How a Composer Used AI to Reach Flow More Reliably: A Case Study

Nadia Osei had a practice room reserved every weekday morning from 7:30 to 10:00 a.m. She had been a composer and session musician for eleven years. She had discipline. She had structure. What she didn’t have was a reliable way to convert the time she showed up for into sessions where she actually created anything worth keeping.

Most mornings, she spent the first 45 minutes in a low-grade state of activity that produced nothing: adjusting MIDI settings, reading about technique, generating ideas in a chat interface, noodling through motifs she’d tried before. By the time she reached something that felt like real work, an hour was gone and her attention had started to fragment.

She called this her “warming up forever” problem. She was always preparing to work and rarely actually working.

This is a case study about what she changed, what happened, and what other creative professionals can take from it.

The Baseline: What the Sessions Actually Looked Like

Before examining the intervention, it’s worth understanding Nadia’s starting conditions precisely.

She composed primarily for film and television — instrumental work, episodic, with specific emotional and technical briefs. The work required deep engagement with harmonic and structural problems that had no single right answer. It was exactly the kind of task that benefits most from flow: complex, subjective, requiring her own cognitive architecture rather than reproducible techniques.

She had been using an AI assistant for about eight months. Her typical use pattern: she would open the assistant at the start of a session and use it throughout to explore ideas, get feedback on motifs, brainstorm structural approaches, and work through creative blocks. She described the AI as “like having a collaborator in the room who always has a response.”

On the surface, this sounds useful. The problem was twofold.

First, the AI’s constant availability had eliminated her tolerance for the discomfort that precedes creative breakthrough. Any time she hit a difficult moment — a passage that wouldn’t resolve, a harmonic choice that felt wrong — she would prompt the AI rather than sitting with the problem. The difficulty that would have produced insight was consistently bypassed.

Second, she never reached the state she described as “the good zone” — the condition where she lost track of time, her inner critic quieted, and the work seemed to generate itself. She could identify this state from her years before AI tools. She knew what it felt like. But it had become increasingly rare.

She framed the question clearly when she started experimenting: “I want AI to help me get there, not to replace being there.”

Version 1: Restricting AI to Research Tasks Only

Nadia’s first experiment was to limit AI to factual and research tasks — looking up instrumentation ranges, checking historical precedents, clarifying technical questions — and to exclude it from creative generation and feedback during sessions.

This was an improvement. Sessions with clear research tasks were more focused. But the intervention was too narrow to change her overall session quality. The real problem was not what she was using AI for during sessions — it was that she was using it at all during the critical first 20 minutes when flow entry would otherwise have occurred.

The sessions still started without a clear single objective. She would arrive at the piano, open her current project, feel the blank-slate difficulty, and start doing something — anything — rather than committing to a specific creative problem.

After two weeks, she recognized the pattern: her flow problem was not mid-session AI use. It was session-start ambiguity.

The Redesign: Flow Runway in a Creative Context

Nadia’s second version implemented a more complete structural change, adapted to her specific work.

Pre-session (15 minutes, AI active):

She shifted from arriving at her studio and opening the project to spending 15 minutes at her desk before going to the piano. The pre-session work had three components:

Task definition. She used AI to convert her current project brief into a single session-sized task. For compositional work, this meant moving from “work on episode 4 score” to something like: “Compose the 40-second underscore for the flashback scene in act two — sparse, unresolved, built around the motif from the opening.”

Prompt she used regularly:

I'm scoring [scene description] for [project brief].
My current emotional target for this cue is [what she was trying to achieve emotionally].
Define a single compositional task I can complete or meaningfully advance in 90 minutes, 
specific enough that I'll know when the session was productive.

Blocker identification. Before leaving her desk, she asked what might interrupt her mid-session. Common blockers in her work: uncertainty about the scene’s timing, questions about the instrumentation brief, unresolved choices from the previous session. She resolved what she could in 5 minutes and wrote down explicit deferrals for the rest.

Challenge calibration. This was the piece she had never done before, and she found it transformative. Some mornings the task she’d defined felt overwhelming; others it felt too familiar. She had a short prompt for each direction, and the act of consciously assessing where on that spectrum she was — and adjusting accordingly — changed how she approached the piano.

During-session (90 minutes, AI off):

She set a physical timer. Phone in another room. AI closed. She allowed herself to sit with difficulty without reaching for anything.

Her first few sessions without AI available were, by her account, uncomfortable in a way that surprised her. “I realized how conditioned I’d become to having an escape,” she said. “Anytime the work was hard, I’d been outsourcing the hardness.”

By the third session, the discomfort at the beginning was still present but recognizable. She had begun to trust that it would pass. By the seventh session, she recorded what she identified as the first genuine flow state she’d reached in months — a 40-minute stretch where she was, in her words, “just inside the music.”

Post-session (10 minutes, AI active):

After each session, she spent 10 minutes at her desk before doing anything else. She captured what she’d produced, noted what felt promising, and answered two questions via AI prompt:

I spent [time] working on [task]. Here's what I produced: [description].
What should the next session build on?
What conditions were different about today compared to sessions where I'm stuck?

Over five weeks, the post-session log became her most valuable resource. She could see the pattern clearly: her best sessions happened in the first half of the morning, after at least 8 hours of sleep, when the task was pitched slightly above what felt comfortable, and when she had resolved all technical questions about the scene before starting.

What Changed: Stable State

By week six, Nadia’s session structure had stabilized. The pre-session ritual took 12–15 minutes and she ran it automatically. The during-session period averaged 85 minutes before concentration broke naturally. She reached what she identified as genuine flow — the “good zone” — in roughly 60% of sessions, compared to perhaps 5–10% before the intervention.

The quality shift was visible in her output. She described two specific changes:

First, the compositional ideas that emerged from flow sessions were qualitatively different — more original, less derivative of her own previous work. She attributed this to the transient hypofrontality effect: the inner critic that would normally filter out unusual harmonic choices was quieter, allowing ideas through that she would previously have self-rejected before they materialized.

Second, the debrief log had made her self-aware in a way that transferred across projects. She now had a reasonably accurate model of her own cognitive conditions — and she used it to schedule her most demanding creative work during her highest-reliability flow window rather than distributing it across the day.

The AI, notably, had not become less useful. It had become more useful, more intentionally. The pre-session work was more focused and more consistently valuable. The post-session debrief had become a planning document she actually used.

She summarized the shift simply: “The AI is more helpful now because I know what I’m asking it to do.”

What This Case Study Suggests for Other Creative Professionals

Three observations that transfer broadly:

Session-start clarity matters more for creative work, not less. The conventional wisdom is that creative work requires open, exploratory conditions at the start — not tight task definitions. Nadia’s experience, consistent with the flow research, suggests the opposite. Exploration belongs to the pre-session phase. During the session, you need a clear creative problem to engage with. Exploration with no defined target produces drifting, not flow.

The debrief log compounds. Five sessions of post-session notes tells you almost nothing. Thirty sessions tells you your personal flow profile in enough detail to be actionable. The 10-minute post-session debrief is the single highest-leverage habit change in this framework for long-term practitioners.

AI as preparation support differs from AI as in-session collaborator. Nadia did not lose a collaborator — she gained a clearer preparation process and a more reliable execution state. The AI became more useful precisely because its role was better defined.

For composers, writers, designers, researchers, and engineers who are doing creative work with real AI tools available, Beyond Time offers session structure that’s built around this kind of preparation-execution separation — planning happens outside the block, not inside it.

The Lesson Behind the Case Study

The “warming up forever” problem Nadia described is not a creative problem. It is a structural one. The session begins without a clear target, which means the creative instinct has nothing to aim at, which means the early session produces low-quality activity that prevents flow from forming.

The fix is not a creativity technique. It is a session design decision: define the problem before you walk in the door.

Define your next creative session’s single task before you open your project, and track whether the session reaches absorption earlier than usual.

Related:

• The Complete Guide to Flow State and AI Tools
• The Flow Runway Framework
• How to Enter Flow with AI Tools
• The Science of Flow State

Tags: flow state, case study, creative work, AI tools, composer productivity