A finance director gets a Slack message at 4pm asking for Q3 P&L by EOD. She'd normally spend three hours reconciling accounts across five systems, matching invoices to payments, calculating accruals, and formatting the report. She types a prompt into Doe describing what she needs. The AI agent pulls data from all accounting systems, reconciles discrepancies, generates the P&L, and delivers it formatted in 20 minutes. She reviews it, makes two adjustments, and sends it up the chain.
Doe executes multi-step work through natural language prompts. You describe a task, research competitors, build a financial model, create a prospect list — and the agent handles the execution by connecting your existing tools and delivering completed results. This isn't a chatbot that answers questions. It's an executor that does the work.
A product manager needs to understand which features actually drive revenue. Manually, she'd export data from analytics tools, cross-reference with billing records, build comparison tables, and calculate ROI across 15 features. With Doe, she asks for feature ROI analysis. The agent pulls usage data, matches it to revenue changes, builds the comparison tables, and delivers the analysis. She can then ask follow-up questions or request different cuts of the data.
The system handles complete workflows. A marketing manager can ask Doe to launch a campaign, and the agent researches the target audience, creates content, schedules posts, and monitors performance. An engineering lead dealing with a P1 incident can prompt Doe to identify root cause, rollback problematic deploys, and coordinate the response team. A finance team doing M&A work can request acquisition models, and the agent builds financials, pulls comparable company data, runs DCF analysis, and generates investment memos.
Tasks consume credits based on complexity. Simple queries burn through fewer credits than multi-step processes involving large datasets or multiple AI models. The free plan provides 150 credits monthly with a three-seat maximum and documentation support. That's enough to test the system but won't cover daily operational work for most teams.
Standard plans run $45 per user monthly when paid annually, providing 1,500 credits per user with unlimited seats and the Loops feature for recurring tasks. Pro accounts get 8,000 credits per user monthly at $225 annually, plus usage analytics, Slack support, and early beta access. Enterprise customers get custom credit volumes, dedicated account teams, SSO, VPC deployment, and uptime SLAs.
The credit system creates planning friction. You can't predict exactly how many credits a complex task will consume until you run it. A financial model request might use 50 credits or 500 depending on data volume and calculation complexity. Teams need to monitor credit burn rates and adjust usage or upgrade plans accordingly.
Doe works best for teams drowning in structured, repeatable work — monthly closes, pipeline analysis, competitive research, campaign launches. If your team manually builds the same spreadsheets, pulls the same reports, or performs the same multi-system workflows repeatedly, Doe automates that pattern.
It doesn't fit exploratory or creative work well. A brand strategist developing positioning concepts or a researcher conducting unstructured interviews won't get much value. The system executes defined tasks efficiently but doesn't replace strategic thinking or qualitative judgment.
Small teams experimenting with automation can start free. Growing teams running regular operational workflows should budget for Standard. Finance and ops teams processing high volumes of complex work will hit Pro-level credit needs quickly.