Running an AI Music Project: Week 2 Progress Report
Clafacio Lobo · Project Manager, Musinique · Humanitarians.ai
We are two weeks into Musinique and the project is moving faster than I expected. Five contributors, four workstreams, and a publishing pipeline that is already producing real output. This is a progress report, honest about what is working, honest about what is not, and specific about where we are headed.
What Musinique Is and Why It Matters
Musinique is an AI music research project operating under Humanitarians.ai. The core question we are trying to answer is this: how can independent artists use AI tools to compete on platforms that were not designed for them?
That question runs through everything we do. The Substack articles, the Ghost Artist tracks, the research papers, the collaboration pitches to platforms like SubmitHub and Artist.tools , all of it is aimed at building a research-backed, systematically documented framework for independent artist strategy in an AI-driven music industry.
The team is small and the work is ambitious. Five contributors working across content creation, research, technical development, and music production simultaneously. My job as project manager is to make sure those four workstreams stay connected to each other , that the research informs the content, the content builds the audience that gives the research credibility, and the technical infrastructure makes the whole system scalable.
Week 2 was the week we found out whether that coordination model actually works under real conditions. Here is what I found.
What the Team Built This Week
The output volume this week was genuinely strong.
Shruti published three Substack articles on music industry analytics — covering why high-follower playlists consistently underdeliver for independent artists, what Spotify’s native dashboard hides from the artists using it, and what streaming numbers actually mean for revenue. She also completed the research collaboration pitch to SubmitHub and Artist.tools, proposing a data-sharing agreement and potential co-authored research paper. The pitch is complete and strategically sound. It is not yet sent,held pending data from Artist.tools , but it is ready.
Ragamalika published two articles on the Ghost Artist concept, produced two YouTube tutorials walking through the Ghost Artist creation process, and completed an original track using the Ghost Artist methodology on Suno. She also began early research into Spotify for Artists as groundwork for the next article in her series.
Nidhi matched that output,three published articles on music discovery and AI music workflows, two YouTube videos, and a produced Ghost Artist track. Nidhi also began scoping the process for setting up Spotify and Twitter accounts for Ghost Artist identities. That workstream is currently blocked, and I will come back to that.
Nixon spent the week on foundational work a deep review of Artist.tools documentation and the start of the first Musinique Substack article on artist-led discovery tools. The article is in outline phase. The documentation review produced useful output that fed directly into the collaboration pitch.
Sakshi worked on the technical infrastructure specifically investigating why Claude embedding is not loading correctly on irreduciblyhuman.xyz. The diagnosis is getting clearer: CSP conflicts and iframe restrictions are the leading suspects. The fix requires sign-off on the site architecture from Professor Nik, which introduces an external dependency on her timeline.
The Coordination Layer - What It Actually Looks Like
I want to be specific about this because it is the part of the project that is least visible from the outside.
Managing a project like this is not administrative work. It is continuous context maintenance. At any given moment this week I was tracking: where each contributor’s primary deliverable stood, which workstreams had external dependencies, which blockers were mine to escalate versus mine to wait on, and where the gaps between workstreams were quietly opening up.
Monday started with four hours of rolling check-ins, not a single meeting, but back-and-forth across the morning that confirmed alignment and surfaced scope questions early. That investment paid off by Wednesday when several of those early clarifications prevented downstream confusion.
The most time-intensive coordination session of the week was the Nidhi blocker. She needs platform access to a tool to proceed with Ghost Artist social media account setup. Tracing that escalation path, figuring out who owns the access decision, documenting what is blocked and why, and flagging it with the right urgency, took the better part of a session. The blocker is not resolved. I will carry it into next week with a named owner and a hard deadline.
The Shruti pitch review was the highest-stakes editorial work of the week. Two revision cycles, focused on making sure the value proposition was framed from the recipient’s perspective and not just ours. The pitch improved significantly across those two cycles and is now ready to send the moment the data dependency clears.
What Is Not Working Yet
I want to be direct about this. A progress report that only counts the wins is not useful.
The upload confirmation process is broken. By end of week, nearly every contributor had artifacts listed as pending upload to the project site. The content was being produced. The articles were going live. But the operational layer, confirming uploads, documenting links, closing the loop, was running behind. I caught this on Thursday going into Friday. A more attentive PM catches it Wednesday. That gap is mine and I am fixing it next week with a standardized checklist and direct tracking.
Nixon’s article needs more than a directional note. I reviewed his outline and told him to narrow the thesis to one clear argument. That is correct feedback. But delivering it as a note in a check-in rather than a working session may not have been sufficient. Finding the argument is the hard part, and I gave direction instead of time. Whether a complete draft arrives next week will tell me whether the feedback landed.
The Nidhi escalation has no confirmed owner. Escalating a blocker without naming who is responsible for resolving it is not a resolution, it is a transfer of anxiety. That needs to be fixed at the start of next week.
Where We Are Headed
Week 3 has six specific things that need to close, not ongoing tasks, not continued coordination, but actual done conditions.
Nidhi gets access and begins account setup. The upload checklist goes out Monday and confirmations are tracked by Thursday. Nixon delivers a complete draft. Both collaboration pitches get confirmed send dates. My own Substack article gets published. And the coordination dashboard gets confirmed live on the project site.
The project is building real momentum. The publishing pipeline is active, the research pipeline is scoped, and the Ghost Artist content series is establishing itself with both written and video assets. The coordination model is working, imperfectly, and with gaps I have named here directly, but working.
More next week.
Clafacio Lobo is the Project Manager for Musinique, an AI music research project at Humanitarians.ai. Follow the project at musinique.net · humanitarians.ai/clafacio-lobo