Artifacts have always been a part of software delivery; I mean software itself is an artifact. But, with the ever acceleration of AI into our workflows, it is becoming clear that artifacts must take absolute precedence.
My journey to seeing the importance of artifacts has expanded throughout my career:
- Software is an artifact, duh!
- Software tickets are artifacts.
- Git commits are artifacts! Yeah!
- Automated test results are artifacts.
- Documentation is an artifact.
- GitLab merge requests are artifacts!
- PROMPTs are artifacts!
- Everything is an artifact!
I've gotten to a point in my career where I'm not certain I've done any meaningful work if there isn't an artifact left behind showing that work. This conclusion has led to some interesting results of late.
Tokenomics
token + economics = token·omics
For instance, I've been receiving pressure to justify the use of AI tokens in my workflow. At first, I found this daunting; I mean, wouldn't I want the best model at all times? However, I see the point that we should be good stewards of the resources we spend for the business.
My Solution: Build a
tokenomicsartifact. Include:
- The model I chose
- The reason I chose that model
- The amount of tokens (credits) I have remaining
- The cost of that model
- The amount spent on the model
- The time spent on the task
- The level of engagement (1-10) that I had with the model
Change Log
My next AHA came with merging that artifact with the change log! Our stakeholders want to know what changed and what model was used to produce that change. It's crazy to me that I can now just drop a link to a change log to stakeholders and that artifact takes on a life of its own. My software changes and its tokenomics are being actively discussed in meetings I had no knowledge about.
Side Note: I cannot overly stress the importance of setting up the automation to publish your artifacts in your pipelines. The pipelines that deploy the versioned software should also publish the change log.
It might sound daunting at first, but keeping tabs on your changes with tokenomics has been very satisifying. I mean, I leave work everyday with the confidence that my work was solid because of it!