you are reading now set no cookies at all until you sign in.

We do not sell it, and we do not share it for advertising. It is shared only with the infrastructure providers needed to run the service — Supabase for the database and authentication, Groq and Google for model inference, and the hosting platform — each acting on our instructions.

No, and they say so at the top. They are honest descriptions of what the software actually does, written by the team that built it, not documents reviewed by a lawyer. They must be reviewed by qualified counsel before this product is sold commercially.