ItalyProfessional services2026 – ongoingJAKALA
Claude agents working beside senior delivery teams on client campaign optimization
JAKALA cut repetitive work by about 70% and shortened campaign optimization from 5-7 days to under 24 hours with Claude agents
Problem
JAKALA, a Milan-based data and AI company with more than 3,600 staff, had senior planners who spent 40-60% of their week assembling data, building reports and stress-testing hypotheses. A full optimization loop for a client took 5-7 working days and pulled data from 8-12 systems.
Approach
JAKALA rolled out Claude Enterprise to more than 3,000 employees and built production agents on the Claude Agent SDK. Each major client account runs a network of performance agents that connect to ad accounts and JAKALA's geo-intelligence engine, while a human pod reviews and carries out the work. It uses Opus 4.6 for the hardest analysis, Sonnet 4.6 as the default for agent workflows and Haiku 4.5 for high-volume steps. New agents now reach client environments in weeks, where the first one took about a quarter.
Results
Takeaway
The saving came from moving data assembly and reporting to agents while people kept review and client judgment. The company did not say how it measured the time savings.
Categories
- Industry
- Professional services
- Business function
- Marketing and salesInternal work and knowledge management
- Innovation type
- Process innovation
- AI technology
- AI agentsGenerative AI
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Anthropic's customer story describes how JAKALA uses Claude Enterprise internally and runs Claude-based agents on client accounts alongside human pods. It reports about 70% less time on repetitive work, an optimization cycle cut from 5-7 days to under 24 hours, about 30% higher team productivity and 35-40% higher client satisfaction on hybrid accounts. The page carries no date; the Internet Archive first saved it on June 18, 2026.
Full details are in the original sources.
