Our own product
Seven agents that find, design and run your automations.
The Borek AI Suite is the system an AI department works with. It interviews your whole organisation, maps every automation candidate with a business case, turns each one into a build-ready specification on a technical framework, orchestrates the running workflow with people at the gates, and shows in one cockpit what runs, what it saves and what comes next.
For the first time, an objective view of your automation potential.
No gut feeling, no workshop opinion, no consultant’s slide. Today, nobody in a mid-sized company holds the complete map of what is automatable. It sits in the heads of the people doing the work, and capturing it by hand takes weeks per department, surfaces a handful of ideas and ends in one or two pilots that never reach production. The Suite replaces that with a conversation: five to ten minutes per person, on any device, by voice or text, in German or English.
Manual discovery
- Weeks per department
- a handful of ideas, the rest stays hidden
- 1–2 pilots that never go live
- no business case per idea
- nobody owns the next step
With the Suite
- Whole company in two to three weeks
- typically 60+ candidates
- a business case per row (benefit, effort, risk, payback)
- build-ready specification per release
- the next wave visible in the cockpit
Discover. Design. Run. Seven agents, one closed loop.
-
1 · Discovery Agent
- What it does
- Interviews every level (management, division heads, team leads, employees, IT) in nine structured conversations (vision & priorities, potential scan, process map, process walk-through, daily tasks, task deep-dive, technology & data check, review & confirm, big ideas). Captures trigger, volume, rule, steps, systems and a sample document per task.
- Output
- Validated task inventory across the company
-
2 · Process Map Agent
- What it does
- Assembles the inventory into the process landscape: areas, hand-offs, shadow processes that run on inboxes and Excel, manual transfers between systems.
- Output
- Automation map per department and chain
-
3 · Business Case Agent
- What it does
- Scores every candidate on the impact profile (Result, Capacity, Risk, Market, Knowledge) plus effort, tier and payback. Commercial, not technical.
- Output
- Scored backlog with business case per row
-
4 · Specification Agent
(the Framework Engine)- What it does
- Turns a released candidate into a build-ready technical framework: components, data sources, rules, model use, frontend, human gate, KPIs, acceptance per stage. Today / future process side by side.
- Output
- Specification your Borek team builds from
-
5 · Build Agent
- What it does
- Scaffolds connectors, rule modules, prompts, tests and documentation on the platform foundation; your builders review, complete and harden. The second block is faster than the first, the tenth faster still.
- Output
- Code, tests, runbook on the foundation
-
6 · Orchestration Agent
- What it does
- Runs the workflow in production: time- and event-driven, deterministic rules for every figure, language models for reading and drafting, exceptions routed to the named person with full context. Write-back only through approved gates.
- Output
- Running automation with people at the gates
-
7 · Steering Agent
- What it does
- The cockpit: automation level per department, hours saved, value contribution, exceptions, errors, dark-processing rate per block, and the proposal for the next wave.
- Output
- Steering instead of gut feeling
Employees simply talk and can share documents.
-
Agent
“Which task do you do almost daily and consider routine?”
-
Employee
“Matching invoices with purchase orders, about 60 a day.”
-
Agent
“Understood. Can you share a current invoice?”
What the Suite captures
- trigger
- invoice arrives in the AP mailbox
- volume
- ~60/day
- rule
- block at >2 % deviation from PO
- steps
- read → match → book
- systems
- Outlook, SAP
- sample
- invoice_4711.pdf
That becomes a build-ready requirement. Two scans, then one deep-dive: everyone names their manual tasks first; depth only where it counts.
Sixty-one found automations are not fate. They are a calculation: team size times time.
-
Finance
- Candidates
- 14
- Tier 1 / 2 / 3
- 9 / 4 / 1
- Potential hours/month
- 310 h
-
Customer service
- Candidates
- 12
- Tier 1 / 2 / 3
- 7 / 4 / 1
- Potential hours/month
- 260 h
-
Logistics
- Candidates
- 9
- Tier 1 / 2 / 3
- 6 / 3 / 0
- Potential hours/month
- 180 h
-
Purchasing
- Candidates
- 8
- Tier 1 / 2 / 3
- 5 / 2 / 1
- Potential hours/month
- 140 h
-
HR
- Candidates
- 7
- Tier 1 / 2 / 3
- 5 / 2 / 0
- Potential hours/month
- 95 h
-
Sales
- Candidates
- 6
- Tier 1 / 2 / 3
- 4 / 2 / 0
- Potential hours/month
- 80 h
-
IT
- Candidates
- 5
- Tier 1 / 2 / 3
- 3 / 1 / 1
- Potential hours/month
- 70 h
-
Total
- Candidates
- 61
- Tier 1 / 2 / 3
- –
- Potential hours/month
- ~1,135 h
Illustrative map for a mid-sized company. Filled with your numbers, not estimated. Team S (2 juniors + PM) works the backlog in ~18 months, Team M in ~12, Team L in ~9: the same backlog, three speeds.
One cockpit: what runs, what it saves, what comes next.
- Backlog
- Automation map
- Reach
- Rollout
- 58 % automation level
- 23 automations live
- ~1,240 hours saved per month
- €55k value per month
Per department
Running
- invoice check (three-way)
- returns triage
- delivery-date bot
- onboarding package
Architecture
-
Borek AI Suite
The intelligence layer
-
Platform foundation
- Borek connector
- integration layer with one read-only server per source system
- data foundation
- AI & rule services
- orchestration & operations
- identity & security
- cockpit
-
Your systems
- ERP
- M365
- DWH
- specialist systems
Three pillars
Delivered, not built.
- Running automations, not roadmaps
- Fixed prices you know before you start
- A German project lead you can call
People at the gates.
- AI drafts, checks, orchestrates
- Your people decide
- Three maturity levels, earned by accuracy
Above your systems, not instead of them.
- SAP, AS/400, M365 keep leading
- Read-only connectors, gated write-back
- Foundation built once, paid once
We automate tasks, not people.
Automation only succeeds if your teams help find it. So the message from leadership, managers and the agent itself is always the same: we take away the dull work and give the time back. The agent asks about tasks and processes, never about people. Your ideas decide what gets built, and you get the hours back first.
Every pipeline ships with a works-council briefing, an employee FAQ and a communication plan, written for German companies.
Plans: the Suite is a subscription by company size plus a one-off set-up; maintenance, updates and hosting included; EU hosting in your tenant or ours. No prices on the site: the plan is named in the demo call.
How we work
Start where it makes sense for your balance sheet.
Automation
Project
- Fixed scope, fixed price, fixed date
- Price known before you start
- Ideal for the first automation
Process-as-a-Service
- Rent running processes monthly
- No capex, operation included
- Three months’ notice, any time
Integrated
AI Team
- Your permanent AI department
- Same builders every month, your backlog
- Monthly rate, three months’ notice
FAQ
Is this process mining?
No, mining reads system logs; the Suite interviews people and finds what runs outside the systems.
Do we need a licence to start?
No, the Discovery Agent already runs inside every free AI Assessment.
Who sees the interviews?
Your project sponsor and the Borek project lead; evaluated per task, never per person.
Which models?
EU endpoints behind an abstraction layer; interchangeable; nothing trains on your data.
Can our own developers build from the specifications?
Yes, the specification is yours, with or without a Borek team.
Start with one process.
A 45-minute call. We look at two or three of your processes and tell you honestly whether automation pays. If it does, you get your complete pipeline designed free of charge, and the first automation live in 30 days.