A company can buy AI tools and still struggle to choose a useful next step. Leaders may disagree about priorities; teams may lack access or confidence; pilots may have no owner. Another general introduction will not resolve every one of those problems. Start by identifying which barrier is actually holding the organisation back.
Updated 4 October 2026: this guide now focuses on adoption decisions, with open planning templates and links to the relevant private AISB formats.
What enterprise AI adoption involves
Enterprise AI includes predictive models, generative tools and systems that can act across workflows. Their usefulness depends on the task, inputs, access, review and operating context. A model's capabilities do not by themselves establish whether it belongs in a particular business process.
Use this guide to decide what to clarify next. It is not a market-adoption survey, a maturity assessment or a claim that every organisation should follow the same sequence.
Diagnose the barrier
| What you observe | Check first | Useful next output | Possible next step |
|---|---|---|---|
| Leaders discuss AI but disagree about priorities | Is the disagreement about facts, goals or trade-offs? | Shared question or priority record | Briefing for one information gap; workshop for group decisions |
| Employees have tools but do not use them | Do they have access, permission, a relevant task and support? | Defined skill or access issue | Resolve access; then practise the actual task |
| Many ideas, little testing | Are the tasks specific and comparable? | Shortlist with inputs, reviewer and evaluation | Use-case discovery and a bounded experiment |
| Several demos, no further work | Who owns the next decision and has time to act? | Handover, scope and review date | Assign ownership before adding another event |
| A team wants outside perspectives | Which assumption could another practitioner help test? | Comparison notes and business implications | Targeted discussion or learning visit |
| A wider audience needs context | What should each audience segment take away? | Common language and routes to follow-up | Internal AI day with appropriate tracks |
One organisation may have several barriers. Give each one an owner; do not collapse them into a single “AI readiness” score.
Choose a bounded experiment
Describe a workflow in enough detail to compare the current and proposed approach. A generic aspiration such as “use AI in marketing” needs a narrower task, permitted inputs and a quality criterion.
For example, a fictional first test might compare manual and AI-assisted drafting of an internal summary using synthetic notes. A reviewer checks omitted decisions and incorrect actions; the team records total effort, including review. This illustrates a test design. It is not a reported client result or evidence that the task is suitable for every company.
Use the use-case discovery worksheet to record the baseline, reviewer, dependencies and stop conditions. Include a simpler process or ordinary automation as an alternative when it could solve the problem.
Match the programme to the need
A private executive briefing helps illuminate one question. An AI strategy workshop helps leaders compare initiatives and agree priorities. A private AI event supports a wider programme and audience within an agreed scope.
Training develops a defined skill through practice and feedback. A hackathon tests a bounded challenge with appropriate tools and support. An expedition brings outside perspectives into a learning question. These are different formats, with different dependencies; the comparison guide explains when each is useful.
AISB's published private offers are the workshop, briefing and event. This guide does not imply a confirmed training curriculum, available hackathon mentors or guaranteed host access. Additional delivery requirements need confirmation in a proposal.
Evaluate before expanding
Pick measures before the experiment. Record quality, total effort, review requirements, failures and the conditions under which the observations apply. Do not equate a fast first draft with a completed task.
Keep separate records for participation, satisfaction, learning, continued use and operational outcomes. A positive event survey answers a different question from an evaluated workflow. Avoid describing one as evidence of the other.
Use the thirty-day follow-up plan to decide whether to continue, change, pause or stop. A documented stop can be useful when it prevents a larger commitment to an unsuitable approach.
Governance and ownership
The NIST AI Risk Management Framework Playbook provides voluntary guidance organised around Govern, Map, Measure and Manage. It is a reference for risk questions, not a certificate that a workshop or tool makes an organisation compliant.
For your experiment, identify the permitted inputs, approved tool, human reviewer, failure-handling process and person authorised to decide what follows. The EU AI Act guide provides a separate introduction to the regulatory context; the scope of an actual deployment requires appropriate internal review.
When systems can take actions, make the allowed actions and approval boundaries explicit. A demonstration is not authorisation to connect an assistant to production tools or sensitive records.
Prepare the next conversation
Use the executive readiness checklist to clarify the question and decision-makers. Use the provider brief to specify the output and responsibilities. Use the scope and budget framework to compare inclusions.
AI Summit Barcelona connects its public Summit programme, recordings and private company formats through these planning resources. The Digital Pass is a separate product: review the available catalogue and access conditions before using recordings in a team learning plan.
Methodology
About this adoption guide
Updated by AISB's editorial team on 4 October 2026. The diagnostic and fictional example are editorial planning aids. No universal adoption statistics, client savings or return-on-investment guarantee are asserted.
Find the next step for your organisation
Share the barrier, audience and decision you want to move forward.
Sources
Frequently asked questions
What is enterprise AI?
Enterprise AI is AI used within an organisation's workflows and systems, with defined access, review, ownership and measures. It can include predictive models, generative tools and systems that take actions; the organisational context matters as much as the technology.
Where should a company start with AI adoption?
Identify the actual barrier and business task. Clarify the owner, current baseline, approved tools and permitted inputs, then define a small experiment with a reviewer and a next decision.
What if employees have AI tools but rarely use them?
Check whether the barrier is access, relevance to the task, skill, unclear permission or lack of support. Training can address a skill gap but cannot fix a missing approval or an unsuitable workflow by itself.
How should we assess AI value?
Compare the chosen task with a stated baseline, including output quality, total effort, human review and rework. Record assumptions and limitations. A demo or positive reaction does not establish business ROI.
Change log
- — Shortened search metadata, aligned the named author and added contextual learning links; removed statistical callouts.
About the author
- Tanguy WinckerCo-founder, AI Summit Barcelona
Tanguy is a co-founder of AI Summit Barcelona. His editorial work covers company programmes, event planning and the Barcelona ecosystem.
Related reading
Which AI Learning Format Fits Your Business Problem?
Start with the decision or workflow you need to improve. This comparison explains what each format can produce, when to choose another approach and how to write a useful brief.
AI Use-Case Discovery: Agenda and Prioritisation Template
Start from a task rather than a tool. Record the baseline, permitted inputs, expected output and owner; then use evidence and unresolved dependencies to decide what to test.
After the AI Workshop or Hackathon: A 30-Day Plan
A practical follow-up template for turning a decision record or demonstrator into an owned experiment. Separate event satisfaction, learning and operational evidence.