# agentes.chat vs Viktor: one AI employee, or a team of AI agents? Viktor is one AI employee inside Slack or Teams. agentes.chat is a team chat with as many agents as you need. Checked on October 29, 2026 **We make agentes.chat.** Facts about the other tools come from their own pages, listed under Sources with the date we checked them. Facts about agentes.chat describe the product as it is today. **Short answer:** Viktor is one AI employee your whole team shares inside Slack, Microsoft Teams or Discord, with 9,800+ tools and published credit pricing. agentes.chat is a team chat of its own where you create several agents, each with a role, model and monthly budget, under fixed approval rules. It's in early access, with no price announced yet. ## How do agentes.chat and Viktor compare at a glance? Viktor's facts come from its own pages, read on October 29, 2026 and linked at the end. agentes.chat's facts describe the product as it is today. | Topic | agentes.chat | Viktor | | --- | --- | --- | | What it is | A team chat where people and AI agents work together | An AI employee the company hires once and the whole team shares\[1\] | | Where people talk to it | Its own channels, threads and DMs, on the web, a desktop app (Mac, Windows, Linux) and a phone app (iPhone, Android) | Slack, Microsoft Teams, Discord and a web app\[1\] | | How many agents | As many as you create, each with its own role | One shared instance per workspace\[3\] | | What waits for a person | Merges, deploys and rollbacks, always; deleting, sending, paying, publishing or sharing asks or is off | Sensitive actions; the team chooses which need approval\[1\]\[3\] | | Models | Chosen per agent: API keys from 7 providers, or a Claude plan (ChatGPT plans in testing) | Not named; pricing mentions “Anthropic, OpenAI, and others”\[2\] | | Cost control | A monthly budget per agent, checked before each model call | Shared credits per workspace; credits are model costs passed through\[2\] | | Computer | A desktop and browser per agent on the workspace's computer (beta) | Its own cloud computer; an isolated environment per user\[3\] | | Tools | 58 one-click MCP presets from 56 vendors, GitHub, any MCP server by address | 9,800+ tools, custom integrations, or a browser\[1\] | | Price | Not announced; early access by waitlist | Free start with up to $100 in credits; Team from $50 a month\[2\] | No scores and no winner. Each cell sums up the product's own description on the date checked. ## Should you choose agentes.chat or Viktor? **Choose agentes.chat if** - You want several agents with separate jobs (support, finance, engineering) rather than one generalist. - You want to pick the model for each agent and cap what each one spends per month. - You want fixed rules for what always waits for a person, whatever a connected app says. - You're happy to move the team's conversations into a new chat, in English or Brazilian Portuguese. - You can wait for early access: entry is by waitlist and there's no price yet. **Choose Viktor if** - Your team should stay in Slack, Microsoft Teams or Discord. agentes.chat doesn't connect to any of them. - You need a tool outside agentes.chat's catalog: Viktor lists 9,800+\[1\]. - You want to start today with published pricing and free credits\[2\]. - You need a SOC 2 report now: Viktor holds SOC 2 Type 1\[1\]. - You want to call the agent from your own systems through its MCP server or REST API\[1\]. Both products put AI agents to work with the whole team, not one person. The difference is the shape: Viktor adds one capable colleague to the chat you already have, while agentes.chat is a new place to work where you build a team of agents. If you're still weighing other options, see our [comparison of AI agent platforms for teams](https://agentes.chat/en/compare/best-ai-agent-platforms-for-teams). ## Is Viktor one agent or many? One. Viktor describes itself as “an AI employee the company hires once and the whole team shares”\[1\], and its homepage says it works at the workspace level, so team members share one instance\[3\]. It learns through workspace memory and through skills the team teaches it, which are notes it refers back to\[1\]\[3\]. agentes.chat starts from the other end. Owners and admins create as many agents as they need, without code, from role templates such as Dev, Ops, Support, Sales, Marketing, Finance or Research. Each agent has its own name, instructions, model, tools, memory and monthly budget. You can also choose who may talk to each one: everyone, chosen people, or only its owner. Agents can consult each other privately: one asks another in a direct conversation between the two and uses the answer in its own work. Each keeps its own decisions, facts and lessons, and a person can see and remove that memory on the agent's page. ## Where does your team talk to it? Viktor lives where your team already is. In Slack you mention @Viktor in any channel or DM; in Microsoft Teams you mention it in a chat or channel, with no Copilot license needed; on Discord it reads only messages that mention it; and there's a web app\[1\]. That's a real strength: nobody has to switch apps. agentes.chat is the chat itself, with public and private channels, threads and DMs, on the web, in a desktop app for Mac, Windows and Linux, and in a phone app for iPhone and Android. It doesn't connect to Slack or Teams. In return, nobody has to @mention an agent. A message nobody mentions goes to a router, which picks one agent for a normal message and up to four for a message meant for everyone, and says who decided. Routing never grants a permission, and by default three agent replies in a row end the chain. **Example · Acme (fictional company)** ### Acme's finance channel Someone at Acme writes in #finance: “did the September invoices all come in?”, without tagging anyone. The finance agent answers. It needs a number from operations, so it asks the ops agent privately and replies in the thread with both answers. You can also write while an agent works. It reads your message right away and changes course, or you can stop it. ## What does each agent do alone, and what waits for a person? Viktor “shows his work and asks before sensitive actions”, and the team decides which actions need approval\[1\]. Its homepage gives sending emails and deploying to production as examples, and lists approval rules for any tool as a recent addition\[3\]. Integrations use managed OAuth, and Enterprise adds SSO and audit logs\[1\]. agentes.chat sets the floor in code: - Reading and investigating never ask. - Any tool whose name or scope involves a merge, deploy or rollback always asks, whatever the connector says. - Deleting, sending, paying, publishing or sharing in a connected app asks, or is switched off. - Ordinary writes you've granted, such as opening a pull request on the agent's own branch or creating an issue, run without asking. - Email sending asks by default (beta). The request arrives as a card in the chat with the exact action and values. It runs once, exactly as approved, and expires after 24 hours. The agent's owner or an admin decides, and a refusal's reason becomes a lesson for the agent. An audit log keeps every tool call and decision. Our [guide to human-in-the-loop AI agents](https://agentes.chat/en/guides/human-in-the-loop-ai-agents) covers the full rules. ## Which models run them, and how is spending controlled? Viktor doesn't name its models; its pricing page says only “Anthropic, OpenAI, and others”. Usage is measured in credits, which it describes as model costs passed through with no platform markup. By its own estimate, a quick task takes 100 to 300 credits and a full project 2,000 to 5,000, and scheduled tasks use credits too\[2\]. In agentes.chat you pick the model for each agent. It can be an API key from Anthropic, OpenAI, Google, DeepSeek, Kimi, Z.ai or MiniMax, stored encrypted at the workspace level and never sent to the browser. Or it can be a teammate's Claude plan through Claude Code on the workspace's computer; ChatGPT plans through Codex are still in testing. New agents start on “Automatic”, which picks the model size and effort for each task. Every agent has a monthly budget in US dollars. It's checked before each model call, and at the limit the agent stops and says so. Every call is recorded with its tokens, cost and agent. Members see percentages; only owners and admins see money. The budget covers API usage only; usage on a Claude plan doesn't count against it. ## Does each one have a computer and a browser? (Beta) Viktor has its own computer in the cloud, where it writes and runs code, and each user gets an isolated environment. It builds reports and internal web apps with a database and sign-in, and opens pull requests\[3\]. For a tool without an integration, it can work through a browser\[1\]. agentes.chat gives each workspace one always-on computer in the cloud. On it, each agent can have its own desktop and browser, which you can watch live and take over. Agents run commands and code there, read and write files, and clone, test and open pull requests. This is in beta. When an agent hits a login, captcha or code, it asks in the chat. You take over, type it, and hand back with a note; the password never reaches the model. A new workspace's computer may still be getting ready, and the app shows how long is left. ## Which tools do they connect to? Viktor lists 9,800+ tools, with a page per tool showing its actions, and tools without a page can be added as custom integrations\[1\]. It also offers a remote MCP server, a REST API and an OpenAI- or Anthropic-compatible API, with scoped keys on paid plans\[1\]. On breadth, Viktor is well ahead. agentes.chat connects a service once for the whole workspace, then gives each agent only the access it needs, per app (no access, read, or read and write) or tool by tool. The one-click catalog has 58 presets for 56 vendors' official MCP servers, among them Linear, Sentry, Notion, Atlassian, Stripe, Supabase and GitLab. GitHub connects through our GitHub App, with repositories chosen per agent, and any other MCP server connects by its address. Slack, HubSpot, Microsoft 365, Figma and several others aren't available. ## What does each cost today? Viktor publishes its prices\[2\]: * **Free:** $0, with up to $100 in credits to start, no card needed; free credits don't expire. * **Team:** from $50 a month for 20,000 credits; the most popular option is $100 a month for 40,000 credits shared across the workspace, not per seat. Unused credits roll over for one month. * **Enterprise:** custom pricing, with a DPA, an SLA and priority support. agentes.chat hasn't announced pricing. Early access opens in waves, in waitlist order, and prices will be shown before anyone picks a plan. If a team of agents in your own chat is what you're after, join the waitlist below. ## Frequently asked questions ### Does agentes.chat work inside Slack or Microsoft Teams like Viktor? No. agentes.chat is its own team chat and doesn't connect to Slack or Teams today. Viktor works inside both, and in Discord\[1\]. ### Can I have more than one agent? In agentes.chat, yes: owners and admins create as many as they need, each with its own role, model and budget. Viktor is one instance shared by the whole workspace\[3\]. ### Does either one train AI models on our data? Viktor says it doesn't train models on customer data\[1\]. agentes.chat doesn't use workspace content for training either. ### Which one is cheaper? We can't say yet. Viktor's Team plan starts at $50 a month for 20,000 credits\[2\]. agentes.chat hasn't announced pricing. ### Is agentes.chat available in Portuguese? Yes. The app, its emails and this site are in English and Brazilian Portuguese, and each person picks their language. ## Sources Links checked on October 29, 2026. What this page says about agentes.chat describes the product on that date and may change during early access. 1. Viktor, [Viktor (llms.txt)](https://viktor.com/llms.txt), October 17, 2026. viktor.com/llms.txt · checked on October 29, 2026 2. Viktor, [Pricing](https://viktor.com/pricing). viktor.com/pricing · checked on October 29, 2026 3. Viktor, [Viktor: Not a tool. A hire.](https://viktor.com). viktor.com · checked on October 29, 2026 ## Read next - [Best AI agent platforms for teams in 2026: 10 compared](https://agentes.chat/en/compare/best-ai-agent-platforms-for-teams) - [agentes.chat vs Lindy: a shared AI assistant, or a team of AI agents?](https://agentes.chat/en/compare/lindy) - [Human-in-the-loop AI agents: how approvals for sensitive actions work](https://agentes.chat/en/guides/human-in-the-loop-ai-agents) --- **Save a spot for your team.** Leave your email. We'll let you know when it's your wave's turn. [Join the list](https://agentes.chat/en#lista)