How long AI implementation takes.
With Ortelian, it starts with a two-week audit on site. Then we build in two-week cycles. By the end of the first two-week cycle, you already have a first part of your world model live, your people can use it right away, including from their own AI tools through MCP that we set up with you, and a first piece of work is automated. From there it is iterative: every next two-week cycle extends the world model and automates more.
This page is an honest range for buyers, not a promise. For the step by step method, see how to deploy AI agents in an enterprise. For how an engagement is structured, see how we work.
The short answer
A two-week audit, then a first part of your world model live after the first two-week cycle.
Ortelian is a platform plus a forward deployed team, not a consultancy. The audit takes two weeks on site, with the people who do the work and the systems they use. Then delivery runs in two-week cycles. At the end of the first cycle, a first part of your world model is live and your people can use it right away, in Ortelian and from their own AI tools through MCP, set up as part of the engagement. A first piece of work is automated in that same cycle. Each further two-week cycle extends the world model and automates the next piece of work. There is no big launch at the end; value arrives every cycle.
When the sources and the standard are already clear, it can be faster still. In one public sales example, a ranked account list, a daily refresh and meeting prep shipped about a week after the first call. The write-up is one week at one company.
What the calendar looks like
Audit, first cycle, then iterate.
- Weeks 1 and 2: audit on site
Time on site with the people who do the work, plus reading the systems and procedures. Output: a map of the work, the constraint, a redesigned process, the scope of the first build cycle and a benchmark to measure against.
- First two-week build cycle: world model live, first work automated
By the end of the first cycle, a first part of your world model is live and your people use it right away, including from their own AI tools through MCP that we set up with you. A first piece of work is automated and evaluated on real examples against the agreed standard.
- Every next two-week cycle: extend and automate more
Each cycle has a written scope and a pass bar. It extends the world model to more of the company and automates the next piece of work. Your team runs it or we keep running it; either way we host and operate the platform in the EU and evaluations keep running on live work.
What actually drives duration
Clarity of sources and standard, not model choice.
The calendar stretches or shrinks with the work, not with which language model you prefer. The drivers we see most often:
For the redesign step, see redesign the work before you automate it. For the standard, see agree the standard before you build. For why pilots die without those, see why AI agents fail in production.
- Sources and identity
If the facts the work needs already live in joinable systems, with one customer or order meaning the same thing across them, the first build moves. If the territory lives in spreadsheets, registries and three CRM spellings of the same company, identity work comes first.
- A written standard on real examples
Agreeing what good looks like, including the messy cases, before code starts is what keeps cycles short. Without it, every demo restarts the debate. More in agree the standard before you build.
- How much redesign is needed
An agent on the current process makes waste faster. The first weeks often redesign what belongs to software, agents and people. That time is not optional if you want a lasting result.
- Access and an internal owner
Agreed access to systems, time with the people who do the work and an owner who can decide. Missing any of these pauses the calendar more than any technical choice.
- Scope of the first job
One clear job (daily account list, invoice matching, planning prep) beats a company-wide assistant. Wide scope without a standard is why many pilots never end.
What this is not
Not a six-month transformation programme by default.
Industry surveys keep showing a gap between AI use and business result. In McKinsey's 2025 survey, 88% of companies use AI in at least one function and 39% report any EBIT impact. Long programmes that pick a tool first and define done later are a common way into that gap. Ortelian's answer is the opposite shape: start from one piece of work, audit it on site, agree done on real examples, then build in short cycles.
We are also not a self-service product with a setup wizard. Access to the platform comes with an engagement. There is no public API or public MCP server. Inside an engagement, your own agents can use the platform through scoped tools, and it can plug into tools your team already uses.
Examples of pace
What we can say from public Ortelian writing.
We do not invent case study durations. These are the timings already stated on ortelian.com:
See AI agents for sales and AI agents for operations for the kinds of jobs. See the platform for what agents run on after launch.
- Audit
Two weeks on site.
- Build cadence
Two-week cycles. After the first one, a first part of the world model is live and usable, and a first piece of work is automated. Each further cycle extends it.
- Narrow sales first build
About one week from first call to a live ranked list, daily refresh and meeting prep, when sources and standard were clear (one week at one company).
- Functions we write about
Sales and operations role pages describe the jobs we start with. The audit still decides which job comes first for you.
How to shorten the path
What you can do before the audit.
None of this replaces the on-site audit. It makes the audit faster to scope. Start from the AI readiness assessment if you want a checklist, or from what an AI deployment partner does if you want the engagement shape.
- Pick one piece of work
Name the job, the owner and what a correct result looks like on a few real cases, including awkward ones.
- List the systems of truth
Where do the facts actually live today, including spreadsheets and outside sources the CRM never held?
- Name an internal owner
Someone with authority to decide and clear blockers, ideally a founder or leadership sponsor.
- Check readiness early
If you want a structured self-check before a call, use the AI readiness assessment, then talk scope.
Where Ortelian fits
Where this timeline fits, and where it may not.
Ortelian fits mid-market and enterprise knowledge work where the constraint is context no single system holds: sales, operations and similar. We work on site not because our product needs it, but because your company does. After launch, your team runs it or we keep running it. Detail is on how we work.
Where it may not fit: if you need a multi-year transformation across dozens of teams with a global systems integrator, a different shortlist is better. If you only need a chat workspace for every employee, start with ChatGPT Enterprise or Copilot, not with us. If the work is a stable, high volume click path with little judgement, an RPA specialist may be the better first call. If nobody inside can own the work or grant access, fix that before any vendor starts a clock.
Questions
Questions buyers ask about AI implementation time.
How long does AI implementation take with Ortelian?
A two-week audit on site, then two-week build cycles. By the end of the first cycle, a first part of your world model is live, your people can use it right away, including from their own AI tools through MCP that we set up with you, and a first piece of work is automated. Every further cycle extends the world model and automates more.
Can a first AI agent go live in a week?
Sometimes, when the job is narrow and the sources and standard are already clear. One public sales example shipped a ranked list, daily refresh and meeting prep about a week after the first call. As a rule, plan on the first piece of work being automated by the end of the first two-week build cycle.
When can our people start using the world model?
At the end of the first two-week build cycle. The first part of the world model is live then, and your people can use it in Ortelian and from their own AI tools through MCP, set up as part of the engagement. It grows with every cycle after that.
Why do AI projects take months or years elsewhere?
Often because done was never written down, the process was automated as it stood, identity across systems was left unsolved or the programme started from a tool instead of one piece of work.
What slows an Ortelian build down?
Unclear sources, no written standard, missing system access, no internal owner and a first scope that tries to take on the whole company at once.
Is the audit included in the timeline?
Yes. The audit is the first two weeks, and it sets the scope of the first build cycle.
Who runs the system after the build?
Your team or Ortelian. Either way we host and operate the EU-hosted platform and keep evaluations running.
Bring one piece of work and a rough clock.
On a call, we say whether an audit is worth it, what would drive the duration in your case and where Ortelian fits or does not.
