Forge Flows · active
Forge Flows
A part-time company context for turning successful AI work into shared organizational capability.
Co-Founder
DeepSOM → systems architecture
Controlled change, carried through a complex system.
DeepSOM explored a compact lattice whose local edits could steer the texture of a much richer 3D object. I keep returning to that shape in systems work: find the smallest explicit structure that governs behavior, change it deliberately, then verify where the effect travels.
At Forge Flows, the analogous structure is a shared capability—context, workflow, evaluation, and delivery bound together. A revision should reach its known consumers without becoming an uncontrolled rewrite of everything around it.
See the earlier lattice ideaIllustrative model: one reviewed capability change reaches explicit consumers; unrelated workflows remain unchanged.
The method
- ObserveSee how consequential work actually happens, within explicit boundaries.
- DistillTurn a useful pattern into a clear, testable capability.
- DistributePlace it where the relevant people and agents already work.
- ImproveUse evidence to revise it with provenance, review, and rollback.
Problem and context
AI leverage accumulates person by person. A useful method can remain trapped in a private conversation, a one-off prompt, or an individual habit—even when the whole organization would benefit from learning it once.
My role
Part-time
I co-founded Forge Flows and lead product discovery and technical direction across systems architecture, workflow infrastructure, evaluation, and capability distribution.
Constraints
The personal site may describe only public work. Private repositories, partner work, telemetry, contracts, customer data, and internal roadmaps remain outside this case study.
What I built
I work from one consequential workflow outward: find the real constraint, make the governing context and process explicit, test the capability, place it into the working environment, and build the feedback loop that can improve it safely.
Key decisions
The company story is larger than any single tool or acquisition experiment. Skills, evaluation, cross-surface delivery, and auto-improvement are machinery inside one coherent learning system—not a product catalog.
Evidence
The public standard is controlled change: behavior tested, affected consumers identified, provenance preserved, and rollback available. The lattice above is illustrative; it explains the design principle without implying a literal model of an entire organization.
Results
Development teams are the first proving ground because their work produces dense feedback, inspectable artifacts, and clear failure signals. They are the wedge, not the boundary of the company thesis.
Limitations and status
Forge Flows is active and early. This page is a bounded account of my role and public methods, not a substitute for the company site and not a claim about private customer outcomes.