this+that vs Viktor
An AI coworker you summon in chat, or one that works the email nobody has read yet.
Viktor is an execution engine that lives in the Slack and Teams channels you invite it to. You mention it, or you put it on a schedule, and it writes and runs code in its own cloud, deploys apps, and hands back finished work. this+that starts from the other side: your email and chat channels flow into one place, what arrives there becomes tracked work, and your workflows write what they learn into a shared brain as they run, so the whole team keeps working from the same picture.
What each one accumulates
Viktor launched in February 2026 and has moved faster than anyone in this category, with a $75M Series A led by Accel in May and Slack's own cofounders among its backers. It is an AI coworker inside Slack and Microsoft Teams: mention it in a channel and it executes, running code in an isolated cloud environment, deploying web apps, opening pull requests, auditing ad spend, and returning the result as a file, an app, or a commit. It goes deeper on execution than we do, and the traction is earned.
The clearest place to see the difference is email. Viktor can work a mailbox: point it at Gmail and it will triage, categorize and draft. But it starts when you mention it in a channel or when a schedule fires, so a request sitting in a thread nobody opened stays sitting there. Here the arrival is the trigger. Gmail and Outlook are read continuously, and the work buried in them surfaces whether or not anyone thought to ask. That is the difference between an agent you remember to use and one that has already read your morning.
The second difference is not what either product can do this afternoon. It is what each one accumulates. Viktor accumulates execution ability: more tools, more skills documents, a deeper bench of things it can build for you. this+that accumulates what your company knows. Your email and chat land in one place, and your workflows write what they learn into a knowledge layer we call the Brain as they run, which then grounds whatever the AI does next.
That is a problem big companies solve with headcount. Chiefs of staff, program managers, sales ops, account teams, people whose real job is making sure everyone else is working from the same current picture. A twenty person company has exactly the same problem and nobody to spare for it, so two people answer a customer differently, or someone quotes off numbers that changed last month. That is the job the brain does.
It matters for the AI too, because agent output is bounded by context now rather than by model. Any capable model can write the pitch deck. Whether the first draft is usable depends on whether it knows your pricing, your product, and what this particular customer has already told you. Viktor's context lives in skills documents that somebody writes and maintains, and hand-maintained knowledge decays, which is the reason wikis fail. Ours is written by the work: your workflows put what they learn back into the brain as they run, and the people layer builds itself out of the accounts you already connected, rather than someone remembering to update a page.
So, where each one fits. If your company lives entirely in Slack or Teams, and the job is handing an agent something you can describe in a sentence, Viktor is a strong choice and it will be useful on your first afternoon. If the problem is that the work is buried in an inbox nobody has read, and you want the AI to get better at your business the longer it runs, that is what we built. Our first afternoon is less impressive. Our fourth week is a different product.
Feature by feature
How the two products compare across the things that matter most.
| Topic | Viktor | this+that |
|---|---|---|
| Where work starts | You mention @Viktor in a channel with a request, or you put a job on a schedule. It also proposes automations based on workflows it observes. Either way the trigger is you or a clock. | With the messages that arrive. Every incoming message is read for what needs doing, whether or not anyone asks and whether or not it was expected. You can also ask the assistant directly, or run a workflow on a schedule. |
| Gmail is one of its connectors, and it can triage a mailbox, categorize messages and draft replies when you ask it to or put it on a schedule. What it does not do is sit on the mailbox and start work because something landed. | Gmail and Outlook are first-class inbound channels, read continuously. The message arriving is itself the trigger, with nobody to mention and no schedule to wait for. | |
| Surface | Slack, Microsoft Teams and, since August 2026, Discord, in the channels and DMs it has been invited to, plus its own web app chat. | A rebuilt inbox app where every connected channel is read and replied to in one place, plus a Chrome extension for Gmail. |
| Messaging and customer DMs | Slack, Teams and Discord, which are places colleagues talk rather than places customers do. | Slack, Microsoft Teams, Google Chat, and WhatsApp Business. |
| Task management | Work comes back as artifacts: files, dashboards, deployed apps, pull requests. No shared task manager is described. | DoBox: tasks extracted from messages across every connected channel, with per-person assignment, lists, priorities, due dates, comments, and attachments. Every task links back to the message it came from. |
| Code | Writes and executes code in its own isolated cloud environment, deploys web apps with databases and auth, opens pull requests, triages bugs. | Delegate hands coding work to an agent running locally on your machine, on your own AI subscription, and it comes back as a draft pull request. Mac and Linux today. |
| Knowledge and memory | A knowledge base of documented "skills" it references, plus team context carried across conversations. | The Brain: a writable team knowledge layer with personal and team spaces. Workflows read from it and write back to it as they run, so it grows out of the work instead of being maintained by hand. Keeping the operational layer current with no flow to set up is coming. |
| The people layer | Team context is carried across conversations in the channels it sits in. No contact record is described. | Contacts assembles itself from the accounts you already connected, resolves the same person across email and chat into one record, and attaches your correspondence to each of them. Nobody fills it in, and your edits survive the next sync. This part genuinely maintains itself. |
| Integrations | A large catalog of prebuilt connectors, stated as more than 3,200 tools, plus an OpenAI-compatible public API and its own MCP server, so other tools can call Viktor as well as the other way around. | Direct integrations plus the open MCP standard, so any tool with an MCP server or an API plugs in. The difference is which direction the work flows: integrations here are mostly places work arrives from, rather than tools to point a task at. |
| Approvals and control | Sensitive actions wait for approval; credentials are kept away from the model; no training on your data. | Approval gates on workflow actions, drafts rather than sends by default on anything outbound, and provenance from every task back to its source message. |
| Security and compliance | SOC 2 Type 1 and CASA Tier 3, GDPR aligned. SOC 2 Type II and ISO 27001 stated as in progress. | CASA Tier 2 certified, most recently revalidated in February 2026, which is the assessment Google requires for apps with restricted-scope access to Gmail data. SOC 2 Type I as of July 31, 2026, with Type II underway. Message content, tasks, knowledge pages, and connected-account credentials are envelope-encrypted under dedicated customer-managed keys, one per data class. No this+that employee can decrypt customer content, and your data never trains a model. |
| Pricing | $100 in free credits, then Team from $50/month for 20,000 shared workspace credits, scaling up from there, and Enterprise on custom terms. Credits are pooled per workspace rather than charged per seat, and map to model cost with no platform markup. | Free during beta. |
- It executes rather than drafts. Its own cloud environment means it can run the code, deploy the app, and hand back something finished, which is a genuinely higher bar than most agents clear.
- Slack and Teams native, in the App Directory, sitting where a lot of companies already run their day, with Discord and a web app added in August 2026.
- Workspace credits instead of per-seat pricing, so rolling it out to more people is not a budgeting exercise.
- CASA Tier 3, a tier above ours, with credentials kept away from the model and no training on customer data. SOC 2 Type II and ISO 27001 are stated as in progress. We both hold SOC 2 Type I.
- Sustained context across projects that run for weeks, not just single-turn tasks.
- A very large prebuilt integration catalog, which means less setup before the first useful result.
- A real developer platform as of late July 2026: an OpenAI-compatible public API, an MCP server exposing Viktor to clients like Claude Desktop and Cursor, and published docs. Other systems can drive Viktor programmatically, which few tools in this category offer.
- The fastest traction in this category by a distance: public in February 2026, a $75M Series A by May, thousands of customer organizations.
- Email is where the unread work actually is, and it is an inbound channel here rather than a tool to be pointed at. Gmail and Outlook are read continuously, so a request buried in a thread nobody opened still becomes tracked work.
- The Brain is written by the work rather than by somebody keeping a doc up to date: your workflows put what they learn back into it as they run. A design partner put it best, that a pitch deck built on accurate product, pricing, and customer knowledge comes out close to ready, where the same deck from a generic automation tool needs rewriting.
- The people you deal with assemble themselves. Contacts builds from the accounts you already connected and keeps itself current, with the same person across email and chat resolved into one record. It is the part of the brain that needs no setup at all.
- The work starts itself. Nobody has to notice a message and think to delegate it, which is the step that actually fails on a busy week.
- DoBox turns messages into tracked, assignable work for the whole team, with a link back to the source every time.
- Workflows are written in plain language and triggered by what arrives, with approval gates on the actions that matter.
- It does the job a bigger company would hire for. Keeping everyone working from the same current picture is what chiefs of staff and sales ops teams are for, and a twenty person company has the same problem with nobody to spare for it.
- Deciding what deserves action is the hard part, and it is the part we build for. Two hundred messages arrive, most of them need nothing, and a handful are the ones that matter. Judging an unfiltered stream is a different problem from executing a task somebody already wrote down.
- Free during beta.
What is Viktor?
Viktor is an AI coworker you message in Slack, Microsoft Teams, Discord or its own web app, and it does a job: writing and running code in an isolated cloud environment, deploying web apps, building reports and decks, and automating across a large integration catalog. Jobs can run for days rather than minutes. It is built by Zeta Labs in Warsaw, founded in 2023 by two ex-Meta engineers, and launched publicly in February 2026. The company raised a $75M Series A led by Accel in May 2026, the largest Series A in Poland to date, with angel investors including Slack cofounders Stewart Butterfield and Cal Henderson. Reported traction at ten weeks was more than 2,000 organizations and a $15M ARR run rate.
How much does Viktor cost?
Viktor starts with $100 in free credits, then Team plans from $50 a month for 20,000 shared credits, scaling up from there, with Enterprise on custom terms. Credits are pooled across the workspace rather than charged per seat, and they map to model cost with no platform markup. That makes a straight per-seat comparison misleading in both directions: a small team running heavy jobs can spend more than a per-seat tool would cost, and a large team running light ones can spend considerably less. this+that is free during the open beta.
Who should choose Viktor?
If the work you want done is well specified and someone is there to ask for it, Viktor is the stronger tool. It executes long-running jobs, writes and ships real code, and reaches a very large integration catalog. Since late July 2026 it also has a public OpenAI-compatible API and its own MCP server, so other systems can drive it programmatically. It holds SOC 2 Type 1 and CASA Tier 3 today. We hold SOC 2 Type I as of July 31, 2026, so we match on that and they are a CASA tier above us. Teams that live in Slack or Teams and want a capable executor on demand will get more out of Viktor than out of us.
Looking for a Viktor alternative?
The usual reason to look elsewhere is channel coverage. Viktor works in Slack, Microsoft Teams and Discord, and while it will send email on your behalf, it does not read an inbox as a channel. If the requests that matter arrive by email from customers, suppliers, or candidates, Viktor cannot see them. this+that reads Gmail, Outlook, Slack, Microsoft Teams, Google Chat, and WhatsApp Business, and starts work when a message arrives rather than when somebody remembers to ask.
See what's waiting in your inbox
Connect your inbox and this+that starts reading it. The tasks, follow-ups, and open commitments buried in your messages become tracked work, and what your team knows starts collecting in one place.
Free during beta · No credit card required