MR
consulting · rollins.io
  For Engineering Leadership · Onsite or Remote

Bring agentic engineering to your team.

Live working sessions for engineering leadership. One or two days, onsite or remote.

Your team builds something real with agentic tools. The structural pain that's been hiding surfaces. You walk away with a private debrief on team dynamics and which engineers are positioned to lead in this new model.

The shift

Coding stops being the bottleneck. Everything underneath it shows up.

Adopting agentic engineering isn't a tooling decision — it's a leadership decision. When AI takes coding off the critical path, the limits of your planning, infrastructure, trust models, and team composition surface fast. Most orgs are not ready for what they expose.

Trust models break

Senior code review can't keep up with agentic velocity. Trust has to move to tests, multi-concern AI review, and deterministic build gates.

Bottlenecks migrate

Velocity exposes planning, ticketing, and infrastructure as the next limiters. Engineers idle on Jira while DSO chokes.

The deck reshuffles

When coding skill is leveled, broad thinkers and self-starters rise. Some engineers will resist hard enough that reassignment is on the table.

Approach

A working session, not a slide deck.

The first day stands on its own: your team builds, the room talks, the structural pain surfaces, and you walk out with a private read on the team. A second day extends into a guided hackathon and a deep-dive scoped to whatever Day 1 surfaces.

01
The Shock
A one-hour build challenge with a real prize. Engineers ship something concrete with agentic tools. The room sees what's now possible — viscerally, not in theory.
02
Trust in Engineering
A facilitated discussion that reframes how you trust code. From "a senior engineer will catch it in review" to "what infrastructure would let me ship a junior's code unsupervised?" Tests, AI reviews, deterministic build gates.
03
The Pain Snake
A structured exercise that surfaces the structural pain that's been hiding. Group catharsis, plus a prioritized list of engineering priorities that weren't visible to leadership yesterday.
04
Agentic Mechanics
Hands-on patterns: how to structure ideas for the model, how to work with non-determinism, how to run a daily workflow that compounds. Prompt engineering demystified — talk to the model about the prompt.
05
Leadership Debrief
A private session at the end of the day. Observations on team dynamics, individual standouts, friction points, and concrete recommendations on team structure. Followed by a written report.
Day 2 · optional
06
Guided Hackathon
An aggressively-scoped goal with live coaching. Engineers pushed past where they think reasonable ends — that's the point. About half a day.
07
Custom Deep-Dive
Scoped from Day 1 observations. Candidates: fine-tuning workshop on your hardware, infrastructure or DSO acceleration, build-pipeline design, restructuring of dev environments, or a team-specific challenge surfaced in the Pain Snake.
Outcomes

What changes after the session.

Immediate · in-session
  • Engineers ship a working app in under an hour with agentic tools.
  • Leadership gets a clear mental model of the new trust framework.
  • A prioritized list of structural pain — items that have been hiding.
Short-term · 30 days
  • 5–10× velocity on infrastructure, containerization, and pipeline tasks.
  • Testing and automated review begin to replace manual code review as the primary trust mechanism.
  • First fine-tuned models deployed where on-prem GPU capacity allows.
Medium-term · 30–90 days
  • Team right-sizing becomes apparent. Output of 30 engineers is achievable with 5–6 in this model.
  • Planning shifts from tickets to feature-level assignment. Engineers get discretion.
  • Downstream bottlenecks (DSO, infra) get addressed before they choke velocity.
Strategic
  • An engineering culture that selects for broad thinkers and self-starters.
  • Soft skills, design sense, and domain knowledge become force multipliers.
  • Sovereign or near-sovereign AI capability where compute and constraints allow it.
Format

How it runs.

LengthOne day, or one + one
ModeOnsite (preferred) or remote
Team sizeUp to ~30 engineers · larger orgs split into cohorts
Pre-sessionTeam dossier · model access resolved · harness pre-installed
DeliverableWritten leadership debrief with team-structure recommendations
ConfidentialityMutual NDA on request
Engagement

How we work together.

Engagements are scoped per team. Pricing depends on team size, mode (onsite/remote), travel, and Day 2 scope. Reach out and I'll come back with a fit-check and a quote within a few days.

Looking for ongoing advisory, embedded engineering support, or a multi-team rollout? Let's scope it.

Get in touch

Tell me about your team.

Drop me a line with team size, current state, and what's pushing you toward this. I'll come back within a few days with a fit-check and a proposed shape for the engagement.

FAQ

Questions, asked & answered.

Is this for engineering leadership or for individual engineers?
Both. The session is run for the engineering team, but the leadership debrief and structural recommendations are for you. The strongest engagements have an engaged audience and an engaged sponsor.
Does my team need to be using AI tools already?
No. Some teams arrive having barely tried it; some are already deep. The shock-and-build exercise calibrates everyone to the same starting line within an hour.
Can you run this remotely?
Yes. Onsite is preferred — a chunk of the value comes from observing team dynamics during unstructured time, and that's harder over Zoom. Remote works well for distributed teams and is often the right call for a Day 1 pilot before committing to onsite.
What if our environment restricts model access?
That's a feature, not a bug. Defense, healthcare, and other constrained environments can stand up self-hosted open-source models on existing GPU infrastructure. The constraint usually points toward a sovereign capability that most peers are years away from.
How do you handle confidentiality and NDAs?
Mutual NDA on request before any pre-work. The leadership debrief and written report are private to you and your designated leadership group.
What's the team-size sweet spot?
Up to about 30 engineers in a single session. Larger orgs typically split into two cohorts run on consecutive days, sharing the leadership debrief.
How far out do I need to book?
Usually 4–8 weeks for onsite, less for remote. Faster turnaround is possible for the right engagement — ask.
What if some of our engineers will resist?
Some will. Agentic engineering tells people that what they believe is "good engineering" is changing, and that's a hit to identity. The leadership debrief includes candid observations on which engineers are positioned to lead in this new model and where friction is likely to emerge.