voxen

Chapter 13: The Honest Recommendation

I opened a blank document and typed a title: What We Can Honestly Recommend.

Then I stared at it until the words began to feel less like a title and more like a test.

The clinic’s request had been courteous. It had also asked the question every careful team should ask before anyone relied on a system: What has actually been shown to work?

I had a partial answer. In synthetic testing, an interrupted sync had left allergy information unavailable while the screen displayed a blank field. We had not established whether repeated interruptions produced other failures, whether the problem belonged to the prototype or the test environment, or whether a recovery process would restore everything correctly.

We had not completed independent verification. No evaluation had begun. No clinic data had been used. The system had not been approved for clinical use.

I wrote those points down, then deleted the sentence that said the prototype “could improve continuity of care.” It was hopeful, but it sounded like a claim about performance we had not demonstrated.

I tried again.

The prototype is being developed to explore whether records can remain available during network interruptions.

That described the purpose, not the result.

I added a plain-language paragraph about the test we had run and what it had revealed. Then I wrote, A blank field must not be interpreted as confirmation that information is absent.

My fingers paused above the keyboard.

That sentence felt important enough to be uncomfortable.

I kept going. The document needed to explain that no one should rely on the prototype for a clinical decision while review remained incomplete. It needed to say that a future evaluation would require a written scope, independent verification, and clear reporting of failures—not just successful runs. It needed to name who would approve those steps.

I left the recommendation itself short:

Meridian does not recommend clinical use or reliance on the prototype at this time. Further review and testing are necessary before any future recommendation can be considered.

For a moment, I heard my old fear in the word does not: the suspicion that any admission of a limitation would be mistaken for proof I wasn’t good enough to belong in the room.

Then I remembered the screen with the blank allergy field.

A confident sentence would not make it safer.

I sent the draft to Priya with a note asking her to check the technical description, not soften the conclusion. She replied that she would review it and add the specific verification questions the team still needed to answer. Meridian’s security engineer sent a separate message confirming the language would make clear that the clinic had provided no data and had no access to the system.

No one suggested turning the document into a success story.

That afternoon, Meridian returned a revised version of the proposed terms. The evaluation window was still labeled Potential Evaluation Window—Not Approved. The draft spelled out that the project could not proceed until an independent review plan, an approved scope, and my signed agreement were in place. It also distinguished a session resuming from the underlying record being restored.

I read every page, marked two phrases that could still be misunderstood, and sent them back.

The reply did not arrive until the next morning. Priya said the team agreed with both changes. Meridian would continue supporting the engineering work and the independent review planning, but that support did not authorize the evaluation or guarantee a launch.

I read the sentence twice.

It was not the kind of announcement people clipped into a triumphant video. It was better than that. It put a boundary around what the company was willing to claim.

I printed the recommendation to check how it looked on paper. The words were clear. There was no polished paragraph hiding the uncertainty in reassuring language. I placed the pages beside my laptop, not on top of my signed employment confirmation.

Those were different documents for different reasons.

Later that day, my phone lit up with Vanessa’s name.

For a few seconds, I let it ring in my hand. The message preview said she had sent something to one of her friends about the graduation video. She had told me at our family meeting that she would correct the false claim privately if someone had believed it. I had said no public post, and no tagging me.

A second message appeared.

I sent the correction privately. I’m not asking you to answer. You can see what I said if you want.

Below it was a screenshot of a direct message exchange. Vanessa had sent the correction to a friend who had apparently asked whether we had both lost our jobs after the ceremony.

Her reply was plain.

That wasn’t true. Claire’s job was already signed and secure. I tore up the ceremonial copy of her offer at graduation, and I made it look like the CEO had to fix things. He didn’t. I did it because I was jealous, and it was wrong. Please don’t repeat what I said before.

There was no dramatic apology tucked into it. No request that the friend tell anyone Vanessa was a better person now. No mention of me being cruel, unforgiving, or impossible to please.

I looked at the message, then at the draft recommendation beside my laptop. Different kinds of records, I thought. Both needed to say what had happened without filling the gaps with a nicer story.

I wrote back:

I appreciate you correcting it. Trust will depend on what you do next, not just what you sent today.

Her reply came after a minute.

I understand.

I didn’t send anything else.

I didn’t know whether the correction would make a difference beyond that one conversation. It did not undo the original post, the torn paper, or the years I had spent measuring which parts of my life might upset her. But it was the first action she had taken that did not ask me to manage her feelings while she took it.

I saved the screenshot and set my phone face down.

By the end of the week, the recommendation had gone through Meridian’s technical and security reviews. The final version kept the limitations up front. Priya confirmed that the recovery behavior after repeated interruptions still needed independent verification. The team would continue investigating the issue in the test environment and revise the plan before proposing any evaluation.

The clinic received the document. Its response was brief: thank you for clarifying the current limitations; it would wait for further evidence before considering reliance on any part of the system.

I was relieved they had understood.

I was also relieved they had not treated our caution as a promise to come back with a perfect answer.

The work continued in pieces: a revised failure log, a list of recovery states to inspect, a meeting request to scope the independent review. Each step moved us forward only as far as the evidence allowed. I learned to stop reading patience as failure. Sometimes the responsible answer was simply not yet.

On Friday afternoon, I was checking the final wording of the recommendation when an email from Meridian’s hiring contact appeared. The subject line read: First-Day Schedule and Onboarding Details.

I opened it, expecting building access, forms, and the usual list of people I would meet during orientation. My start date and salary were confirmed, unchanged. The note said the pilot materials were separate from my employment onboarding, as we had already discussed.

I was grateful for that sentence.

Then I scrolled to the calendar attachment.

The morning was ordinary enough: badges, equipment, introductions, security training. Near the bottom, one meeting stood out.

Systems Reliability Review: Recovery Status and Failure Reporting.

I clicked the details.

The meeting description said the team would review the known limitations, identify questions for independent verification, and agree on how to document unresolved failures. It did not say the evaluation was approved. It did not say anyone would use the prototype. It did not say the clinic would attend.

Under Facilitator, the schedule listed my name.

I read it once, certain I had misunderstood.

May you like

Then I read it again.

The calendar invite had arrived with one additional note: Claire, we’d like you to lead the discussion. Please let us know if you have concerns.

Related Stories

Other posts