On working with your team, and with AI.
A live online session for a team to build a shared frame for AI, see the architecture under an AI-native team, and look together at where AI might fit in actual work.
At Keftek, we design and build AI Operating Systems for specialised firms, the foundation their work runs through. Along the way, I keep meeting individuals who are excited about AI, but real change only sticks when the whole team moves together.
That’s why we designed this workshop. It’s for the team to come together, get the basics, and have every voice in the room heard. In 2026, AI isn’t just a chatbot, and it isn’t just a tool to implement. It is a new way to build company-level intelligence into how the team operates.
Tools cycle every quarter; the architecture under them compounds. This session works at the architecture level: what an AI harness is, what its parts are, how a team’s judgment and shared context become part of the system, and how to read any new tool against the same frame.
The format brings the team together, because the compounding lives one layer up from the individual. Most AI productivity stories are still about a single role: the engineer who shipped twice as much, the analyst who reads ten reports a day. The shared layer, context, rhythm, and decisions captured over time, is where one person’s progress turns into everyone’s.
By the end of the session, the team shares a baseline for AI, the language for its harness, and two frames for looking at the work together.
Where AI is, and where the work is going.
The first hour gives the team a baseline that holds. The arc starts with the seven problems that started AI in the 1950s, traces the research progress that brought us to today, and looks forward to what work becomes when each person has an agent of their own. The point isn't the history; it's shared language and common ground for everything that comes next.
The architecture under an AI-native team.
The second hour works at the architecture level. A reliable AI system is a harness: context, tools, memory, feedback, and a team that's part of the architecture, not outside it. The hour gives the room a frame to read any harness against, the language to describe its own, and a guided conversation about the team's shared context.
From frames to a concrete read on the team's work.
The final hour brings the architecture down to one workflow. Two frames, the Work Map and the Fit Frame, surface the choices a team has to make: what to automate, where humans stay in, what could become an internal app. The session closes on the rules of the road: limits, risks, governance, and what to write down before adoption goes wider.
Mostly alignment, with a side of training. The team learns the concepts and the two frames, then applies them to one example from the room. The outcome is shared understanding the team can act on, not a strategy deliverable.
Three configurations work well. A whole team in a specialised firm, usually 5 to 12 people who do the work together. A leader plus their champions, one or two leaders bringing 3 to 4 people who'll lead the rollout inside the team. A leader plus a knowledge manager or ops lead, when the work of organising the team's context sits in a specific role.
No. The session is built for non-technical leaders and teams. Technical concepts get explained only as far as needed for the team to make sound decisions.
No live build. The third hour walks through one example using the Work Map and the Fit Frame, with side examples to make the harness tangible. Teams that want a hands-on build go on to coaching or a harness build afterwards.
Bring one or two real workflows the team cares about. We pick one as the shared example for the third hour. If you don't bring anything, we use a generic workflow from your industry.
No. It works best with mixed levels. People already experimenting share their language; people still unsure get a clear frame to work from.
The team can take the frames home and apply them internally, or ask Keftek to scope the next step: a coaching retainer, an ecosystem build, or a fuller AI Operating System.
Share a few details and we’ll get back to you with an agenda tailored to your team and industry. If we’re not the right fit, we’ll say so and point you to whoever is.