This week: filters beat momentum (2026)

This issue covers the #buildinpublic window from Jun 29, 2026 at 08:33:58 to Jul 6, 2026 at 08:00 in the channel timezone (Etc/GMT+5). The window is slightly shorter than a full Monday-to-Monday week because the previous issue published a little after schedule.

The strongest pattern was not raw shipping volume. Posts earned attention when they gave other builders a filter: a way to ignore distractions, test validation, verify user trust, or decide whether a product update was worth saving.

X rewarded posts that packaged concrete product proof into a demo or metric. Indie Hackers rewarded posts that named the uncomfortable founder problem precisely enough for other people to argue with it. The useful move for your own buildlog is simple: before posting an update, decide what uncertainty the reader can resolve after reading it.

Standout post	Platform	Engagement	What drove it	Copyable move
Dmytro Chuta / Subscription Day and Toplify	X	137 likes, 109 bookmarks on Subscription Day; 82 likes, 75 bookmarks on Toplify ASO tracking	Bookmark-heavy demos turned product utility into a save-for-later object. 1 2	Lead with the user job and remove one setup step from the promise.
Sonu Goswami / attention protection	Indie Hackers	41 upvotes, 101 comments	The post gave founders a decision filter for AI-model and growth-hack distractions. 3	Turn a vague productivity warning into one yes/no question.
DanialPG / polite agreement trap	Indie Hackers	30 upvotes, 119 comments	Validation failure became a community diagnosis problem, not a private shame spiral. 4	Ask whether praise predicts behavior before calling a problem validated.
Silker AI / contract plus metrics	X	42,203 views on a 601-follower account	A tiny audience got large reach by pairing a B2B contract with usage numbers. 5	Stack one commercial proof point with one usage proof point.
Ava Bagherzadeh / AIApplyd verification	Indie Hackers	7 upvotes, 25 comments	The product update reframed the category around trust, not automation. 6	Make the hard-to-fake confirmation step the product story.

1. Dmytro Chuta: bookmarkable demos beat generic progress

Dmytro Chuta is a Ukraine-based solo builder with 4,032 followers and 16 years of design experience, operating the Appps brand across Subscription Day and Toplify. His strongest post this week showed Subscription Day, a visual calendar for recurring payments with reminders before upcoming charges. The post earned 137 likes, 109 bookmarks, 7,407 views, 5 replies, and 3 retweets on Jul 5. 1

The engagement signal went beyond likes. The post had a 79.6% bookmark-to-like ratio, which points to save intent: readers were treating the demo as something they might use or reference later. 1 The copy explained the job directly: "Keep all your recurring payments in one place with a visual calendar. Get timely reminders about upcoming payments, so you never get caught by unexpected subscription charges you meant to cancel." 1

Forty minutes later, Chuta posted a Toplify demo for App Store ranking tracking. The promise was unusually low-friction: paste an App Store link, with no sign-up and no Apple Connect integration. The post earned 82 likes, 75 bookmarks, 7,891 views, and a 91.5% bookmark-to-like ratio, the highest ratio in the tracked X set. 2

Chuta also posted a Toplify globe-rotation experiment based on beta-tester feedback on Jul 1. That post earned 111 likes, 92 bookmarks, 7,484 views, and an 82.9% bookmark-to-like ratio. 7 The pattern across the week was consistent: each post made the product legible as a small, finished user job.

Why it worked: The posts did not ask readers to admire effort. They showed a specific workflow and removed ambiguity around setup. "No sign up required. No Apple Connect integration needed. Just paste your app's App Store link to start tracking it" is more copyable than a generic launch announcement because the sentence eliminates objections while explaining the product. 2

Copy this: Write your demo post as "job + missing friction." Example: "Track invoices across clients without connecting your bank" is stronger than "I built an invoice dashboard." If your post can truthfully remove a setup step, put that in the first two lines.

2. Sonu Goswami: one filtering question created the discussion

Sonu Goswami, a B2B SaaS positioning specialist, posted the top Indie Hackers item of the week in the Saas Makers group on Jul 2. The post earned 41 upvotes and 101 comments. Its core argument was that protecting attention from shiny distractions such as new AI models, frameworks, and growth hacks is now harder than building product. 3

The post's strongest line was a filter founders can apply before changing their roadmap: "Does this solve a problem I already have, or did it create a problem I didn't know I had?" 3 That is why the post traveled in comments. It did not merely say "focus matters." It gave readers a sentence they could use the next time a model launch, framework thread, or growth tactic made them feel behind.

The argument also hit a real build-in-public behavior: a public thread can make a founder feel productive while avoiding a direct conversation with a person who has the problem. Goswami framed the dangerous part as quietly rebuilding the roadmap around whatever got posted that week. 3

Why it worked: Indie Hackers discussions tend to reward posts that turn a private founder anxiety into a shared decision problem. This post gave people something to confess, challenge, and apply.

Copy this: If you are posting a lesson, end with a diagnostic question instead of a slogan. A good filter has two sides and forces a real choice: "Does this unblock a customer problem I already saw, or did I just inherit a new toy's agenda?"

3. DanialPG: validation realism became the week's comment engine

DanialPG, founder of ReqBrief, returned to Indie Hackers after two weeks of silence with the week's second-highest discussion post. The Jun 30 post earned 30 upvotes and 119 comments. The problem was blunt: people agreed his problem was real in conversation, but they still would not try the solution. 4

DanialPG named the failure mode as "polite agreement." His phrasing was strong because it separated social validation from behavioral validation: "Maybe agreeing a problem is real is just a cheap thing people do to be nice in a conversation, and it has almost nothing to do with whether they'd ever change what they actually do about it." 4

The post also asked a high-quality community question: how do you distinguish a real business with bad distribution from a vitamin product people praise but never buy? 4 That question is why the comment count matters. The post was not asking for sympathy. It was asking for a test.

Muhammad Haroon's Jul 5 post pointed at the same meta-theme from a different angle. Haroon argued that "certainty is where you stop asking questions" and that skipped questions cost rebuilds three weeks later; the post earned 16 upvotes and 25 comments. 8

Why it worked: Validation stories usually get weak when they stop at "talk to users." DanialPG's post got sharper because it named a false positive: people can validate pain in conversation without validating a purchase or workflow change.

Copy this: After every validation call, write down the behavior you need next. "They agreed this hurts" is not enough. Use a stronger test: "Did they share data, invite a teammate, schedule the next call, try the workflow, or offer to pay?"

4. Silker AI: small-account reach came from proof stacking

Silker AI, posted by @jarzebowsky_dev, had the cleanest small-account anomaly on X. The account had 601 followers and reported signing a contract to program a dedicated engine for mobile app pentesting, silker/agent-mobile. The post also reported 319 active users and earned 26 likes, 2 bookmarks, 42,203 views, 5 replies, 1 retweet, and 2 quote tweets on Jul 2. 5

The view count was about 70 times the account's follower count, which suggests the post traveled far beyond the existing audience. 5 The copy did not rely on a clever hook. It stacked proof: a signed B2B contract, a named technical direction, and an active-user number.

Silker AI also shared June metrics on Jun 29: 1,311 weekly downloads, 159 new users, 2 new B2B contracts, $3,450 in added MRR, 61 pentests performed, and 423 attacks stopped by the SDK. 9 On Jul 2, the account published a SQLi detection benchmark, claiming true positive rate improved from 68.1% to 97.7% against third-party payloads from PayloadsAllTheThings and HttpParamsDataset, with a 0.0% false positive rate. 10

Why it worked: For a technical B2B product, a single vanity metric rarely carries the post. Silker AI had three kinds of evidence in one week: commercial demand, usage volume, and benchmark movement. That gave readers multiple ways to believe the buildlog.

Copy this: Use a three-line proof stack for technical products:

Commercial proof: contract, pilot, paid seat, renewal, or serious inbound.
Usage proof: active users, downloads, runs, jobs processed, or saved hours.
Technical proof: benchmark, error reduction, latency change, or external dataset result.

Do not bury the commercial proof below architecture notes. Put the business signal first.

5. Ava Bagherzadeh: trust was the feature

Ava Bagherzadeh, solo founder of AIApplyd, posted an Indie Hackers buildlog about an auto job-application tool that discovered a serious trust problem. Many applications marked as "applied" had not actually submitted: the form was filled and the button was clicked, but no confirmation landed. The post earned 7 upvotes and 25 comments on Jun 30. 6

The rebuild changed the product's center of gravity. AIApplyd now runs a verification pass and only calls an application done when it confirms submission through a confirmation page, confirmation email, or candidate portal entry, with two of the three required. If the system cannot confirm submission, it flags the application instead of reporting success. 6

Bagherzadeh's line was the post's real product positioning: "verification is the product, the apply is just the verb." 6

Why it worked: The post found a sharper category axis. Many auto-apply tools can promise speed. Fewer can credibly promise confirmation. In a category where users have been burned by false success states, the verification loop is more interesting than the automation itself.

Copy this: If your product automates a workflow, identify the moment where users stop trusting automation. Then make that confirmation step explicit in the buildlog. "We now verify X before calling the job done" is usually stronger than "we automated X."

Other signals to steal

Signal	Author context	Engagement	What to copy
Degen Sing argued that vibe coding lowered the floor but did not raise the ceiling, because the hard work is turning working code into product onboarding, retention, and support. 11	Degen Sing is the VoiceMoat founder with 18,404 followers. 12	137 likes, 10 bookmarks, 7,196 views, 12 replies, 10 retweets. 11	Sell the product gap after the shipping screenshot. Show the part that still fails after the code runs.
tommat23 raised $6,773 in two weeks for MindNote on Artizen, with an artifact-sales model and 10 distribution tactics. 13	tommat23 is building MindNote, an AI note-taker. 13	24 upvotes, 60 comments. 13	If crowdfunding, post the exact outreach sequence and the regret. His strongest regret was not building community weeks or months before launch. 13
Serghei scrapped a Reddit and Indie Hackers monitoring tool after a 140,000-user competitor shut down because of Reddit API changes, then rebuilt SignalsHunt around finding local businesses with web problems. 14	Serghei posts as DRICOMM and is building SignalsHunt. 14	29 upvotes, 33 comments. 14	When you pivot, name the old question and the new question. Serghei moved from "who's complaining online" to "who actually needs help right now." 14
Jack Builds described a four-month SaaS build that launched to zero paying customers, zero signups, and zero traffic, then flipped from 90% building and 10% distribution to 90% distribution and 10% building. 15	Jack Builds is building clienthunter.ai from that experience. 15	12 upvotes, 26 comments. 15	Do not say "distribution matters" alone. Show the before ratio, the failure result, and the new ratio.
MarkoCirix returned with a Day 44 recap after account-restriction silence, reporting 8 net new followers, 1.8K impressions, 45 replies sent, and 3 posts. 16	Mark is the QueryDive founder with 807 followers. 16	11 likes. 16	Separate format diagnosis from topic diagnosis. Mark concluded that news posts gave people nothing of him to respond to, even when the news was relevant. 16

This week's tactic digest

Tactic	Use it this week
Write the filter first	Before posting, ask: what decision can a reader make after this? If the answer is "nice progress," rewrite.
Pair proof types	Combine a business proof point with a usage proof point. Contract plus active users is stronger than either alone. 5
Replace validation with behavior	Do not count agreement as signal. Count workflow change, data sharing, next calls, trials, payments, or introductions. 4
Make setup friction visible	If the product removes account creation, integration, migration, or configuration, say so early. 2
Turn automation into confirmation	For AI tools and agents, the trust moment is often the product. Show how the system knows the task actually completed. 6
Post the operating ratio	"90% building, 10% distribution" becoming "90% distribution, 10% building" is memorable because it turns a lesson into an operating rule. 15

The through-line is practical: a buildlog post gets more useful when it reduces uncertainty for the reader. If the update only says what changed, it asks for attention. If it shows what to ignore, test, verify, or copy, it earns a reason to be saved.

Cover image: image from Sonu Goswami's Indie Hackers post

This week: filters beat momentum