Financial services, government, telecom, legal and risk
Back-office agent workflows with approval gates
How do regulated firms run document procedures with AI agents and human sign-off?
Katonic turns a firm's written procedures into agent workflows. The workflows process each day's Excel and PDF files, carry out the steps, and stop for a named person to sign off. Manual work becomes a reviewed batch run, every step is traced for audit, and the files stay on the firm's own infrastructure.
At a glance
| Industry | Financial services, government, telecom, legal and risk |
|---|---|
| Who it is for | COO, Head of Operations, Head of Compliance, process owners |
| Data that stays in place | Source files, working data, outputs and the audit trail |
| Runs air-gapped | Yes |
| Inputs | Excel and PDF files, processed in daily batches |
| Infrastructure | HPE Private Cloud AI, or HPE GPU servers in an HPE AI Factory design |
The problem
Operations and compliance teams in regulated firms run document-heavy procedures by hand: reviews, reconciliations and batch summaries.
Agent pilots stall because they cannot follow a written procedure, take the day's files, stop for sign-off, or show afterwards what the agent did.
How it works
- The procedure is written as a workflow in Katonic Studio, step by step.
- The workflow picks up the day's Excel and PDF files.
- Agents carry out each step and record what they read and what they produced.
- The run stops at each approval gate until a named person signs off.
- The full trace of the run is kept for audit.
On HPE infrastructure
On HPE Private Cloud AI
HPE Private Cloud AI is HPE's turnkey private AI system, co-engineered with NVIDIA. Katonic AI is a partner in the HPE Unleash AI program. On Private Cloud AI, Katonic connects to the models the system serves and adds the agent platform on top. The workflows run against the models served on the system, and the files never leave it.
On HPE GPU servers
Customers who build their own estate on HPE GPU servers, as part of an HPE AI Factory design, get compute, storage and networking from HPE. Katonic supplies the software layer above the hardware: open-weight models served on the customer's GPUs, the AI gateway, the agent platform and governance. Batch runs are scheduled on the firm's own GPU servers, next to the systems that hold the source files.
More on both routes: Katonic AI on HPE.
Where it stands
Proven in a pilot with a financial-services firm.
Questions and answers
Can an agent complete a step without approval?
Only steps the firm marks as automatic. Any step set as an approval gate stops the run until a named person signs off.
What file types does it handle?
The pilot processed Excel and PDF files. Other sources are connected through the Knowledge Engine and tool connectors.
What does the audit trail contain?
Each step of each run: what the agent read, what it produced, which model it called and who approved the step.
Do the files leave our network?
No. The workflow, the models and the files all stay on the firm's own infrastructure.