Rameo interviews your people, turns their answers into evidence-backed BPMN models and finds where work gets stuck. Next, it builds the fix: quick wins implemented for you, bigger changes built on the full process context.
What Rameo produces
The problem
Workshops sample opinions, consultants leave slides behind, and every automation project starts by asking the same people the same questions again.
The exceptions, the re-keying, the “ask Petra first”: the real process is known only to the people who run it.
Process maps get drawn once for an audit or a project and are outdated by the time they are approved.
Without a shared model nobody can say where work gets stuck, who is affected or what a change would touch.
Copilots and agents are only as good as what they know about how your company actually works.
How it works
Four steps on one model. Each step feeds the next, and every result links back to what people said.
A spoken AI interviewer talks to every participant in their language and follows up where answers stay vague.
InterviewsAnswers become a Process House and editable BPMN models. Every step links to the statements behind it.
BPMN 2.0Rameo surfaces gaps and bottlenecks, drafts TO-BE processes and lets the people affected confirm them.
AS-IS → TO-BEAccepted improvements become working solutions, built and tested inside Rameo with the people who asked for them.
In developmentProduct tour
Four short click-throughs of the real product, on example data from a fictional manufacturer.
Elena opens her tasks: a follow-up interview is waiting for her.
Product
Everything in Rameo traces back to a conversation, and everything can be checked, edited and exported.
A spoken AI interviewer talks to every participant in English, German, French or Italian and follows up where answers stay vague. People answer at their own pace and confirm a summary before anything is used.
Each BPMN step shows the interviews and statements behind it, down to the verbatim quote and the conversation it came from. Models stay editable, are checked for errors and keep every revision.
Management, core and support processes in one Process House: numbered, owned and connected by handoffs and shared systems. From the value chain, one click opens any BPMN model.
Every opportunity comes with a TO-BE model and a visual diff against today’s process. The employees affected confirm it before anyone commits, and an accepted TO-BE becomes the next revision.
The process brain
Everything Rameo learns lands in one versioned model: steps, roles, systems, rules, evidence and know-how. People read it, documents are generated from it, agents query it.
The Process House for leads, “My processes” for every employee and the know-how on each step, visible by role.
Standard BPMN 2.0 export with step notes as annotations. Work instructions and SOPs generated from the model are in development.
A local MCP server hands agents permission-filtered context for any process or step, with every read logged.
Before an agent drafts a work instruction or touches a process, it can load what Rameo knows: the model outline, owners, systems, rules and the know-how people shared — filtered exactly as Rameo filters it for the token’s owner.
◆ Load the Rameo context for supplier invoice matching ↳ rameo · get_context(process: "02.11")# Rameo context: 02.11 Supplier invoice matching and payment releaseLevel L2 · model revision 4 · 02 Core processes## Process summary- Owner: Nora Berger- Systems: Excel, Outlook, SAP, Teams- Downstream: 01.01 Financial reporting; 03.03 Purchase to receipt## Model outline1. Supplier invoice received [startEvent]2. Check invoice for duplicates in SAP5. Enter header and run three-way match in SAP7. Match within CHF 50 tolerance? [exclusiveGateway]… 17 items · 1,124 tokens✓ sha256:8c1d741a…78e4d3 · read recorded in the audit log
Build · in development
Most tools stop at the process map, and a context layer only hands the work to someone else. Rameo goes the rest of the way: documentation, automations and small internal services are specified from your TO-BE, built and tested inside Rameo and run on its infrastructure, hand in hand with the people who do the work.
It starts from what is live today: an accepted opportunity with evidence and a confirmed TO-BE. A person presses Build; nothing starts on its own.
Rameo drafts the brief from the TO-BE and the step context: problem, roles, systems, rules and success criteria. Raw transcripts never enter it.
An agent builds from a tested service kit in an isolated sandbox. Every iteration is tested and checkpointed.
The employees who raised the issue open the prototype from the opportunity they validated, rate it and ask for changes.
Feedback becomes the next brief. The process owner approves every iteration before anyone sees it.
Adopted services run on Rameo infrastructure, isolated per customer and linked to their process step. Usage flows back into the model.
◆ Build a service · “Triage blocked invoices” ↳ 02.11 Supplier invoice matching · step “Block payment in SAP”✓ Brief drafted from the TO-BE · shape: checklist with reminders✓ Shareable text only · no transcripts, no audio✓ Sandbox build · 14 tests passed · checkpoint v1→ Prototype shared with Accounts Payable · 4 testers 4.5 / 5 average · 2 change requests✓ Iteration 2 · change requests addressed · checkpoint v2◆ Waiting for approval by Nora Berger, process owner
Work instructions, SOPs and onboarding guides generated from the BPMN and the know-how on each step. Regenerated when the model changes, approved by the process owner.
Steps your TO-BE marks as automated become running automations between the systems the process already uses.
Small internal tools built for one process step and used by the people in it, from a closed set of proven shapes.
Low-hanging fixes run the whole flow above: specified, built, tested and shipped inside Rameo.
Changes across systems or departments are identified with evidence, scoped in Rameo and built on the same process context at a larger scale: with your IT, your partners or agents over MCP.
Roadmap
What ships today is the foundation. Documentation, automations and services are being built on the same model, and pilot partners shape them first.
Discovery, modelling and improvement on one evidence-backed model.
Work instructions, SOPs and onboarding guides generated from the living BPMN, regenerated when the process changes and approved by its owner.
From an accepted TO-BE to a working solution: quick wins built in Rameo, bigger changes built on the full process context, tested by the people who asked for them.
Trust
Rameo models processes, not people. Who sees what, what the AI may do and what it costs is decided by you, and recorded.
Admin, transformation lead, process owner and employee. Every read and write is checked server-side, per organization.
Let authorized leads open raw transcripts, or keep them with the participant only. Process owners never see them.
AI output is always a proposal. Verified models, history and ownership are never silently replaced.
Changes, settings and every agent context read are recorded with who, what and when.
Daily, monthly and per-operation budgets. Without a budget, AI calls are refused.
Standard BPMN 2.0 export, a personal interview export for every employee and a self-hostable deployment.
Why now
Language models can hold many structured, adaptive conversations in parallel, in each participant’s language. Discovery no longer has to sample a few voices.
BPMN is an open, machine-checkable standard. Generated models can be validated, diffed, edited and exported instead of living in slides.
Companies are rolling out agents faster than they document how work runs. Without that context, automation starts in the wrong place.
FAQ
A 30-minute walkthrough: Rameo on example data, your questions, and what a pilot on your own processes would look like.
