TGH Tech
For founders · ProductPilot · Free beta

Building isn’t the hard part anymore. Knowing what to build next is.

ProductPilot is a tool for founders who are unsure what to build, fix or leave for later. You tell it where you are. It works through one decision with you, keeps what is known apart from what is a guess, and recommends a next step. It is built by YE Stack, the group we’re part of. Free beta: no card, no call, no obligation to hire anyone.

Who it’s for
Founders with an idea or an AI-built first version, and nobody experienced to ask
What you get
A structured review of one decision: what it found, your options and a recommended next step
What it costs
Nothing while it is in beta. No card.
Where it lives
productpilot.work
Who is behind it
YE Stack, the group TGH Tech is part of
Status
Free beta. Some parts are still planned. This page says which.

[01]Where you areProductPilot

It starts from the decision you’re stuck on. Not from a blank chat box.

Six moments ProductPilot was written for. They are situations, not customer quotes.

  1. 01

    I built it over a weekend with AI. Someone wants to pay for it. Now what?

    The decision underneath: launch carefully, stabilize it first, or rebuild.

  2. 02

    It works when I use it. I’m scared real users will break it.

    What blocks a launch, and what can safely wait.

  3. 03

    My co-founder wants to rebuild it in a “proper” stack.

    Launch, stabilize, restructure the code or rebuild, compared on time and risk.

  4. 04

    A developer quoted me a fortune and I can’t tell if it’s fair.

    The smallest first version worth building, and what to build first.

  5. 05

    I have 40 feature ideas and no idea which one comes first.

    Which features to keep, which to delay and which to cut.

  6. 06

    I don’t know if I need a technical co-founder.

    What to prove before you build, and the smallest thing worth testing.

AI tools can now get you working software without an engineering team. That is not the same as a product real customers can rely on. In between sit decisions about scope, risk, order and readiness, and a first-time founder often has nobody experienced to make them with.

[02]A first sessionProductPilot

Open it tonight. Here is exactly what happens. A few questions in. A reasoned first review out.

ProductPilot calls itself a navigator, not an autopilot: it shows the next move, and you decide. This is the session as it runs today.

  1. 01You

    Sign up

    Email, Google or GitHub. No card. Today the account comes before anything else; a preview before sign-up is planned.

  2. 02You

    Say where you are

    Pick your stage, from shaping an idea to growing.

  3. 03You

    Describe the product

    Its name, what it does and who it is for. ProductPilot reads each answer, replies, and offers example answers to tap if you’re stuck.

  4. 04You

    Add what you have

    Optional. A pitch deck, spec or research notes, up to 10 MB each. Or connect GitHub, read-only, to import your README and project files.

  5. 05ProductPilot

    “Here’s where I’d start”

    Three reviews, in order, with a reason for each and the facts it relied on. It will not claim to have read code you haven’t connected.

  6. 06ProductPilot

    Run the first review

    A few minutes. It comes back as a structured, mostly visual report: what it found, the options, the trade-offs and a recommended path.

  7. 07You

    Question it

    Ask follow-ups in the same thread. Push back where you know something it doesn’t.

The report says what it knows, and what it doesn’t.

A review is not a wall of chat. It keeps what you’ve told it apart from what it can see in your files, and a fixed check rejects any review that upgrades a guess.

  • What it found, and what each point rests on
  • The options, with the trade-off for each
  • A recommended path, and the reasons for it
  • A warning when a qualified person should check first
Launch readiness · first read
BackupsYou’ve told us backups run. Nobody has tried a restore.Blocking
Sign-inWorks for you. Not yet tried by a second real user.Blocking
PaymentsThis touches customer money. Get a qualified review before launch.Needs a person
Dark mode, referralsNeither one stops a first customer.Can wait
Recommended pathStabilize first. Rebuilding would push your first customer back by weeks.Your call
Illustrative. A made-up product and findings. Real reports are more visual.

The five reviews you can run today

ReviewThe question it answersWhat it looks atApp’s estimate
Architecture reviewWhat breaks first as usage grows?Your brief and any project files you’ve synced. With no repository synced, it must say it is a hypothesis, not a code review.About 6 min
Next best stepsWhat should you build, fix or drop this week?Reliability, onboarding, evidence and new features, ranked against each other.About 3 min
Launch readinessLaunch carefully, stabilize, restructure the code, or rebuild?Sign-in and access, payments, error handling, analytics, backup and recovery, support and rollback.About 5 min
MVP scope cutWhich features can wait without hurting the core?Your feature list, sorted into keep, delay and cut.About 4 min
Simulated user reviewHow might real users react, and what would stop them?Five AI personas react separately and their disagreements are kept. Marked as not validation. Ends with one real-world test to run.About 8 min

Names and timings as shown in the app, November 2026. A sixth, the tech stack audit, is listed but not available yet.

[03]Who it’s forProductPilot

Built for two kinds of founder. Neither has a product or tech lead to ask.

  1. Built with AI

    You have a working first version, and a first customer in sight.

    Your question: can real customers use this, what needs improving, and what can wait? ProductPilot separates what blocks a launch from what can wait, and compares launching carefully, stabilizing, restructuring the code and rebuilding. It is designed not to default to a rebuild or an immediate launch.

  2. Still an idea

    You know the problem. You haven’t built anything yet.

    Your question: what should I build, should I test the idea first, and how do I work with technology at all? ProductPilot works on whether the idea should become software yet, the smallest thing worth testing, and what not to build, before you spend on developers.

A founder with early customers can use the same tool for a different question: which one bottleneck to fix this week, and what to ignore. If you already have an experienced CTO and a product lead, you probably don’t need it.

[04]Not a chatbotProductPilot

A general chatbot answers the question you asked. ProductPilot is built to find the one that matters.

Row by row, with what is live today and what is still planned.

A general AI chatProductPilotToday
Answers the question you askHelps you find the question you need to ask: it recommends where to start, and says whyLive
May start over without your earlier contextKeeps your product, files and past reviews, and reads them on every runLive
Gives information and recommendationsKeeps what you’ve told it apart from what it can see in your filesLive
Produces answers in a conversationStructures the decision: options, trade-offs and a recommended pathLive
Aims for a useful replyAims for a decision you act on: records what you chose and why, then checks backPlanned

The last row is the design, not the product yet: today the loop stops at the report.

Its own answer to “I’ll just ask a chatbot.”

“Do. Then notice whether it remembers your product next week, or tells you what it doesn’t know.”

[05]Why we send you thereProductPilot

Most founders don’t need a team yet. They need to see the next decision clearly.

So before a call, we point you to a tool. Here is why, and what we get out of it.

Not every founder who reaches us is ready to build. Some are still deciding what to build, whether what they have is ready, or whether they need anyone at all. A free tool that helps you decide well is more useful to you than a sales call.

Who makes it, and what we get from it. ProductPilot is built by YE Stack, the group TGH Tech is part of. The way it thinks through a decision comes from our work: 50+ products shipped with founders since 2020. We have a commercial interest here: some founders who use it will later want a team to build with, and we hope they think of us.

What using it commits you to. Nothing. No card to enter, no call to book, no obligation to hire anyone. It is meant to stay fully useful to a founder who never buys anything from us.

What it will often tell you. To do the next step yourself. Or not to build that yet. Most of what it recommends is something you can do on your own, with AI tools, or with the developer you already have. When that is the right answer, you would hear the same from us on a call.

Before

No pitch. No chasing.

  • Nobody from TGH Tech calls or emails you because you signed up
  • Your product details are not passed to our team without your explicit agreement
  • You are not cut off halfway through a decision to push a paid plan or a call
Now

You get a review, and it is yours.

  • Your reviews stay under your product for when you come back
  • You can copy any report and show it to anyone
  • You can delete a product, or your whole account, at any time

[06]After the reviewProductPilot

Four ways to act on what it tells you. We are one of them, and we’ll say so every time.

ProductPilot’s rule is that outside help is never the only route shown. Here are all four.

  1. 01

    Do it yourself

    Often the whole answer. Test a restore from your backup. Have a second person try to sign in. Decide what not to build.

  2. 02

    Use AI tools

    Keep building with the tools that got you this far. The review gives you a next action to hand them, and flags areas such as payments, sign-in and personal data for a second look.

  3. 03

    Use your own team

    Already have a developer? ProductPilot helps you decide what to ask them for, and how you’ll both know it’s done.

  4. 04

    Get outside help

    A freelancer, a technical co-founder, an agency, or us. With us it usually means a Foundation Sprint, then a live product in 8 to 12 weeks. That route is ours and we earn from it, so weigh it against the other three.

Planned, not live.

ProductPilot is designed to show these four routes side by side, with cost and time for each and our team labeled as ours. That screen is not built yet, so today the route is yours to work out.

If you do come to us, bring what ProductPilot gave you. You can copy any report today. With your say-so, our first call starts from that report, not from a blank page. A built-in handover, with a screen showing exactly what is shared and with whom, is planned. If you never come to us, nothing changes.

[07]More detailProductPilot

How it reasons

Fixed rules around the AI, written down.

For the founder who asks why they should believe it.

RuleWhat it means for you
What you say is recorded as reported, not verifiedIf you tell it sign-in works, a review treats that as “you’ve told us”, not as “it works”.
It cannot raise its own confidenceIt can mark a point as: we don’t know yet, you’ve told us, or we can see it in your files. “Checked by an expert” and “tested and confirmed” can’t be awarded by you or by the AI.
It is told not to invent measurementsIt is told to use only what you supplied. If a number in a report surprises you, ask where it came from.
Every review is checked before you see itFixed rules in code check each result: no evidence label raised, options and warnings present. A failure gets one repair attempt. If it still fails, it is rejected, not shown.
High-stakes topics are flagged for a personWhen a decision touches security, payments, personal data, health or children, it tells you a qualified person needs to check first.

The thinking comes from decisions and mistakes we have seen again and again across 50+ products shipped with founders since 2020. That is where the approach comes from, not a guarantee that every recommendation is right.

Live and planned

What works today, and what is still on the plan.

As of November 2026.

PartStatusDetail
Sign-upLiveEmail, Google or GitHub.
Guided setupLiveStage, name, what it does, who it is for, files, sources.
A recommended first reviewLiveThree reviews in order, with a reason for each.
Five reviewsLiveReports can be copied. Follow-ups in the same thread.
Documents and codeLivePDF, Word, text, Markdown or CSV, up to 10 MB. GitHub, read-only, synced when you choose.
Saved products and review historyLiveDelete a product, or your whole account, at any time.
A first read before sign-upPlannedToday the account comes first.
Recording your decision, a next action and a check-back datePlanned nextToday the loop stops at the report. Weekly check-ins and a product plan follow.
The four routes compared in the product, and a handover with consentPlannedSee “After the review” above.
Export and sharingPlannedToday you can copy a report.
Tech stack audit, and Jira, Linear or Bitbucket connectionsPlannedNot available yet.
Paid plansPlannedFree and unlimited during the beta.
Common questions

Answered the way ProductPilot answers them.

Adapted from ProductPilot’s own answers to match what is built today. “We” here means ProductPilot.

QuestionProductPilot’s answer
Do I need to be technical?No. It asks plain questions. If you connect a GitHub repository it has more to go on, but you don’t need to understand the code yourself.
What if it gets my product wrong?Correct it. Edit your product’s details and run the review again.
Who can see my idea?Your workspace is kept separate from every other founder’s. No one outside ProductPilot is shown your product details unless you explicitly agree. To run a review, your brief and file excerpts go to the AI providers that process it.
Is it free?Right now, yes. Every review is free and unlimited while ProductPilot is in beta. Paid plans will come later, and we’ll tell you before anything changes.
I’m too early for this.Early is the cheapest time to decide what not to build.
Will it try to sell me development services?The part that would, a comparison of your routes, is not built yet. The design: outside help appears only if your situation calls for it, as one of four options, labeled as ours, with nothing shared without your OK.
What it is not

Six things ProductPilot does not do.

ProductPilot is notWhat that means
An app builderIt helps you decide what to build, fix or leave for later. It does not build or deploy anything.
An auditIt is told never to claim an audit, or to call your product secure, compliant or production-ready.
Legal, tax or fundraising adviceIt flags where those touch a product decision and suggests qualified advice.
Customer researchSimulated users raise questions. Real customers answer them.
Proven yetIt is designed to help you decide better.
A way into a sales callNothing in it requires talking to us.

What is the one decision you’re stuck on right now?

Take it to ProductPilot tonight. It is free while in beta, and what it gives you is yours. If you’d rather talk it through with people, we’re here for that too.