spot bottlenecks
map real work and find where time, money, or attention gets stuck. not everything needs automation.
ai adoption your company can trust
fifteen years in ai, focused on your real work: we separate hype from reality, coach your teams, and build ai agents and systems that deliver measurable roi, not flashy demos.
about
senior ai experience
ceo johannes gontrum has dedicated his life to ai since 2011. with a master's degree in language technology and published papers, academic rigor helps him validate ai hype and pick the trends that truly move the field forward.
his work as a senior developer shipping products underlined the need for secure, maintainable, and controllable solutions. apart from consulting, he serves as the cto of qallio, helping small businesses accelerate with voice ai.
risk vs opportunity
agents running wild
"what if our agent deletes our database?" the web is full of horror stories of ai experiments gone wrong. ai is powerful: risk that is acceptable for enthusiasts is unthinkable for established companies.
but agents can run safely and effectively, and deliver explosive roi by speeding up company processes. a recent customer reported saving 45h per month by handing a simple process to an agent that outperformed the existing employees.
our methodology
the SAFE ai loop
applying a decade of senior developer experience, we bring production software practices to ai agents. the result is the SAFE ai loop: a repeatable way to build powerful agents that stay controlled.
adapt systems
connect ai to your company's existing software, apis, and workflows so it supports real operations.
fence actions
true security: shield integrations from the ai and enforce clear permissions on what an agent can and cannot do.
evaluate results
monitor quality, cost, execution, and roi after launch. less guesswork, real numbers.
for complex cases, we collaborate with 4front2market to map commercial and operational workflows, quantify friction and revenue impact, and define what is worth building before we build the safe ai system.
build a SAFE ai agentworkshops
practical ai coaching
claude, mcps, codex, token-maxxing, agents, automation loops. is that your day-to-day language? no? good. you have a company to run.
we help your team use ai in the context of their actual work, from practical chatgpt and claude basics to hands-on workflows for writing, analysis, planning, admin, sales, support, and operations.
some sessions work best as team workshops. others work better as weekly 1:1 coaching around one person's role, bottlenecks, and workflows.
ai basics
a practical introduction to chatgpt, claude, prompting, review habits, and where ai fails. use it better without handing over judgment.
claude cowork
hands-on sessions around real emails, documents, plans, reports, research, meeting prep, sales material, communication, and admin.
automation basics
an introduction to skills, connectors, mcps, and simple automation patterns that stay understandable and controlled.
agents and loops
advanced coaching for leaders, operators, and power users building proactive ai support with limits, feedback, and review.
contact
ready to make ai useful?
tell us what you are trying to improve. we will help map the smallest safe first step, whether that is coaching, an agent, or a custom system.