Map
Eight answers. A conservative range in about a minute.
AI IMPLEMENTATION, PROCESS BY PROCESS
AI implementation for companies of 10–1,000 people. One process at a time. Free savings estimate in 60 seconds.
People · Loaded cost · Time · Rate — a formula, not a model
People Loaded cost Time Rate Human in the loop One process Not a model PreliminaryWho this is for
Not for you if you need a chatbot, a content engine, or a transformation deck. We are not that firm.
How it works
Not a chatbot. Not a transformation programme. A ladder you can leave at any rung.
Eight answers. A conservative range in about a minute.
Optional. We read the map with you — a conversation, not a pitch.
One process. Cost → potential → implementation → ROI. Go or no-go.
One process live. People remain in the loop. Weeks, not a programme.
Scale only after a pilot that earned the next step.
Work · examples
Full stories you can open. They are composite examples from our playbooks — not named Aperanda clients. When a real case has a source, it replaces the example.
Hot now: the three processes we can start mapping this month — not a leaderboard. The other 100 process files are in the catalogue.
Method
Every multiplier has a public source — Eurostat SES 2022, the Labour Cost Survey, McKinsey Global Institute bands. Run it on your process. We will not invent a case to decorate this screen.
AI Savings Finder
Eight answers. Personalised map in 60 seconds.
Free estimate — 60 sec →Services
Prices are on this page, not behind a call. Review stays a free step between them — not a fifth product.
Eight questions. A teaser. A report. Email if you want the full range. Not implementation. Not a guaranteed sum.
Discovery of one process. A defensible number. A go / no-go. Not a live pilot. What’s inside ↓
One process in production. Humans in the loop. Not a company-wide rebuild. What’s inside ↓
Integrations, handover, scale. No invented monthly rate on this page.
Inside the audit · €1.5–3K
Interviews with the people who run it. We measure the desk as it is — volumes, hours, exceptions.
Cost today, savings potential, an implementation plan, and ROI. The same formula as the map — every multiplier sourced.
A written report you can hand to your CFO, walked through on a call. It ends in go or no-go — and no-go is a real answer.
Of your team’s time, across interviews and access to the people who do the work. Then we work.
Typically two weeks, confirmed in writing before you pay. No hourly billing. Fixed scope, fixed price.
From day one you follow the work in a client workspace — tasks, builds, chat. The screen at the end of this page is a preview of it.
Inside the pilot · €10–20K
The one the audit priced — live in the systems you already run. No new platform.
Exceptions, approvals and go/no-go stay with your team, on a visible queue — not hidden in a model.
The client workspace from day one: tasks, builds, chat. The screen at the end of this page is it.
Weeks, not a programme — the exact length is fixed in the audit, in writing, before you pay. Production only if the pilot earns it.
Your data & documents
Written for the person who has to say yes to sharing documents. Commitments of process — not badges.
“Your intellectual property sits above the model” means exactly this: your documents and rules stay yours. The model works under them for the job at hand — it does not keep them, and they do not become training data.
You contract with Aperanda S.L. (en constitución) · Barcelona, Spain. Questions before sharing anything: ceo@aperanda.com · Privacy policy
The studio
Six people. One expensive process. A numbered map, then a pilot you can stop. We do not sell “AI transformation.”
Seat 01
Name forthcoming
The map, the audit, the pilot. Go or no-go stays with the client.
Seat 02
Name forthcoming
The desk as it actually runs. Exceptions, approvals, humans in the loop.
Seat 03
Name forthcoming
Wire the process into the systems you already have. Not a new platform.
Seat 04
Name forthcoming
The formula, the sources, and what the number is allowed to claim.
Seat 05
Name forthcoming
The workspace the client sees. Handover. Pages that stay current.
Seat 06
Name forthcoming
How the corridor runs — orders, TMS, exceptions — before we touch a model.
The team holds vendor certifications — Microsoft Azure AI Apps & Agents, NVIDIA DLI, NVIDIA Generative AI LLMs, Google Cloud Professional ML Engineer, AWS Machine Learning Engineer. Team-held, not a partner programme.
Before you write
No. It is the hook of our campaign and it scales with company size. Your map is a preliminary range from a formula. We will not write “we guarantee €100K.”
No. Humans stay in the loop — exceptions, approvals, go/no-go. We will not advertise “six people replaced.”
Weeks, not a programme. Order of magnitude: a few weeks for one process. Exact length is confirmed in the audit — we will not invent a slogan number here.
Three example stories are on this site so you can see the format. They are composites from the playbooks, stamped Example. A named Aperanda client goes live only with a source. Our founder’s employers are a career track, not Aperanda’s cases.
Yes. The map is free. Review is a conversation. Audit and pilot are paid. You can stop after any step.
Each step sees only what it needs. Not used to train models, processed in the EU, NDA by default, deleted on request when the engagement ends. The full commitments: Your data & documents.
Discovery of one process, a sourced before → after number, a written report walked through on a call, and a go / no-go. Fixed scope, fixed price. Details: Inside the audit.
For the audit — typically 2–4 hours across interviews and access to the people who do the work. Then we work. See Inside the audit.
Generic chatbots as a service. Invented clients or percentages. “AI transformation” programmes. A monthly rate invented for Production. We do one expensive, repetitive process: a conservative number with sources, then implementation with people in the loop.
Our products · under the hood
We sell four engagements. These are the two systems we install inside them — not separate line items, not SaaS.
Memory
Aperanda’s company-memory, implemented inside your processes. Your desk’s decisions, exceptions and rules stay available the next time the same work runs — in intake, review and handover, not in a side chat. More than a search overlay: persistent, governed, and still overridable by a person. Implemented in your stack, scoped after an audit or pilot.
No public fee. No named Memory client on this page. Related to Virtual Organisations — not the same product.
VO
Virtual Organisations — roles, permissions and message rules compiled into the systems you already run. Who may speak, who may approve or block, what a message may carry, and which memory a role may see. Writes to shared records need an approval step by design, and the audit trail comes with it. Implemented inside your stack, scoped after an audit or pilot.
No public fee. No named VO client on this page. Related to Memory — not the same product.
At AI Summit Barcelona?
22–23 September 2026. The number on the shirt is not a guarantee — it is a hook that scales with company size, computed by the same formula as the instrument above. Scan the QR on the shirt, or run it right here.
Contact
One email, straight to a person. Tell us which process eats the most hours — or just say hello.
Get your free savings estimate
60 seconds · no sales call
A person reads it and replies from the same inbox. No list, no sequence — marketing only if you ask for it.
Or connect on LinkedIn →Your email is used to reply — nothing else. No marketing without an explicit opt-in.
Preview — not a live client. This is the workspace you get during audit and pilot: tasks, builds, chat. The screen is a mock-up; the company on it is illustrative.