All products

Live · 14-day trial

Alongpad

A shared pad of priorities and notes for a studio and each of their clients, in each side's own brand.

One shared pad of priorities and notes, drawn twice: in AutomatedPanda's brand on the studio's side and in the client's brand on theirs, with the same five priorities and the same note on both.
Status
Live · 14-day trial
Built
SaaS, one pad per client
Stack
Next.js · Supabase · Stripe · TipTap · Vercel
Since
Nov 2026

The problem

My client work was scattered. Some of it was in Google Docs, some in email threads, and with one client the list of what to do next lived in a WhatsApp chat. Nobody could tell you what the current priority was without scrolling.

So I built a shared pad for me and that one client. A list of priorities you can reorder and tick, notes beside it, and both of us looking at the same thing. It worked well enough that I stopped reaching for the other three places, and after a few months it was obvious it shouldn't only be mine.

Two sides of the same pad

The thing I wanted and couldn't buy is that the client shouldn't feel like they're using my software. They open the pad on an address I've set up for them, and what they see is their logo, their colours and their fonts. I see mine. It's one workspace and one set of data, painted from whoever is looking at it.

That isn't a theme switch, it's one function the whole app renders through:

export function sidesFor(input: { isPrivate, role, org, client }): Sides {
  if (isPrivate) return { mine: org, partner: org };
  return role === "client"
    ? { mine: client, partner: org }
    : { mine: org, partner: client };
}

Every screen draws mine and partner, so there's no screen that can forget. Even the words flip: the section I call "their side" is the one the client calls "your side". And a client who belongs to one pad never sees the folder of clients at all, because they're redirected past it into the only workspace they have.

Isolation is the database's job, not the app's

This is the part I'd want to read about if somebody else had built it, so it's the part I did most carefully.

A studio's other clients have to be invisible, and "the app doesn't render them" isn't good enough for that. So the isolation is in the schema. Every policy goes through one membership check, and every child table (a task, a note, an attachment, a member row) keys on the workspace plus its own id rather than the id alone. That makes pointing at something in another workspace a foreign key that doesn't exist, instead of a rule somebody has to remember to apply. The migrations also revoke the defaults first and grant back only the columns the policies expect, so a signed-in select * on an organisation is refused rather than trimmed.

The decision I'm happiest with is a smaller one. On the top plan a studio can put the pad on their own domain, and a sign-in link is never sent to it. A customer controls their domain's DNS, so a dishonest one could point it somewhere else after it's verified and catch their own people's one-time sign-in links, and those people may be in somebody else's pad too. So links and tokens only ever land on an address I control, and the custom domain is for looking at once you're already signed in. It costs a bit of polish on the nicest part of the product, to close a hole that the customer it protects will never see.

What I built

The pad. A priorities list per project that both sides can reorder, own and tick, and notes beside it. Priorities update live, so a tick lands in the other person's browser without a reload.

Notes that take what you paste. Headings, tables and images survive a paste out of Google Docs. Links come in as cards, files and images attach to a task or a note, and comments anchor to the words they're about rather than sitting at the bottom. Every block is marked with who wrote it.

Brands as data. Upload a logo and the product cleans it, crops it, sizes it and takes a palette out of it. Every colour in the interface is then derived from one accent colour and kept readable whatever is chosen, and the brand reaches the emails, the tab icon, the installable app and the signed-out door, not just the pad. There's a live editor with the real pad rendered beside the form.

A team, and a desk. A studio has owners, admins and members. An admin sees every client, a member sees only the clients they've been added to, and the pads sit on a desk as one folder per client. A member can preview the pad through another member's eyes, which is look-only: the write methods are replaced with refusals rather than hidden.

Sides that never mix, including in the past. Someone who has ever been a client of a studio can't later join its team, and the other way round. Removed people count. If they came back under the other role, everything they'd already written would be relabelled.

Addresses. Every customer gets a free one on sign-up, added to the hosting one subdomain at a time rather than by a wildcard, and the top plan can use their own domain with verification.

Billing that can't be raced. Plan limits are triggers that lock the organisation's row before they count, so two writes at once queue up and the second one sees the first. Two tabs can't buy two subscriptions: the second is cancelled, and refunded automatically if it already charged.

Leaving properly. Download everything as one zip, or delete now. A deletion writes down everything that lives outside the database (stored files, addresses on the host) and a daily job works through the list until each one is actually gone, including leftovers from earlier runs.

What it costs

Solo is $15 a month, or $150 for the year. Studio is $39 a month, or $390 for the year, and adds your own domain and more of everything. Extra people on your side are $8 a month each. There's a 14-day trial.

Your clients never pay and never see a price. Up to 10 people per client on Solo and 25 on Studio.

The outcome

It went live this month, so there's nothing honest to say yet about how it's going.

What I can say is that I'm a customer of it. AutomatedPanda's own pad runs on the same code as everyone else's, on a plan row that marks it as internal, with its address handled exactly the way a paying customer's own domain is. That wasn't the plan originally, it's just what happened when the two-person app I'd built for myself got migrated into the multi-tenant one. The migration refused to run unless the database held exactly one workspace and one owner, which it did, once.

It's separate from Panda HQ on purpose. Panda HQ is where all my own financial records live, and Alongpad has other people's workspaces in it, so they're two products with two databases and no shared tables. Alongpad does report into Panda HQ, in the same shape GT Vault does, but that's a stats feed over a token, which is a very different thing from sharing a database.

It's live at alongpad.com.