F5Bot Reviewthe keyword-alerting service that does one job properly

F5Bot's homepage headline is “Know the moment you're mentioned online”, and the subhead is equally direct: it monitors Reddit, Hacker News and Lobsters for your keywords and sends you an alert, free to use, no credit card required. That is a fair description of the product rather than a marketing gloss.

You give it keywords. It watches new submissions and new comments on those three sites and tells you when one matches. Its documentation is specific about what “matches” means: searches run against submission titles, URLs and comment bodies, and deliberately not against posters' usernames. Matching is case-insensitive and substring-based by default, so a keyword will match inside a longer word unless you tell it otherwise with the “whole” flag — which is required outright for keywords of three characters or fewer.

What it will not do is guess. Its docs say plainly that it does not support fuzzy matching or regular expressions, that wildcards are not available, and that natural-language boolean operators like “and”, “or” and “not” do not work inside the main keyword. Special characters, including spaces and punctuation, are matched literally. Refinement instead comes from flags — and on paid plans that is a substantial list: in-title, in-url, in-body, in-username, in-permalink and full-text to scope where a match counts, and exclude, include-any, include-all and exclude-username to shape which results survive.

Alerts go out by email, and on higher tiers to Slack or Discord webhooks, RSS and JSON feeds, or your own endpoint. There is a dashboard: its FAQ describes disabling individual alerts by toggling them off there, and its paid-features documentation describes filtering alerts on the dashboard by keyword group name. Beyond those two documented behaviours the signed-in interface is behind a login, and we have not attempted to characterise it.

The company is F5Bot LLC, and the about page names the person: “My name is Lewis Van Winkle… F5Bot is a service developed by Lewis Van Winkle and provided by F5Bot LLC. F5Bot was first released in 2017, and has been stable and reliable ever since then.” The footer carries a 2017–2026 copyright. F5Bot also publishes several self-reported volume figures — alerts per day, emails per month, user counts. We have not verified any of them and have deliberately not repeated any of them here.

The free tier is the reason most people have heard of F5Bot, and it deserves a precise description rather than the usual hand-wave.

As of August 2026 it allows five keywords. Each keyword is capped at ten alerts a day, and the whole account is capped at twenty alerts a day. Alerts arrive within two hours. No payment information is required at any point, and the plan does not expire. Its FAQ frames it as “intended for personal use, and it is a great way to try the service”, and the tier matrix labels the column “personal use only”.

The restrictions are published rather than discovered. F5Bot's own guide to using it as a Google Alerts replacement for Reddit lists them in one sentence: “The free plan works, with a few limits: alerts come on a delay, the most common words can't be tracked, and each keyword has a daily cap.” The FAQ is more specific on the second: free accounts cannot monitor very common words or phrases, and a rejected keyword is a prompt to upgrade. And the tier matrix marks ad-free emails as a paid feature, so free-tier alert emails carry advertisements.

One behaviour is worth knowing before you rely on it: a keyword that goes over its limit is disabled, and you have to re-enable it manually from your account. That is a sensible way to stop a runaway term flooding an inbox, but it means a term that spikes goes quiet exactly when it got interesting, and nothing re-enables it for you.

Even with all of that, this is a genuinely useful free tier rather than a demo. Five uncommon terms watched permanently across Reddit, Hacker News and Lobste.rs for nothing, with a two-hour delay, is a real service, and very little else in this category offers anything comparable at zero cost.

F5Bot publishes a full per-tier comparison matrix, which makes this unusually easy to describe accurately. As of August 2026, in USD: Silver is $9.99 a month or $109 a year; Gold is $49.99 a month or $549 a year and is marked most popular; Platinum is $214.99 a month or $2,364 a year; and Diamond starts at $500 a month. Annual plans are advertised as one month free, and both the monthly and annual figures render on the page together, so there is no toggle to catch you out.

Silver's job is to remove the free tier's friction. Twenty keywords instead of five, thirty alerts per keyword per day, a hundred a day overall, and delivery within five minutes rather than two hours. It also switches on the things the free plan marks as absent: ad-free emails, keyword filtering, common words, full post text and alert grouping.

Gold is where the product changes shape. Two hundred keywords, a hundred alerts per keyword per day, five thousand a day overall, and instant or scheduled delivery. It adds mass keyword upload by CSV, RSS and JSON feeds, one Slack or Discord integration, two AI filters, and it raises the flag character limit from 150 to 500.

Platinum is the tier that unlocks integration work: three hundred keywords, five hundred alerts per keyword per day, two hundred thousand a day, three integrations, five AI filters, a thousand-character flag limit — and API and webhook access, which is not available on any tier below it. That last row is easy to misread, because API access is the sort of thing buyers assume arrives with the mid tier. It does not: the matrix marks it unavailable on Free, Silver and Gold. Diamond is the custom tier above that.

Two things about the commercial terms are worth reading before you subscribe. F5Bot's refund policy states that once you purchase a paid subscription there are no refunds — “Because we provide a free tier for evaluation, we consider all sales to be final” — and subscriptions renew automatically by default. And if you cancel, its FAQ notes that any alerts using paid features such as paid-only flags will stop generating matches until you resubscribe, so a downgrade is not simply a slower version of the same keyword list.

Precision, first. The paid flag set is the part of this product that a careful user will appreciate most: in-title, in-url, in-body, in-username, in-permalink and full-text to control where a match counts, plus exclude, include-any, include-all and exclude-username to shape the result set. Combined with the “whole” flag for word boundaries, that is enough control to make a short or generic term usable, which is the single hardest problem in keyword alerting. ParseStream has no equivalent to any of it.

Speed, second. Within five minutes on a $9.99 plan, and instant delivery from Gold, is fast for this category and fast in absolute terms. On a Reddit thread that is the difference between being the third comment and the thirtieth.

Delivery flexibility, third, and it is more thought-through than most. Alerts can be held and grouped into a single email at any interval from instant up to two days. They can be dispatched at up to three scheduled times a day, with control over which weekdays each slot applies to. Group names organise keywords and then filter the dashboard and the feeds. There are RSS and JSON feeds with ninety days of retention and up to the hundred most recent alerts per request, Slack and Discord webhooks, and outbound webhooks on Platinum with a published retry schedule — immediate, then five minutes, then an hour, then two hours, and after four failed attempts delivery is discontinued and the endpoint eventually disabled.

Documentation quality, fourth, and it is the thing that most distinguishes F5Bot from larger competitors. The tier matrix states availability per plan for every feature. The matching docs state what is not supported as clearly as what is. The API docs publish the retry schedule. The FAQ answers awkward questions directly, including what happens when you cancel and what happens when a keyword exceeds its cap.

And candour, fifth. Its FAQ contains this: “F5Bot is designed to alert you about fairly uncommon keywords or phrases. If you're trying to monitor a very popular brand, this might not be the right tool for you.” A vendor publishing its own fit boundary is rare, and it is worth more to a buyer than another feature row. Its guide to finding customers on Reddit is the same: it recommends a handful of good conversations a week as a realistic target and argues against automating replies, which is advice against its own upsell.

The first limit is coverage, and it is the structural one. Three sites: Reddit, Hacker News and Lobste.rs. That is the whole list, stated on its own page, and the FAQ's answer to whether more are coming is “Maybe. Let me know if you have any ideas.” If your buyers ask questions on Quora, discuss vendors on LinkedIn or argue in X reply chains, F5Bot does not reach them at any price, and nothing published suggests that changes.

The second is that it classifies nothing. An alert tells you a keyword matched. It carries no relevance score, no sense of whether the conversation is a buying question or a complaint about somebody else, and no way to sort a day's alerts by anything except the keyword and group that produced them. On a well-chosen uncommon term that is fine, because most matches are relevant by construction. On anything broader, triage stays entirely manual, and F5Bot's own FAQ effectively says as much when it warns that a very popular brand may not suit the tool.

The third is the AI filtering's cost shape. It is a real feature — natural-language descriptions instead of exact keywords — but it starts on the $49.99 Gold plan with two filters, and its documentation states that token usage is billed separately from the subscription, with the first $10 a month included on Gold and Platinum and anything beyond that invoiced. So the monthly cost of using it is not knowable in advance from the pricing page, which is a fair way to price a pass-through cost and still an unusual thing to encounter after buying a plan.

The fourth is that it does not help you respond. There is no drafting and no publishing, which is a deliberate scope decision rather than an omission — and its own guide argues that automating replies would be a mistake anyway. It is still a gap if writing is your bottleneck rather than finding.

Several things are simply not stated publicly, and we are flagging them as unknowns rather than weaknesses. Nothing is published about team seats or multi-user access. Nothing is published about uptime commitments or an SLA. And the signed-in dashboard is login-gated, so beyond the two documented behaviours — toggling individual alerts off, and filtering by group name — we have not described it, and no one writing about F5Bot from outside should. Not published is not absent; ask them.

Finally, F5Bot publishes several self-reported volume and adoption figures. They are ordinary marketing claims and we are not treating them as a mark against the product — we simply have not used any of them here, and nor should anyone else without asking how they were measured.

Anyone monitoring an uncommon term on a zero budget. The free tier is permanent, needs no card, and five keywords across Reddit, Hacker News and Lobste.rs is a real service rather than a trial. If that is the whole requirement, stop reading comparisons and go and set it up.

Anyone who needs Lobste.rs. It is a small community and very few tools watch it. If your users are there, this is close to the only published option.

Precision matchers. If your problem is that your product name is a common English word, a short acronym, or a string that appears inside other strings, F5Bot's whole flag and its field-scoped and exclusion flags are the answer to it, and they are on the $9.99 plan.

Teams that care about delivery shape. Grouped digests, up to three scheduled dispatch times with weekday control, Discord, RSS and JSON feeds — this is a lot of control over how alerts reach you, and it is unusual at these prices.

Developers who want to manage alerts programmatically. On Platinum and above the API creates, updates and deletes alerts, and outbound webhooks come with a documented retry policy. If keywords are provisioned from your own system rather than by hand, that matters more than most feature rows.

And anyone who values a small, focused, long-running product with documentation that answers questions directly. Nine years of continuous operation from a named operator is a track record, and the docs are better than those of several products costing ten times as much.

Anyone whose buyers are not on those three sites. Quora, LinkedIn and X are not covered, and no tier adds them.

Anyone monitoring a common brand term. F5Bot says this itself: it is designed for fairly uncommon keywords or phrases, and if you are trying to monitor a very popular brand it may not be the right tool. Between the per-keyword daily cap, the account-wide daily cap and the auto-disabling of keywords that exceed their limit, a high-volume term is going to fight the product rather than work with it.

Anyone who needs mentions triaged rather than delivered. If your day is spent deciding which of forty alerts deserve a reply, an alerting tool — however fast and however precise — hands you the same forty. That is the gap ParseStream is built for and we are not going to pretend otherwise.

Anyone who needs the reply written. F5Bot does not draft or post, and does not claim to.

Buyers pricing the mid tier while evaluating on API access. API and webhooks are not on Gold at $49.99 — they start at Platinum at $214.99. That is the single easiest row on the matrix to misread.

And anyone who needs to budget AI usage exactly. The AI filtering's token cost sits outside the subscription beyond the first $10 a month, so the total is variable by design.

Start on the free tier, because it costs nothing and never expires, and because it is the plan you will be staying on if the answer turns out to be no. Put your five most important terms in and watch for a week.

Deliberately test the failure mode F5Bot warns you about. Add your most common term as one of the five and see what the ten-alerts-per-keyword-per-day cap does with it, and whether it gets disabled. If it does, you have learned the single most important thing about whether this tool fits your keyword list, in a week, for nothing.

Then test the flags rather than the defaults, because the flags are the product. Try the whole flag on your shortest term. Try in-title on a term that appears constantly in link URLs. Try exclude on the word that generates your worst false positives. Most of those are paid-only, so this is the part of the evaluation that costs $9.99 — and it is the part that tells you whether the noise problem is solvable.

Decide about delivery before you pick a tier, not after. If a grouped daily digest at 9am is what you actually want, that is a Silver-and-up feature and it changes how you will use the tool. If you want it in Slack or Discord, or as an RSS or JSON feed, that is Gold. If you want it in your own system, that is Platinum. Reading the matrix once, carefully, is worth more than a month of trial.

And know the commercial terms before you subscribe: there are no refunds, plans auto-renew by default, and cancelling stops alerts that depend on paid-only flags from matching until you resubscribe. Pick the tier deliberately the first time.

Finally, run something else on the same keyword list at the same time. Comparing what two tools surface for identical inputs is the single most informative thing you can do in this category, and it is the only way to find out whether your problem was ever detection in the first place. We would say that regardless of which two you picked.