Follow the workarounds
Watch where people repeat steps, invent shortcuts, wait, give up, or ask for help. A workaround often points to a need the current process misses.
What do people keep doing despite the inconvenience?
01 / Observe
Look for the space between what is and what could be.
An opportunity is an unmet need, a recurring friction, or a new possibility. Start with evidence from real life before deciding what the solution should be.
Notice what has shifted
New tools, constraints, expectations, and habits make previously impossible or unimportant ideas newly useful.
What can people do now that they could not do a year ago?
Change the viewpoint
Ask who benefits from the current arrangement and who struggles with it. Look at the first use, the handoff, and the edge cases.
Whose experience have we not examined?
Capture, then frame
Write observations as needs or tensions: “New team members cannot tell which decision is current,” rather than “We need a new dashboard.” The first leaves room for better solutions.
02 / Decide
Choose a goal that gives creativity a direction.
Not every opportunity deserves immediate attention. Select one by weighing its importance, the evidence behind it, and your ability to make progress.
A / Value
Does it matter?
Who would benefit, and how much would their experience improve?
B / Evidence
Is it real?
What have you observed? What assumption still needs testing?
C / Agency
Can you act?
What can you change with the time, access, and resources available?
D / Learning
Will you learn?
Can a small experiment reveal whether this is worth a larger effort?
Frame the goal
Help [specific people] achieve [valuable outcome] in [context], while respecting [key constraint].
03 / Create
Generate options for the goal you chose.
A clear goal makes divergent thinking more useful. Produce genuinely different approaches first; judge them against the goal afterward.
Ask “How might we…?”
Turn a fixed problem into an open question. Make it narrow enough to guide ideas and wide enough to allow different answers.
Try: How might we make the next step obvious without adding more meetings?Vary the mechanism
Generate options through different means: remove a step, automate one, change the timing, shift responsibility, or make information visible.
Try: Produce five ideas that work in five different ways.Use a useful constraint
Temporarily limit time, money, technology, or complexity. Constraints can uncover simple ideas that an unrestricted brainstorm overlooks.
Try: What could work with no new software?Make the idea falsifiable
Choose promising ideas by their fit with the goal, then find the smallest test that would show whether each one helps.
Try: What would you need to see to change your mind?A worked example
From a vague complaint to a testable idea.
Imagine a team that keeps losing track of decisions made in meetings. “We need better meetings” is too broad to direct creative work.
Spot the pattern
People repeatedly ask which decision is final and where its rationale lives. New colleagues struggle most.
Name the outcome
Help team members find the current decision and its owner in under two minutes, without adding another meeting.
Try a small change
Add a one paragraph decision record to the existing project page. Ask three colleagues to find a decision a week later.
Put it to work / 15 minutes
Take one real challenge through the sequence.
Use a notebook or work with a partner. Keep the observation, goal, and ideas separate until you have explored each one.
- What recurring friction or overlooked possibility have you observed? Describe one concrete instance.
- Who is affected, and what would a meaningful improvement look like for them?
- Which opportunity scores best on value, evidence, agency, and learning?
- Write one “How might we…?” question tied to that goal.
- Generate at least six different approaches. Pick one small test and decide what evidence would count.
Service design / Security & privacy
Keep session context with your calling agent.
The planned Directed Ideation MCP provides stage guidance without receiving or storing your challenge, conversation, goals, or ideas. The calling agent manages that context and its own provider's policies determine how it is handled.
01 / Access
Authenticated, scoped access
Require secure sign-in and authorize account and team operations against the authenticated user. Isolate account and team records and test those boundaries.
02 / Protection
Limit exposure
Use encrypted connections and encrypted storage for the limited account and team data the service needs. Restrict and audit staff access, and keep session content out of service requests, logs, and analytics.
03 / Context boundary
Session data stays with the agent
The service does not store session history, stage progress, or session content. It cannot control the calling agent's storage, model use, retention, or sharing; check that provider's privacy controls.
04 / Evidence
Show what is verifiable
Publish data-flow and security information, independent security-review summaries, and their dates, scope, findings, and limitations. Provide a way to report vulnerabilities.
Planned controls, not a claim of current certification
The initial service is planned as an invite-only US pilot, with US hosting and processing, clear US-only eligibility, and reasonable signup restrictions. Legal review is required before inviting pilot users; access will not be marketed or intentionally offered in other jurisdictions. These restrictions reduce risk but cannot guarantee that foreign laws will not apply. Session context is managed by the calling agent and its provider, not this MCP service; this boundary does not guarantee end-to-end privacy. The hosted MCP and these protections are still being developed. Do not submit sensitive or confidential information until the service is available and its current privacy and security disclosures are published. Independent reviews and assurance evidence will be identified by scope and date; no single control or certification eliminates risk.