AI in GTM

How Do You Build AI Capability in a Commercial Team?

Most GTM leaders say AI already shapes their go-to-market, but few rely on it for decisions. Real capability starts with one task worth solving, proven, then made to stick.

A black chess pawn in focus on a chessboard, with blurred pieces and blue bokeh lights behind
Olga Papagiannopoulou
Written by
Olga Papagiannopoulou
October 2026 · 8 min read

We asked over 100 GTM leaders across Europe how AI is shaping their go-to-market. Most say it already has real influence. Far fewer have made it part of how decisions actually get made.

Building real AI capability in a commercial team means treating that gap as an adoption problem, not a technology one, and as the starting problem: find one task worth solving, prove it works, then build the behavioural and organisational change that makes it stick, before scaling further.

If you get this sequence wrong, the tool will be quietly abandoned, no matter how good the prompting is.

Why do teams resist, even when leadership is pushing adoption?

In our GTM Confidence Index FY27, 81% say AI already has a significant influence on GTM or is central to it. But for 70%, that influence is still concentrated mainly on enablement and productivity. AI is being used; it isn't yet being relied on.

How AI shows up in GTM today
AI has a significant influence on GTM or is central to it 81%
AI influence is mainly in enablement and productivity 70%
AI-assisted insight is a primary decision input 18%

Source: Magnus Consulting, GTM Confidence Index FY27 (100+ GTM leaders in Europe)

Resistance shows up because leadership frames AI as a technology rollout, while employees experience it as a change to their role, their expertise and their control over their own output.

Adoption is the harder problem. A team can have access to every tool and still work the way it always has, because nobody has shown them where AI fits into the work they actually do. When adoption is done well, the difference shows up in the work itself: briefs start from the same information, repetitive tasks move faster, and people spend the time they get back on the parts of the job that need their experience.

Our data points the same way. Nearly three in four organisations describe their technology as modern but fragmented, so adding more tools rarely closes the gap on its own. Leaders who are very confident of hitting their growth targets are far more likely than the somewhat confident to use AI-assisted insight and feedback loops as a primary decision input: 31% versus 11%.

Real AI capability doesn't come from telling employees to "use AI" across every process. It starts narrower, with leadership naming one specific issue worth solving before AI is even mentioned and giving people a clear line between what they own and what the tool does.

How do you get a team to actually use AI?

Once that answer is clear, adoption follows, because people can see AI closing a real gap: connecting a GTM plan, a value proposition, personas and brand guidelines that currently live in separate places into one system.

From there, three things do the actual work.

  1. Show people the safety net before you show them the tool. Explain what happens if it gets something wrong or pulls in bad data, and what's in place to catch it. That's usually the first thing anyone good at their job wants to know.
  2. Run a workshop against every job to be done in the process, and map, task by task, what a person owns and what AI produces. Without that clarity, people quickly start fearing AI is replacing them rather than helping them.
  3. Start building a shared system: a prompt library, a best-practice workbook, and real examples from colleagues showing where it actually helped.

How do you make sure AI helps your team do the job, without taking the job away?

Once people can see what AI is doing for them, the next question is usually more personal: what happens to my role? It's a fair question, and the answer depends on how AI is brought in. There are two very different ways it can change someone's work, and it helps to name them early.

  • Augmentation means AI supports someone in their job without taking it away from them. The person still owns the task, and AI helps them do it faster or better, as long as the brief is right.
  • Replacement means AI does the task instead, and the person is no longer needed for it.

From the outside, the two can look the same, because both change how work gets done.

Our GTM Confidence Index makes a similar point: the aim is to help teams see more and make better-informed decisions, with people making the final commercial call. So, we have to decide from the start that AI is there to support people and set things up to keep it that way. That comes down to three things:

First, we need to redefine what productivity means before rolling anything out. If productivity is still judged purely by volume of output, AI just produces more of the same, faster, and the time it saves disappears into more work rather than better work. Decide up front whether the measure is quality, task completion, or confidence in the result.

Second, build people, not just workflows. Handing someone a finished prompt library gets them using AI. Teaching them to write, test and refine their own prompts against their own workload gets them thinking with it, and that's what turns a tool into a capability that outlasts any one project.

Third, keep people at both ends of every output. Someone writes the prompt, they review what comes back, and a second person validates it before anything that matters goes anywhere. That chain is what accountability looks like in an AI-assisted workflow, and what determines whether a role has genuinely been augmented or quietly taken over.

What's the sequence that actually works?

  1. Find the real task, not just a willing team. The need doesn't always show up as a stated problem. Often it only surfaces once you start validating an existing process with AI's help.
  2. Narrow to what's repetitive, labour-intensive and disproportionately time-consuming, then pick one task the owning team agrees is worth streamlining.
  3. Build the smallest workable solution for that single task with the team that owns it, not on their behalf.
  4. Pilot it internally only, and ask the pilot group to log every glitch, every piece of feedback, and anything else it should be able to do.
  5. Implement that feedback and retest performance before touching anything else.
  6. Once the first pilot is solid, extend to the adjacent workflows connected to it, repeating the same pilot-feedback-retest cycle each time.
  7. Assign an owner to every workflow and run it as continuous improvement: product-development discipline, not a one-off deployment.
  8. Put governance around all of it from the start, not once something's already gone wrong.

Who owns it as the team matures?

Three models, matched to maturity.

  • Piloting: a small central or semi-embedded unit, deliberately limited in scope, because speed of learning matters more than coverage at this stage.
  • Scaling: the agents and solutions that worked get handed over to individual teams to test, tailor and adopt for themselves; the centre builds them but doesn't run them forever.
  • Operating system: AI connects the organisation's foundations (the GTM plan, the value proposition, the personas, the brand guidelines) into one system that keeps evolving as the business grows. AI performance is a growth design question, not a tooling one.

Most teams are not there yet: in the FY27 report, only 18% use AI-assisted insight as a primary decision input.

What makes it stick after the initial push ends?

The pattern is familiar: AI gets treated as new and exciting, adoption is enforced from the top, and once the mandate loosens, people quietly return to the way they've always worked. A team can nod through a training session and still be completely unconvinced it's worth changing how they work, and that quiet scepticism is easy to miss until, months later, nobody's actually using it. The answer is to show them, concretely, on a task that used to take them ages, not explain it to them again.

The structural answer is treating AI as a permanent operational discipline rather than a project with an end date. A project that solves one task is, by definition, finished once that task is solved. Folding AI into ways of working, backed by frequent training, in-house and external, is what turns a rollout into a habit rather than a memory.

FAQs

Pick something repetitive, time-consuming and genuinely painful enough that the owning team already agrees it's worth streamlining. A task with a clear before-and-after makes the case for scaling far more convincingly than a technically impressive pilot nobody asked for.

Neither works alone. Sequencing starts with one real task the team already owns, not a training programme. Capability gets built by doing that once, piloting it with the people who'll use it, and only then training and expanding outward.

The strongest results come from pairing someone who owns a live task with someone who understands the tool, and working through that real task together until the first person can run it unsupervised. A generic course run by L&D on its own rarely gets a team that far.

The signal is a solid retest, regardless of how long the pilot has been running: the pilot group's feedback has been implemented, performance has been retested against the original task, and the team that owns the work, rather than the AI vendor, is satisfied it holds up before you touch anything adjacent.

Usually some. A platform on its own rarely closes the gap. The GTM Confidence Index FY27 finds that nearly three in four organisations describe their technology as modern but fragmented. What usually moves teams forward is support that combines strategy, technology and execution, so the tool lands inside a process people actually own.

← Back to All Insights Read the report GTM Confidence Index FY27 Related solution Embed & Activate →
Ready to move from thinking to doing?

The first conversation is always about your commercial challenge.

Tell us what you're working on. We'll tell you honestly whether Magnus can help, and what that would look like.

Start the conversation See growth stories