trending
•

2d ago

🤖 How do I build Swingback alone with AI agents?

I started Swingback about 2 months ago, and most of the code is written with AI agents (mostly Claude Code). But it only works because of 3 things:

  1. A plan before any code. Every part of the product is written down first and split in phases, and the agent stops at the end of each phase so I can review it.

  2. One file of rules the agent reads before it starts. For example: every query has to filter by workspace, and internal notes must never reach your users.

  3. Tests that check those rules. One test goes through everything your users' portal can show and fails if an internal note appears. Another one fails if the AI features can ever send an email or touch GitHub. So even if the agent forgets a rule, the tests catch it.

•

2d ago

🤯 How do I ship 5 SDKs on my own?

How do I ship 5 SDKs on my own?

•

2d ago

⚙️ Why did Postgres get 100x slower on the 6th run?

Why did Postgres get 100x slower on the 6th run?
This one took me a while to figure out. After a load test on staging (10,000 workspaces), one query stood out: the trending list took 155 ms a page instead of 1.3 ms, and it was using 60% of the database's time.
Here is what was happening:

  1. The driver I use (pgx) prepares every query, and PostgreSQL plans a prepared query with its real values for the first 5 runs.

  2. On the 6th run, it can switch to a "generic" plan, made without the values, if that plan doesn't look more expensive.

  3. My lists have optional filters (board, status, tags...). Without the values, Postgres expected to find a single post, so it went through every post of the workspace and sorted them all.

The fix was one setting: plan_cache_mode = force_custom_plan. Every query is now planned with its real values, and it costs a fraction of a millisecond.
Have you ever been bitten by this one?

•

2d ago

❓Why did I built yet another feedback tool?

That's a good questions, for me it really came down to 3 things.

  1. I don't like unpredictable pricing, so all platform that have a pay-per-user plan where out of the equation for me.

  2. Most, if not all, alternative that provide a free tier make it almost unusable (example: no auto-signin for your user which add so much friction, even I wouldn't post when I had an issue testing the app).

  3. Only the entreprise platform would provide integrations with GitHub or Linear, thus making their plan really expensive for solo builders.

For me it just comes down to, I want something I can quickly setup for a MVP, don't have to pay a cent and if it grows, then I can pay a reasonable price.
Do you think the pricing is reasonable? Is there something stopping you from using the free plan for your MVPs?

•

2d ago

❤️ What features I am most passionate about?

When I started building Swingback, about 2 months ago, I had 3 main requirements:

  1. It needs to sync with Github and Linear.

  2. It needs to support auto-translate (I live in Switzerland where there is 4 national languages).

  3. It needs to be as frictionless as possible, your users easily get frustrated if there is too many steps for them to give you some feedback, so I made it as direct as possible for your users.

Do you think there is something missing?

•

2d ago

💎 Why Swingback's Pro plan is one flat price

Pro is $20 a month, or $216 a year. Not per seat, not per tracked user, and there's no "contact sales" tier.
The free plan isn't a trial either: unlimited customers, posts and votes, the portal, roadmap and changelog, the in-app widget and one integration, with 3 team members and 3 boards.

Why I went flat:

•

2d ago

⚔️ How Swingback keeps GitHub and Linear in sync without the two sides fighting

The feedback board was the easy part of building Swingback. The hard part was getting a post and its GitHub or Linear issues to agree, in both directions, without one side undoing the other.
Here's what the sync does, and some problems I had to solve along the way.
What it does:

  • From a post, you open an issue in GitHub or Linear, or link one that already exists. One post can link several issues, across repositories and both trackers.

  • When the issues move, the post follows. Close the last open issue and the post is marked completed, and everyone following it hears about it.

  • When you change the post's status, Swingback can update the issues too: their state and a label on GitHub, their workflow state on Linear.

Some problems I had to solve:

  • Webhooks can't be trusted on their own They arrive late, out of order, twice, or not at all. So Swingback doesn't act on what a webhook says. It records that something changed, then asks GitHub or Linear for the issue's current state and works from that. A delivery that arrives twice, or after a newer one, changes nothing.

  • GitHub limits how much an app can create per hour, so writes are paced for each installation. Creating an issue is also safe to retry. On GitHub, Swingback marks what it creates and looks for that mark before trying again. On Linear, it gives the issue an ID of its own, so a second attempt is refused instead of making a duplicate.

•

2d ago

💸 Votes or revenue: how do you decide what to build next?

Here's a trade-off I keep running into.
Request A has 200 votes, mostly from free users. Request B has 5 votes, from three of your biggest customers.

Which one do you build first? How do you get this data (by hand, another tool)?
In Swingback I ended up showing both side by side: every request has its vote count and the combined value of the customers asking. It doesn't make the call for you, but it puts the trade-off in front of you instead of in a spreadsheet.

•

2d ago

📢 Do you tell users when their feature request ships?

When you ship something a user asked for, do they ever find out? How?

From what I've seen, it usually goes one of three ways:

  1. Someone remembers to reply in the original thread, usually only for the loudest customer.

  2. It goes in the changelog, and you hope the people who asked happen to read it (and like my dad always said, hope is not a plan ;).

  3. Nobody hears anything, and the person who asked six months ago assumes you ignored them.

•

2d ago

Swingback - Turn user feedback into GitHub and Linear issues

Collect feature requests and votes in a portal or in-app widget. Turn requests into GitHub or Linear issues. When an issue closes, Swingback marks the request done and emails every voter. AI flags duplicates. Free plan. Pro is a flat $20/mo.