Launched this week
Your AI tools answer questions. Viktor does the work. It lives in Slack, connects to 3,000+ tools across your entire stack, and acts on its own. It watches how your team works, spots problems before anyone notices, and proposes automations built around how your company actually works, before anyone asks. It manages campaigns, builds apps, delivers reports, and writes code. And it runs for weeks without losing context, learning your company deeper every day. Not a chatbot. A coworker.
moltdj
Proactively acting on 3,000+ tools is bold. What guardrails do you have so Viktor doesn't go rogue and start closing Jira tickets on its own?
getviktor.com
@borislovelove So in practice we haven't seen cases where Viktor went rouge like that. But we also have strong guardrails like that by default every single write action (like creating, updating or closing jira tickets) requires you in slack to press a "Approve" button before it will be run. This is also to avoid leaking any info from inside slack to outside apps. You can later change which actions you want Viktor to run automatically, as you build trust with Viktor over time.
Nice one team. How does it deal with content that might be hidden in private pages or channels where there is selective access? ISO 27001 forces a ‘need to know basis’ on us so quite a lot sits outside of public channels.
getviktor.com
@adam_anyszewski Thanks Adam! Viktor only sees the channels you explicitly invite it to - so if something lives in a private channel or DM, Viktor has zero access unless you add it there. You control exactly which channels it joins, so it maps pretty naturally to a need-to-know setup. Same goes for integrations - it only connects to what you grant.
Congrats on the launch! The proactive UTM generation example is wild, that's the kind of 'actually does the work' behaviour that separates real agents from glorified autocomplete. We're building in the AI agent space too (chatbots for agency clients) and seeing similar patterns where the magic is in persistence + context. One question: how do you handle the trust handoff when Viktor wants to execute something sensitive like ad spend changes?
getviktor.com
@cuygun Thanks, and yeah "glorified autocomplete" is exactly the bar we're trying to stay above haha.
Cool that you're building in the agent space too - the persistence + context piece is where it all clicks.
On the trust handoff - Viktor has a tiered approval system. Low risk stuff like pulling reports, monitoring metrics, or drafting content just runs automatically.
But anything with real consequences - pausing an ad campaign, changing spend, executing a database write, sending an email on your behalf - gets flagged as a draft that needs explicit human approval before it executes.
So Viktor will say "hey, this campaign's CPA is 30% above target, I think we should pause it - approve?" and waits for a thumbs up before touching anything. You get the proactive analysis and recommendation for free, but the final call stays with the human.
Over time as trust builds, teams can adjust those thresholds. But the default is always "ask before you spend money." We'd rather Viktor be slightly too cautious on day one than blow someone's ad budget at 3am.
@toni_olendzki, That tiered approach makes a lot of sense, especially defaulting to cautious on day one. We're solving a similar trust problem on the chatbot side, where agency clients need to feel confident the bot won't go off-script with their customers. Appreciate the detailed breakdown.
Congratulations on the launch! How autonomous is it? If I give it my Amazon login and password, will it be able to connect to the site, find a product, and order it under my account?
@mykyta_semenov_ Thanks! Viktor can use a browser, but amazons bot protection is pretty good and thats not the main focus of Viktor. Its an AI coworker that lives in your Slack and connects to your business tools via integrations. You give it a task and it autonomously does the work (reports, emails, data analysis, etc). For sensitive stuff you can set approval levels. Shopping on Amazon would be more of a web agent use case - Viktor is focused on the work stuff :)
FoxyApps
Why do you make it impossible for EU-based companies to upgrade? There is no way to add our company's billing details, and I can't pay using our business card because you won't issue an invoice for our company.
This surprises me the most, because you are EU founders too! You must understand it's a problem.
Implementation is just a "checkbox" you enable in Stripe to automatically collect VAT.
It's a great product...but, common guys. Don't block other EU companies from using your platform.
getviktor.com
@ichangethewayThanks for the pointer! This was indeed an oversight on our side. Fix has been deployed, please try again
FoxyApps
@darthwade ok guys, you won me back! Love you 😘
I love the idea of connecting it to our meta/google ad accounts, but how do we know it's not going to accidentally do a lot of damage?
getviktor.com
@jjc Viktor can't touch your ad accounts without explicit approval. Every action that affects spend shows up as a draft: what it wants to do and why. You approve or reject, and only then does it execute.
In practice: Viktor monitors your campaigns 24/7, catches a CPA spike at 2am, and sends you a message. "This campaign is burning budget, here's what I'd do. Want me to go ahead?" You wake up, see it, and decide.
We've had it catch spend anomalies overnight that would've wasted hundreds of dollars by morning. Your judgment on strategy stays yours. Viktor just makes sure nothing burns while you're asleep.
Launched from a warzone is a hell of a backstory and already builds trust damn. The "runs for weeks without losing context" claim is the part I want to hear more about since that's exactly where every long-running agent stumbles. Congrats on shipping Viktor.
getviktor.com
@shubhampalriwala Thanks! Yeah the warzone launch was... not planned haha.
On the memory side - Viktor uses a skill system where every workflow, preference, and lesson learned gets written to persistent files that survive across sessions. So when it learns that your team prefers reports in a certain format, or that a specific client is sensitive about pricing, it remembers that weeks later without you repeating it.
It's not a giant context window trick. It's actual structured memory that compounds over time. The longer Viktor runs with your team, the better it gets - because it's literally building up knowledge about how you work.
That's the part most agent frameworks get wrong. They treat every session as a fresh start. Viktor treats every interaction as another day on the job.