Forgeron3 or Viktor: which one for an SME?

Viktor, published by the US company Zeta AI, presents itself as an AI employee that works in Slack, Microsoft Teams and email, as well as on the web and in Discord. It learns the team’s ways of working as skills, connects to many tools and bills usage in credits shared across the whole workspace.

Forgeron3 and Viktor meet on messaging and calendar, scheduled routines, MCP connection and publishing content by link. They differ on where data is hosted, the free plan, how usage is counted and the role given to video meetings.

Viktor lives where the team already talks: Slack, Microsoft Teams and email, with no new tool to open. It keeps the team’s methods as skills that can be created, edited or imported, and claims more than 9,800 integrations. It also builds and hosts small web apps, including dashboards that can be published by link. It asks for human approval before any irreversible action, does not train models on customer data, and lets the administrator choose between models from Anthropic, OpenAI and Google.

Viktor bills in credits shared across the workspace, $25 excl. tax per 10,000 credits, from $50 to $5,000 per month, with no annual commitment, so the bill follows the whole team’s consumption. Forgeron3 bills per user, in euros excl. VAT: €12 on Apprentice, with 10 million tokens per person per month, and a Free plan that gives 100,000 tokens again each month.

Viktor’s data is hosted in the United States, with AWS, and European hosting is reserved for the Enterprise contract. Forgeron3 hosts in France, in Strasbourg and Gravelines, whatever the plan. On meetings, Viktor’s pages mention meeting follow-up and audio file transcription. Forgeron3 produces video meeting notes from the Free plan.