Services: consulting, training and coaching
Built from the practice we run inside the studio.
We run workshops on weekends, on-site consulting on weekdays, and retainers for teams who want a sparring partner after the kickoff. There are three kinds of engagement: agentic CLI training, OpenClaw and Hermes Agents setup, and coaching in vibe coding, harnesses, context and memory to get a team past the demo.
Claude Code, Codex CLI and Gemini CLI
Hands-on training on the three agentic CLIs your engineers will use this year.
What we cover
Install, identity and credentials, project setup, slash commands and skills, MCP servers, hooks and settings, plan and agent modes, sub-agents and worktrees. We also cover what catches teams out in week two: cache misses, runaway tool loops, context drift and prompt injection in retrieved content.
Format
A half-day intro workshop, a two-day full workshop, or a four-week embedded coaching block where we sit with your engineers on real work, in your codebase, on real tickets.
What we leave out
Vendor cheerleading. We use these tools every day and we'll tell you where each is strong, where it's weak and when to use which. You also won't get an "AI will replace your team" deck, which is no help in week two when a tool has just blown up a migration.
OpenClaw and Hermes Agents setup
For teams setting up an internal agent platform instead of buying SaaS.
When this fits
You've decided you need an internal agent platform with an action bus, a memory tier, an eval harness, a refusal policy and an audit trail, and you'd rather operate it yourself than pay per seat for a closed SaaS product.
What we bring
Reference architectures we've shipped, deployment templates, and the routine decisions on auth, secrets, networking, observability and audit, settled before the first crew runs. We've already made the mistakes, so you can skip them.
How the engagement runs
A four-to-eight-week setup, with an optional retainer for the first quarter of operation. Then we hand it back: the platform, the runbook and the on-call are all yours.
What it doesn't include
We don't bundle a managed-service contract on top of someone else's open source. If a vendor relationship makes more sense than self-hosting, we'll tell you, and that's a different conversation.
Vibe coding, harness, context and memory
Four disciplines that decide whether an agent works outside the demo.
AI vibe-coding discipline
When to let the agent run, when to step in, and how to read agent output critically. These daily habits decide whether the team's gains compound, or its mess does.
Eval-harness engineering
Golden, aspirational and adversarial sets. Writing evals that catch regressions as well as confirm happy paths. Costing the harness and budgeting for it as you would a test suite.
Context engineering
What to load, when to compact and when to clear. Cache-aware prompt structure and retrieval that justifies its tokens. This is the work that decides whether your agent is fast or slow, cheap or ruinous.
Agent memory tiers
Working, session, long-term and organisational memory: where each lives, what each costs and how each decays, plus which decisions belong in human review instead of a memory write.
Who joins
The engineering manager as well as the engineers. The discipline doesn't stick without the manager's support, and the new failure modes are managerial as often as technical.
Eval runs by commitIllustrative
✓ Pass× Fail– Not passing yet
Commit 7: merge blocked, fixed in commit 8
Brief us
Tell us about
your team.
We scope workshop bookings, embedded coaching slots and platform setups during the paid diagnostic week. You keep a written scope and a calendar either way.
Brief us