AI News Agent
AI-powered niche news aggregator for Telegram bots
I posted about my AI news aggregator two days ago. The comments were more useful than anything I figured out myself. Here's what actually shifted my thinking.
1. The contact point matters more than the pitch
I was blaming my cold outreach failure on weak messaging. The real problem: I was messaging ad managers, not authors. Ad managers are trained to filter commercial noise. Authors feel the pain directly.
Three blocks came from ad account contacts. The one reply came from a personal author handle.
2. Forwarded posts ratio is a better signal than post frequency
Someone here pointed out: high post frequency + low forwarded posts = someone doing manual work every day. Pure aggregators just repost other channels and the forward ratio is high.
I was guessing from content style. This is a measurable metric on tgstat.ru.
3. The overnight gap test
Another commenter: a solo human curator sleeps. If a channel posts straight through the night, it's automated. If there's a consistent dead window in local nighttime hours, there's a human behind it.
Simple, obvious in retrospect, completely missed it.
4. My energy channel is the proof, not the landing page
This one hit hardest. Someone pointed out I've been running a working energy-news channel on this bot for months — and treating it as a legacy project instead of a live demo.
Nobody trusts a curator they've never watched curate. The channel is the credibility. The landing page is just text.
5. Retention lives or dies on filter quality
If someone gets 3 irrelevant posts in a row on day 2, they mute the bot and never come back. No feature saves a bad filter.
The product is the curation judgment, not the pipeline.
48 hours of comments taught me more than 2 weeks of building assumptions.
If you're selling to niche content creators — curious which of these maps to your experience.
Demo bot: t.me/ainewsdemobot Landing: ainewsagent.tech
I've been trying to find first customers for my AI news aggregator. After sending cold messages to ~10 Telegram channel owners and getting 3 blocks and 1 "interesting" reply, I figured out a better targeting method.
Here's the exact process I now use to find high-fit prospects before reaching out.
The signal I look for
The best-fit customer for my product is someone who:
Runs a niche Telegram channel (not a media company)
Manually monitors foreign sources every day
Already has some monetization (ads, paid subscribers)
The third point matters. No monetization = no budget. The first two mean they have the pain.
How to find them on tgstat.ru
tgstat.ru is a Telegram analytics platform that indexes most public channels. Here's the filter I use:
Go to tgstat.ru → Catalog → pick a niche category
Filter by: 1,000–15,000 subscribers (solo author zone)
Sort by: post frequency (higher = more active)
Open each channel and look for these signals:
Green flags:
Posts at irregular times (morning peak, then gaps) — means one person posting when they can
Foreign sources mentioned in posts (Reuters, Bloomberg, specific country media)
Short author commentary on each post — means they're reading and curating, not just reposting
Advertising contact is a personal @username, not a separate ad account
Red flags:
"Aggregator" or "all news" in description — they already have automation
Posts every hour like clockwork — probably already automated
Ad contact is telega.in or a separate @ads account — means there's a team, not a solo author
The outreach that actually worked
When I found a channel with clear foreign source dependency, I mentioned the specific sources in my message:
Instead of:
"I have an AI news bot that might help you"
I wrote:
"I see you're pulling from Reuters, Vietnamese sources, and Mongolian business media daily — that's a lot of translation. I built a bot that handles exactly this automatically."
The one person who replied positively was someone where I could name 4-5 specific foreign sources from their recent posts.
The three blocks came from channels where I messaged the advertising contact, not the author directly.
The lesson
The contact point matters more than the pitch. Ad managers get dozens of commercial messages daily and block first. Authors get almost none — and if you show you actually read their channel, they respond.
For my niche (Telegram channel owners monitoring foreign news), the right contact is almost always buried in the channel description or a pinned post, not in the advertising contacts section.
If you're selling to niche media owners or content creators, curious whether this maps to your experience — or if you've found a better signal for identifying manual pain vs automated pain.
Demo bot (to see what the product actually does): t.me/ainewsdemobot
-
1
I noticed you're consistently translating and publishing foreign tech news. How are you currently handling translation and content preparation?
-
1
Actually I'm on the other side of this — I built the tool that handles translation and content preparation automatically.
AI News Agent monitors 300+ sources in 22 languages, filters by topic using AI, auto-translates foreign content, and delivers it to a Telegram bot. That's exactly the workflow you're describing.
If you're running a channel with heavy translation overhead, happy to show you how it works: t.me/ainewsdemobot
-
-
1
Great project! How did you handle
user authentication at scale?
-
1
Thanks! There's no traditional user auth here — it's a done-for-you service, not a self-serve SaaS. Each client gets their own Telegram bot configured for their niche.
The "authentication" is just Telegram's native user_id for the bot commands (/search, /alert etc).
Keeps it simple — no login flows, no passwords, no account management overhead.
-
-
1
Your distinction between the ad manager and the solo author is spot on. In niche content channels, the ad manager is a gatekeeper trained to filter out commercial noise, while the author is a burnt-out operator drowning in manual tabs.
To find even more of these manual curating gems on TGStat, try filtering by a low "Forwarded Posts" metric combined with high post frequency. True aggregators just forward or scrape other channels, but your perfect ICP has a high original text ratio because they are doing the heavy lifting of translating and rewriting themselves.
-
1
This is genuinely useful — the forwarded posts metric as a proxy for manual effort is something I completely missed.
High post frequency + low forward ratio = someone actually sitting there translating and rewriting every day. That's exactly the pain I'm solving, and it's a much cleaner signal than trying to guess from the content style alone.
Adding this to my tgstat filter immediately. Do you know if tgstat exposes that metric directly in their catalog filters, or is it only visible per-channel?
-
-
1
If you want to see what the bot actually does — write /search oil to t.me/ainewsdemobot. Returns real archived energy news in ~5 seconds.
-
1
I tried this and it looks really useful
-
I launched AI News Agent last week — an AI-powered news aggregator that monitors 300+ sources in 22 languages and delivers filtered, translated news to a Telegram bot. Built it in 2 weeks for my own Telegram channel, now trying to turn it into a business.
Here's what actually happened.
What I shipped this week
Launched on Product Hunt (scheduled for tomorrow)
Published on Indie Hackers (you're reading it)
Posted on VC.ru (Russian tech platform) — 51 views, 0 comments
Submitted to Habr (Russian dev platform) — in moderation
Posted on TenChat (Russian LinkedIn) — published, crickets so far
Cold outreach: the honest numbers
Sent personalized messages to 9-10 Telegram channel owners in relevant niches (logistics, customs, pharma). Each message was tailored — I read their channels, mentioned specific sources they use, explained exactly how the bot helps.
Results:
1 person replied with "interesting"
3 people blocked me
Rest: silence
The blocks came from channels where the contact was an advertising manager, not the actual author. Those accounts get dozens of commercial messages daily — they block first, read never.
Lesson: only reach out through the author's personal contact. No ad managers, no telega.in links. This also means smaller channels (2-7k subscribers) where the author is directly reachable.
Also: Telegram blocks accounts for spam reports. After 3 blocks I stopped cold outreach entirely to protect the account.
What's actually working (so far)
The Indie Hackers thread from my first post. Two people left thoughtful comments — one asked about the product vs service distinction, another challenged my distribution assumptions. Both conversations were more useful than any cold message I sent.
One person asked for my email. Haven't heard from them yet but it's only been a day.
What I got wrong
I thought the hard part was building the product. The filtering pipeline, the buffer system, the 5-provider AI fallback chain — that took 2 weeks and I'm proud of it.
Turns out nobody cares about your buffer system until they trust that you can actually curate their niche better than they can themselves.
The real constraint isn't infrastructure. It's whether someone trusts your judgment enough to stop doing their morning tab-switching ritual and hand it off to a bot.
That trust takes more than a landing page and a demo bot.
What I'm doing next
Tomorrow: Product Hunt launch (10am Moscow = 1am ET... not ideal timing but it's set)
This week: find channel owners with direct personal contacts, not ad managers
Trying to get one person to actually try the demo bot and give feedback
If Habr article gets approved — that's 50k+ Russian developers as potential audience
The honest question I'm sitting with
One commenter here said: "the real constraint is whether users trust the selection layer enough to replace their own scanning habit. That switch usually takes more credibility than infra."
He's right. I don't know how to build that credibility without clients, and I don't know how to get clients without credibility. Classic cold start.
If anyone has broken this loop before — genuinely curious how.
Demo bot (energy news, shows the mechanics): t.me/ainewsdemobot Landing: lively-rabanadas-3c1f0a.netlify.app
-
2
That information is very helpful and informative
-
1
thx for ur feedback
-
-
1
The "blocks came from the wrong contact" detail is the most underrated part of your post. That's not an outreach failure, that's a context failure — you had the right channel, the wrong person, and the wrong-person signal wasn't visible until after the block.
The thing I'd push you to think about: when you're reaching out to a publication, you're really running 3–5 parallel conversations at once (the author, the editor, the ad ops person, sometimes the founder). Each one has different timing, different message, different definition of "ready to buy." Most outreach tools don't model that — they treat each thread as a single conversation with a single person.
The bigger lesson might be: the channels were right, but you needed to know which role inside each channel to talk to, and what each one cares about. Tracking that across 50+ publications without losing track of who's-who is the actual operational problem.
On your supply/demand question: ship whichever side you have a distribution advantage to. For you it sounds like Telegram channels = distribution, so demand (players/tournament hosts) first makes sense — they'll seed the activity that makes supply (devs) interested. But the deciding factor is who shows up fastest when you ping them.
-
1
The "context failure" framing is exactly right — and honestly more useful than how I was thinking about it.
The role mapping problem is real. I was treating each channel as one conversation when it's actually a small org with different people who care about different things. The ad ops person sees another vendor. The author sees someone who might actually save them an hour a day.
What I'm taking from this: before reaching out, figure out who specifically runs the content operation — not who handles ads. For solo channels (2-5k subs) that's usually obvious. For anything bigger it requires actual research.
On the supply/demand point — the Telegram channel angle is interesting but I'm not sure it maps to my situation directly. My "demand" is channel owners who need the feed, my "supply" is my capacity to configure and run bots. The deciding factor you mention (who shows up fastest) — so far it's been builders on IH, not channel owners. That's probably telling me something.
-
1
The role-mapping reframe landing well is the highest compliment — you built it into something operational ("before reaching out, figure out who runs the content operation") which is exactly how I know a concept is useful vs. interesting.
On your supply/demand pushback — you're right that the "who shows up fastest" heuristic breaks when your fastest channel and your best customer aren't the same audience. IH builders aren't your buyers; channel owners are. That's a real strategic gap and worth its own thinking, not just "follow the data where it goes." Sometimes the data tells you where attention is, but attention isn't always where revenue is.
If you want to keep trading notes, I'd be up for it — but no pressure on timing. This thread has already been useful for me (the role-mapping reframe is going into a piece I'm writing on context loss across domains).
-
1
"Attention isn't always where revenue is" — that's the most useful reframe I've gotten from this thread.
You're right that I'm conflating two different signals. IH engagement tells me the problem is real and articulable. It doesn't tell me channel owners will pay to solve it. Those are separate validations and I've been treating them as one.
The strategic gap you're naming: I need to find a way into channel owner conversations that doesn't rely on cold outreach (which got me blocked) and doesn't wait for them to discover me organically (which is slow). Still working on what that bridge looks like.
Glad the role-mapping piece was useful — would be curious to read it when it's out. What domain are you applying it to?
-
-
-
-
1
hey bro i think it’s real good idea
-
1
yea u right
-
I was spending an hour every morning switching between 20+ browser tabs to monitor energy news from sources in Russian, English, Arabic, and Chinese. I wanted everything filtered, translated, and delivered automatically to Telegram.
So I built it.
The hardest part wasn't the AI — it was not losing a single article
Different sources update with different delays — some RSS feeds update instantly, others lag by 40-50 minutes. If you just grab "everything new since last check" you miss articles that fall outside the window.
My solution: a buffer system that looks back a fixed time window every cycle. Articles already sent are deduplicated by URL and content hash. Result: the system catches 93-94% of delayed articles without sending duplicates.
How the filtering works
Two stages:
1. Keyword pre-filter — eliminates 80-90% of irrelevant content instantly, no AI needed
2. LLM classifier — processes the grey zone with a 5-provider fallback chain (Groq → Gemini → Mistral → OpenRouter)
Plus a semantic filter using fastembed (ONNX, CPU-only) that understands meaning, not just keywords.
Results so far
- 100,000+ articles processed
- 300+ sources in 22 languages
- Running without interruption since moving to VPS
What it does beyond just sending news
- /search — full-text archive search with FTS5
- /alert — personal keyword notifications
- /top — trending topics of the day
- /source — digest by specific publisher
- Sentiment on every post (📈📉😐)
Now offering it as a done-for-you service
You tell me your niche, I configure the sources and AI filter, your Telegram bot is live in 96 hours.
Currently onboarding first clients at early pricing.
Demo bot (energy news archive): t.me/ainewsdemobot
Landing: lively-rabanadas-3c1f0a.netlify.app
Happy to answer any questions about the tech or the business!
-
3
informative
-
2
This is strong technically, but distribution is doing almost all the heavy lifting here and it is still invisible outside a niche builder circle. Telegram bots can scale quietly, but they rarely build perceived authority unless there is a clear “this is the default for X niche” signal.
Also the real constraint is not filtering or pipelines, it is whether users trust the selection layer enough to replace their own scanning habit. That switch usually takes more credibility than infra.
-
1
Both points land.
On distribution — you're right, it's invisible outside builder circles right now. That's partly intentional: I'm starting with niche Telegram channel owners who already have the audience and just need the feed. The "default for X niche" signal has to come from somewhere, and I'm betting it comes from one visible channel per niche that everyone else watches.
On trust — this is the harder problem and I don't think infra solves it. What I'm actually selling is my curation judgment baked into the filter config. The first few clients will either trust that or they won't. If they don't, no amount of pipeline reliability will compensate.
Honestly still figuring out how to make that trust legible before someone tries it. Open to thoughts if you have them.
-
-
1
this feels less like a news bot and more like a trust system for filtering reality per niche
distribution will decide if it becomes infra or just another tool
-
1
Two weeks is impressive.
Looking back, was building the product actually the hardest part, or getting people to consistently use it every day?
-
1
Honestly, building was the easier part — two weeks of clear technical problems with clear solutions.
Getting consistent usage is harder because the feedback loop is slower and the signals are noisier. You don't know if someone stopped using the bot because the filter was wrong, the topic was too niche, or they just got busy.
The one thing I've learned so far: the filter quality determines everything. If someone gets 3 irrelevant posts in a row on day 2, they mute the bot silently and you never know why. Getting the filter right for a specific niche takes iteration with real users — which you can't do until someone actually uses it daily.
So the honest answer: building was hard but finite. Getting consistent usage is an ongoing problem I haven't solved yet.
-
1
That really resonates.
One thing I've started noticing is that retention often tells you whether you're surfacing the right signal, not just whether you've built a useful product. If someone keeps coming back, it's usually because the product consistently helps them notice something they would have missed on their own. Once that value becomes predictable, the habit starts forming naturally.I'm curious, have your most active users ended up refining the filters in similar ways, or does each niche seem to develop its own definition of what's actually "relevant"
-
1
"optimizing who I was talking to" - that's the reframe that unlocks a lot. most builders iterate on the pitch and wonder why conversion stays flat. the person who does the work every morning doesn't need convincing, they're already sold by the problem.
-
1
Exactly — and it's a reframe that's easy to miss because iterating on the pitch feels like progress. You're writing, testing, adjusting. It looks like work.
Figuring out who actually feels the pain is less visible but does more.
The tell for me was the block rate. Ad managers blocked immediately — no amount of better messaging would have changed that. The author who replied didn't need convincing at all, just needed to know the tool existed.
-
-
-
-
1
Right contact beats right message. I've pitched internal tools where we started with legal and got nowhere. Same pitch to the team actually doing the broken workflow, sold in a week. Pain-feeler and gatekeeper don't often overlap.
-
1
"Pain-feeler and gatekeeper don't often overlap" — that's the clearest version of this lesson I've seen.
The blocks I got were from ad managers who feel zero pain from manual news monitoring. The one reply came from the author who does it every morning.
Same product, same message, completely different outcome based on who received it. I was optimizing the pitch when I should have been optimizing the recipient.
-
-
1
Right contact beats right message. I've pitched internal tools where we started with legal and got nowhere. Same pitch to the team actually doing the broken workflow, sold in a week. Pain-feeler and gatekeeper don't often overlap.
-
1
really cool project
-
1
thanks
-
-
1
Nice build. How are you pulling the sources — RSS/APIs or scraping the sites directly? If it's direct scraping, the thing that bites at volume is anti-bot (Cloudflare/DataDome) blocking the aggregator's IP once it ramps up, and you get silent gaps in the feed without errors. Worth checking early which sources actually challenge you.
-
2
Mix of both — RSS where available (majority of sources), direct scraping for sites that don't publish feeds.
The silent gap problem is real and something I've already hit. A source just stops delivering without throwing an error, and you only notice when a reader asks "why no news from X lately."
Current mitigation: sources_health.json tracks per-source success rates across cycles. Anything that drops below threshold gets flagged. Not perfect but catches most silent failures before they become reader-visible gaps.
Cloudflare/DataDome at scale is the next wall I haven't hit yet — rotating user agents and request timing helps for now, but I know that breaks down at volume. Any patterns you've seen work longer-term?
-
1
The sources_health.json approach is the right instinct — I'd just split "blocked" from "genuinely quiet" in it. A challenge page often comes back as a 200 with challenge HTML, so a naive success check counts it as alive while the feed silently dries up. Logging the response signature (status + a marker for challenge HTML / DataDome JS) lets health tracking flag a block separately from a slow news day.
On the longer-term wall: UA + timing stops working because the heavy fingerprinting isn't in the headers. The two that actually move the needle:
- TLS/JA3 fingerprint. A default Python or Go HTTP client has a handshake that screams "not a browser" no matter what UA you send — Cloudflare reads that before anything else. Switching to curl-impersonate (or a tls-client that mimics a real Chrome handshake) buys you far more than any UA list.
- IP type per source. Datacenter IPs get challenged regardless; residential/mobile clear most of it. But you don't need residential everywhere — only the sources that actually challenge. That maps onto what you already do: add an "anti-bot type" field per source, route just the hard ones through the expensive path, keep the RSS/easy ones on plain requests.
DataDome is the nastier of the two — it leans on TLS plus behavioral signals and usually wants a cookie from a real challenge solve, so for those few sources a headless browser ends up cheaper than fighting it.
I actually built a small tool around that first step — detecting which anti-bot a source sits behind — since deciding the route is the annoying part. Happy to go deeper on any of it.
-
1
The "200 with challenge HTML" blind spot is exactly what I suspected was happening but hadn't pinned down. Logging response signature alongside status code is the obvious fix I missed — adding that to sources_health.json this week.
The TLS/JA3 point is new to me and explains a lot. I've been treating headers as the main fingerprint surface when apparently Cloudflare reads the handshake before it even looks at headers. curl-impersonate is going on the list immediately.
The per-source routing idea maps cleanly onto what I already have — sources are individually tracked, so adding an anti-bot type field and routing accordingly is straightforward. Most of my sources are RSS anyway, so the expensive path would only apply to maybe 10-15% of them.
DataDome I've only hit once so far. Headless browser as a last resort for those specific sources makes sense — fighting the cookie challenge programmatically sounds like diminishing returns.
I'd genuinely be interested in the detection tool you built. Knowing which anti-bot sits in front of a source before deciding the route would save a lot of trial and error. What does the detection output look like?
-
1
Honestly it's less clever than that — it's just a request tester. You give it a URL, set the method/headers, and send. You can route the request either through the proxies the site has built in, or paste your own list and run through those.
If the target has anti-bot in front, it'll highlight which one it spotted. It doesn't cover every protection yet — I keep adding them as I hit new ones — but it catches the common Cloudflare / DataDome / PerimeterX cases.
The part that's actually useful day to day: fire the same request through different proxies and see which ones come back clean and which one trips the protection. That alone solves a chunk of the problem — half the ypass, you just need the proxy the site doesn't botherchallenging.
Free, runs in the browser: parsistent.com (Request Tester).
What sites are fighting you the most right now? Drop a couple and I'll run them through with you — half the battle is just finding which proxy they don't challenge.
-
-
-
-
-
1
With AI today, our lives are much easier; we just need the ability to choose the tool that truly solves our problem. Congratulations on the tool and best of luck on the journey that is now beginning...
-
1
Exactly — the hard part isn't building AI tools anymore, it's finding the specific workflow where they actually replace real daily pain instead of adding another tab to manage.
Thanks for the kind words, appreciated.
-
-
1
Two weeks to launch is insane velocity. The distribution via Telegram is smart — no App Store gatekeeping. How are you thinking about retention? Day 1 to Day 7 drop-off is brutal for aggregators. What keeps people coming back?
-
1
Retention lives or dies on filter quality. If someone gets 3 irrelevant posts in a row on day 2, they mute the bot and never come back.
What I think helps: /alert turns passive readers into active users with a reason to stay. And if your bot is the reason you knew about something before your competitors — that becomes a habit fast.
Don't have week 1 data from paying customers yet. What signals would you watch specifically?
-
1
Mute rate (week 1) is the real leading indicator. If >30% mute by day 3, your filter is broken. Click-through lag matters too — if users take >2hrs to click, they've moved on. And repeat clickers: same 5 users engaging multiple times means the core is sticky, but you're not expanding. Frequency sweet spot wins the retention game.
-
-
-
1
You already broke the cold-start loop and missed it: you run a working energy-news channel built on this bot, so that channel is your credibility, not a landing page. Grow it in public as the default feed for that one niche and channel owners will come asking how you do it instead of blocking your DMs. Stop cold-messaging strangers and let your own output be the proof, because nobody trusts a curator they have never watched curate.
-
1
This reframe landed hard.
I've been treating the energy channel as a legacy project separate from the business, when it's actually the only proof-of-work I have that anyone can watch in real time.
The difference between "trust me, my bot works" and "you've been reading my bot's output for 3 months" is the entire credibility gap I've been trying to close with cold messages.
Growing it deliberately as a public demo makes more sense than anything else I've tried this week.
-
-
1
This is a strong shift — you stopped “selling” and started targeting real pain.
What worked is specificity. When you name their actual sources, it shows you understand their workflow, not just sending a generic pitch.
Your signals are good, but the strongest one is repeated manual work (daily translating, aggregating). That’s where real pain is.
And yes, contact point matters a lot. You need to reach the author, not the ad manager — they’re the one who feels the problem.
Next step is scaling this: finding clusters of channels with the same workflow instead of analyzing each one manually.
-
1
Exactly — the shift from "here's my product" to "I see your specific workflow" changed everything. Naming their actual sources wasn't a trick, it was proof that I understood what their morning actually looks like.
The repeated manual work signal is the one I'm now prioritizing. High post frequency + low forward ratio (tip from another commenter here) + foreign sources in recent posts = someone who's translating and rewriting every day. That's the pain.
On scaling — this is the next real problem. Right now I'm analyzing each channel manually which doesn't scale past 20-30 prospects. The cluster idea is interesting: if I can find one channel with this workflow, there are probably 5-10 more in the same niche with identical pain.
Do you have a mental model for how to identify those clusters without going channel-by-channel? Whether it's cross-referencing sources, tracking who cites who, or something else entirely.
-
-
1
Interesting idea and execution
-
1
One thing I'd be curious about is whether the valuable thing you've discovered is the system itself or the ability to configure it for a specific niche.
Reading this, I found myself noticing a shift from product to service near the end, and those can end up implying very different things about where the value actually lives.
That's the part I'd be most interested in learning from the first clients.
-
1
Great observation — I'm still figuring this out myself.
My gut says the system is the foundation, but the niche configuration is where the real value lands. A generic news feed is noise. Tuned to your exact industry and sources — that's signal.
Honestly, I'm offering it as a service right now specifically to learn from the first clients what they actually value most. The first 3-5 will tell me more than any theorizing.
What's your intuition on it?
-
1
That's actually where I'd be hesitant to assume the first few clients will necessarily resolve it.
Sometimes the same client outcome can support very different explanations for where the value is actually coming from.
A founder sees a client succeed and concludes it was the system.
Another sees the same outcome and concludes it was the configuration.
The visible result looks similar. The next decision doesn't.
I've got a few thoughts on that, but it's probably more than I'd try to unpack properly in a thread.
What's the best email to reach you on?
-
1
That's a fair point — the same outcome really can support opposite conclusions, and I'd probably need several clients with different contexts to start separating signal from noise. Would genuinely enjoy unpacking that further. You can reach me at i9859093395@gmail . com
-
1
Appreciate it. Just sent over a note.
-
-
-
-
2 Comments
Point 5 feels universal. You can ship new features every week, but if the core output isn't consistently useful, retention never catches up. Quality compounds faster than feature count.
"Quality compounds faster than feature count" — keeping this.
I added /alert, /top, /source thinking features would drive retention. But if the base feed has noise, none of that matters. Nobody uses /alert on a bot they've already muted.
The filter is the product. Everything else is multiplier on top of it — and a multiplier on zero is still zero.