VALAn AI mentor for project managers
60 pain pointseach routed to the framework that addresses it.
- 10Frameworks, each a module with its own completion rules
- 7Domain knowledge bases behind every diagnosis
- 15+Decision nodes doing the routing
- Every sessionPersists, so nobody re explains their situation
The challenge.
The methodology that made the practice special was the thing stopping it from growing.
Ten project management frameworks, refined over years, held as knowledge across a handful of senior facilitators. Sixty catalogued pain points, where matching a client's situation to the right framework depended on who happened to be in the room.
Deep knowledge across seven domains sat in documents nobody could navigate in the middle of a conversation.
New facilitators took months to learn which framework to deploy for which problem, and most never fully got there. Two facilitators sent the same problem to different frameworks. Follow up advice varied with experience. And every engagement started from scratch, because nothing remembered the last one.
Underneath all of it, capacity was capped at facilitator hours. Growing meant hiring and training, slowly, expensively, with quality risk attached to every hire.
If that shape looks familiar, the consultation is where we go and find yours.
Start with the free consultationWhat we built.
VAL is the system that runs the method.
A project manager describes a problem in plain language. VAL identifies which of the 60 catalogued pain points they are actually facing, selects the framework that addresses it, and runs the full facilitated diagnostic from beginning to end.
It opens the way a good facilitator opens, with questions rather than answers. No intake form, no dropdowns, nothing to schedule.
It is built on ten established project management frameworks, each one encoded as its own module with its own diagnostic logic, its own question sequences and its own completion criteria, so a diagnostic cannot end half finished. A routing layer of more than fifteen interconnected decision nodes chooses between them. Between them they cover planning ambiguity, scope drift, root cause analysis, unrealistic timelines, decision paralysis, ownership gaps and reactive risk management.
After the diagnosis it does not hand over a checklist. It classifies the situation, draws on seven domain knowledge bases, and returns concrete decision options, a seven day action plan, a milestone and a reflection question, written in the user's own language and grounded in their own context.
Every conversation persists. People pick up where they left off, and the system remembers what was discussed, which frameworks were used and what was advised, with a full history across the client base and an analytics layer over the patterns in it.
And it answers at 11pm before a difficult stakeholder meeting, and on a ten minute lunch break, at the same quality both times.
Underneath: a three layer state machine with more than fifteen graph nodes, ten specialised sub graphs, deterministic routing across all sixty mapped scenarios, and conversation state that survives every turn. Python and LangGraph for the engine, Claude Sonnet for generation and Haiku for routing, Supabase for data and auth.
The results.
The method stops being a person and becomes infrastructure.
Ten frameworks that lived in people's heads run as reproducible modules. Seven knowledge bases deliver the same depth every time. Adding a framework means building one more module, not retraining anyone.
The same problem routes to the same framework every time, regardless of who is asking or when they ask. Every framework enforces its own completion criteria, so nothing ends half diagnosed. Context carries across turns, and across sessions.
One voice, across every interaction, with no quality variance at scale.
And the ceiling that used to be facilitator hours is not there any more. The frameworks reach a project manager at 11pm, which is exactly when a project manager needs them.
Tell us what yours looks like.
Bring us the part of your business that runs on someone remembering. We will tell you what it would take to build it out.