United StatesFinance and insurance2026 – ongoingOffDeal
Archie, a single Claude-based agent for M&A advisory work
OffDeal's bankers built a deck in about an hour with an AI agent, against 30 to 40 hours for a traditional team
Problem
OffDeal is an investment bank that helps founder-led businesses sell. Each banker handles 5 to 8 deals at once and meets 2 to 3 new potential clients a day. Building buyer lists and decks by hand took analyst teams days or weeks.
Approach
OffDeal's five engineers moved from more than 12 separate agentic workflows to one general-purpose agent, Archie, built on the Claude Agent SDK. Archie connects to more than 20 data sources. A buyer-sourcing agent runs on its own for up to 4 hours across 10 sourcing methods. After the switch to the SDK and changes to tool design and prompting, the internal eval score rose from 25% to 85%.
Results
Takeaway
A small bank put the work of analysts into one agent and kept bankers on client contact. The company did not say how it measured the deck and cost comparisons.
Categories
- Size
- Small business
- Industry
- Finance and insurance
- Business function
- Internal work and knowledge management
- Innovation type
- Process innovation
- AI technology
- AI agents
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Anthropic's customer story describes how OffDeal consolidated more than 12 agent workflows into one agent, Archie, built on the Claude Agent SDK. It gives time and cost comparisons for decks and buyer lists, an internal eval rise from 25% to 85%, and 8 closed deals worth $91M in the firm's first year. The page carries no date; the Internet Archive first saved it on April 15, 2026.
Full details are in the original sources.