Too Good To Be True?

Anyone who has been around software development for a while has confronted the insistence by stakeholders that we conjure up a silver bullet to make all their dreams come true, such as unlimited features on a minimal budget, along with algorithms that make costs and effort magically disappear.

(OK, I am exaggerating. Slightly.)

got AI?

Right now, business stakeholders are very excited about the prospect of AI (large language models, or LLMs) being the ultimate silver bullet. This is exciting and promising, but allow me to share some reality-infused perspective:

• AI will be a game-changer.
• Eventually.
• After more false starts.
• And after we iterate a few more times to get it right.

The ultimate silver bullet it is not. (Yet).

Don’t get me wrong. I use ChatGPT (and its cousins) every day. These tools are a game-changer for accelerating personal productivity in many areas. ChatGPT has replaced Google search for me 90 percent of the time, and I’m not alone. These AI tools can quickly summarize documents and spreadsheets, and they are approaching middle school levels of sophistication in their assembly of words that logically (statistically) go together.

• However, “AI can replace good authors” is not a claim we can make any time soon.

Like many software professionals, I also use AI tools like ChatGPT daily to vastly accelerate software coding mechanics.

• “Mechanics” is the key concept here. Going through the motions is not programming (or writing).

Yes, these AI tools can produce software code in most mainstream programming languages, and they do that very quickly (with increasing accuracy as the models become better tuned), but they cannot yet produce production-ready software.

What AI tools can do is make competent software developers produce production-ready software faster. Think of the current (and likely the next 20 or so) iterations of AI tools as cheap programming interns, fresh out of their model-training schools and eager to help. Like human interns, they know the basics well, but they lack the big-picture context to be effective contributors without expert guidance. This doesn’t mean they don’t add value; rather, they are powerful tools that need to be in skilled hands to be effective.

AI cannot replace your programmers. But by using those tools, they should be measurably faster–maybe not on Day 1, but quickly.

Silver Bullets: Been There. Endured That.

A close cousin (chicken vs. egg?) of the perpetual stakeholder desire for silver bullets is the pitch by software vendors claiming to offer silver bullets–specifically to software buyers, practitioners and their business leaders.

Just buy this (NOW WITH AI!!), and your projects will be implemented flawlessly. (And purple unicorns will graze on your lawn under vibrant rainbows.)

The appeal is understandable–almost irresistable.

We all want software to be easier to create and maintain. Who doesn’t want a silver bullet to make software, which is very difficult to get right under the best of circumstances, much faster/easier/cheaper?

Want vs. Belief

We all want magic solutions, but we also hear many inflated claims about breakthroughs. We all have tried a few, and we have been burned. This is why we don’t believe most silver bullet claims of process improvement, implementation acceleration or better bread slices.

“Show me!” is the pragmatic response to such claims, but to get to that response, we have to believe that the claims could be true.

What do we do when it just sounds too good to be true?

Obviously, we dismiss claims that sound too good to be true. That is a perfectly logical reaction.

But What if it IS true?

This is the dilemma we face with Touchstone, our healthcare interoperability validation ecosystem that objectively measures the conformance of interoperability implementations to interoperability specifications.

If that description seems too nerdy, let’s rephrase it as:

AEGIS, with our Touchstone platform, helps you get interoperability right.

Does that sound too good to be true? Let’s translate this into business value language.

Touchstone for Business People

Quick Background

FHIR® is shorthand for HL7® Fast Healthcare Interoperability Resources, the latest data exchange format standards being adopted by the healthcare industry.

When data being sent from (for example) a provider’s system to a payer’s system is structured and populated as expected by industry standards like FHIR, that vastly increases the likelihood (in this example) of the claim sent by the provider being paid promptly by the payer.

• Sending data that matches the expectations of the other party is the key to interoperability.

Touchstone objectively measures how well FHIR implementations conforms to those FHIR specifications. The more it conforms, the more interoperable messages are between implementations with high levels of FHIR conformance.

Testing Early & Often

When developing software, validating that it meets expectations (conformance to requirements) gives implementation teams and their stakeholders objective evidence about implementation health. The earlier defects are found, the easier–and less expensive–they are to fix.

When Touchstone is made part the software development lifecycle (SDLC) during FHIR implementations, project stakeholders (including the developers) see immediate–and objective–data about conformance to FHIR requirements. This empowers the whole team with a shared language for making course corrections when necessary and celebrating measurable project progress along the way.

You may have heard about Test-driven Development (TDD). This is the practice of individual developers testing their code (using automated tests) prior to merging their work with the code of the rest of the team. Some TDD practitioners actually write the test before writing the code; they then write the code to make that test pass.

Touchstone provides this TDD-like continuous feedback–but at system interoperability level.

Touchstone is TDD for FHIR Implementations

Business Value

To summarize, incorporating Touchstone into FHIR integration projects makes them:

• More transparent via continuuous progress metrics
• More interoperable
• Go faster
• Cost less