Demand Owner Flow
End-to-end: from a new demand idea to a scoped, resourced initiative.
The demand owner is the person with an idea: a product initiative, a capability gap, a customer problem. The original Trmeric onboarding dropped them onto a blank canvas with a chat interface and no context. The insight that changed everything: show someone what good looks like before asking them to create. The result, eventually, was Trmeric Missions: a system that compresses an 8-week enterprise project initiation process into a 10-minute conversation.
The actual conversation, recreated
Click through it. This is the real Tango intake flow: three quick choices, an enrichment pass, and a populated demand canvas.
Initial request and clarification hell. Incomplete form submissions, missing context, email ping-pong with 4–7 exchange cycles, 2–3 days between each response.
Solution architecture and estimation. Manual searches across 4–6 systems for similar projects, resource history, budget constraints, strategic alignment data.
Resource hunting and allocation. Chasing resource managers, checking availability across teams, negotiating allocation percentages, resolving conflicts.
Approval chase and project setup. Waiting for stakeholder attention, answering clarifying questions, re-sending with updates, manual project canvas creation.
Natural conversation. Tango conducts intelligent intake. No forms. Adaptive questioning based on role and context.
AI enrichment. Automatic context gathering from organisational data: similar projects, budget history, strategic alignment.
Smart resource matching. Context-aware suggestions filtered by skills, availability, and role requirements. Not bulk lists.
Execution-ready. Fully formed project canvas. Team assigned. Milestones defined. Ready for planning, not setup.
Across all three personas, the emotional arc told the story: enthusiasm in week one curdled into confusion by week three, then frustration, then resignation by week eight. That arc, not the calendar, was the real design brief.
Three roles, three completely different mental models. Each needed a different entry point into the same system.
Two paths to execution
Inherits roles and costs from the demand but allows a fresh start at execution.
Bypasses estimation entirely for work that doesn't need a governance gate.
One click, fully formed
Click convert. Every field an approved demand already has gets inherited; everything else, Tango generates.
01 · Replace 37 fields with 5 smart questions
Generic forms optimise for data collection, not human understanding. When questions adapt to role and expertise, users provide richer context in fewer steps, and never abandon mid-way.
02 · Show the AI's work, no black box
Black boxes destroy trust, especially with AI. Showing each enrichment step as it runs turns 30 seconds of wait into 30 seconds of confidence-building.
03 · Let users start where they think
Rigid workflows create workarounds. Resource managers often spot available talent before demands exist. Forcing a sequence ignores how real allocation decisions happen: opportunistically, from either side.
04 · One click converts approval to execution
Approved demands already contain all the data needed to start. Re-entering it in a separate system wastes hours and introduces errors. The data should follow the decision, automatically.
What Tango discovers, on one real project
Run it. Every step is shown, nothing inferred silently. This is the actual sequence: source scan, then insight reveal.
No forced sequences
Team-first, scope-first, or parallel: the system never locks a step because a previous one isn't complete.
Show your work
AI processing is never invisible. Each step surfaces in real time: scan, match, score, output, always visible.
Smart defaults, easy overrides
AI suggests a 12-week estimate; the user can set 8. The system assists, it never dictates.
Graceful degradation
Draft state is always saveable. No validation gates block progress, supporting the messy, non-linear reality of planning.
Conversation ends, work begins
Tango outputs directly to a structured project canvas. A 73% reduction in post-conversation data entry, zero manual re-entry needed.
"I can finally build my team and define work in parallel. The system doesn't force me to finish one before starting the other. That alone saved us two weeks."
"The AI enrichment is like having a research analyst who never sleeps. It finds connections and context I would have missed entirely."
"We went from reactive firefighting to proactive portfolio optimisation. Conflict visibility changed everything."
Learnings that apply beyond this project
Show the AI's work
Users distrust opaque systems. Expose the reasoning, expose what changed, let people override it. Transparency is the foundation of trust, not an optional feature.
Avoid one-size-fits-all
Enterprise users carry different mental models. Engineering leadership thinks in sprints and resources; a division manager thinks in budgets and timelines. Design for the mental model, not the data schema.
Governance is not friction
Embed compliance into the flow so it happens automatically, not as a blocking gate. Speed and governance coexist when they're designed together, not traded off.
Conversation beats configuration
Every dropdown, checkbox, and form field is cognitive load. Natural conversation captures the same data with less friction. The best interface is the one users already know how to use: language.
Replacing forms with natural conversation was a risk that paid off. Users didn't need training, they just talked.
Showing users exactly what the AI did, and letting them override it, built trust. The 'show your work' approach turned skeptics into advocates.
Building the resource system to learn from organisational patterns meant recommendations got smarter with every project created.
I waited too long to bring stakeholders into the design process. Earlier alignment would have prevented two rounds of rework on the governance model.
Testing in larger batches meant some usability issues compounded. Smaller, more frequent sessions would have caught issues earlier.
As a solo designer, I under-invested in documenting design decisions. This made handoff harder and slowed onboarding of a second designer later.