The Square Canisters Apollo 13 Could Not Fit, and What They Almost Cost

Ghanaian bilingual speakers should compare four things before choosing an AI companion: whether it handles Twi and Ghanaian English in one natural conversation, how voice interaction feels, when it sends information to the cloud, and what control they have over conversation history. The best choice depends on your own words, your comfort with cloud processing, and whether you can clearly see and change the settings that affect your data.

Start with the way you actually speak

In 1970, Apollo 13 was headed away from the Moon when the spacecraft’s carbon dioxide levels became a serious problem. The crew and mission control faced a practical mismatch: the lunar module had lithium hydroxide canisters, but the command module’s square canisters would not fit its round receptacles. NASA engineers in Houston had to devise an adapter from materials already available to the crew. The outcome was uncertain until the improvised solution worked.

The lesson is plain: a useful system has to work with the real conditions in front of it. It cannot demand that every part of the situation reshape itself first.

That matters when you speak naturally in Twi and Ghanaian English. A family message may begin in Twi, switch to English for an appointment detail, then return to Twi for emphasis. A question about work can include a name, a place, a date, and a phrase that carries more feeling in Twi. If an assistant makes you pick one language before every turn, or requires you to repeat the sentence in a different form, the conversation starts serving the tool.

Nkomo is built for typed and spoken Twi and Ghanaian English conversations, including ordinary code-switching. Put that to the test with a sentence you would actually send: “Mɛkɔ Kumasi tomorrow, but can you remind me what I need to take?” Then correct yourself halfway through. Add the English detail you would naturally use. Listen for whether the reply follows the whole meaning rather than reacting only to one language.

For more on why one-language-at-a-time interaction can create an unnecessary pause, read The Language Picker Nkomo Removed, and What a Pause Can Cost.

Test voice as a conversation, not a demo

ChatGPT Voice, Gemini, Meta AI, and Nkomo may all appear in the same shortlist because they offer AI conversation. A quick demo can make them sound similar. Your own speaking habits reveal more.

Try a short real task. Ask for help wording a message to an auntie, planning what to say in a voice note, or turning a mixed Twi-English thought into a clear list. Speak at your usual pace. Pause. Change your mind. Interrupt the reply if you need to add context.

Nkomo’s hands-free voice mode lets you hold to speak, hear a reply, and interrupt naturally. That interaction is worth evaluating in the moments when your hands are busy, when you are walking, or when you need to correct a detail before an answer runs too far.

Do not treat a polished first response as the whole test. Check whether the assistant keeps up when you restate a name, mix languages, or add “no, I mean…” in the middle of the thought. These are small moments, but they decide whether voice feels useful after the first day.

Make cloud consent a decision you can see

A voice conversation can contain more than a simple request. It may include a health concern, a family disagreement, a school matter, or a draft message you are still thinking through. Before sharing that kind of detail with any AI service, find the data controls and read what each option means.

Nkomo presents explicit cloud-consent choices: never, ask each time, or this session. These controls help you decide when information can reach the cloud instead of leaving that decision vague or buried in a setup screen.

The right setting can change with the conversation. You may be comfortable using cloud processing for a recipe or a rewritten message. You may prefer to be asked before discussing something personal. The important part is knowing which choice applies before you speak, and being able to change it without hunting through menus.

This is also where a side-by-side comparison becomes more honest. Do not assume that similar-looking voice buttons have identical data practices. Open each product’s current privacy settings and policy, then write down what you can control, what is explained clearly, and what you cannot confirm. Product policies and settings change, so verify them on the version you are using.

Check what happens after the conversation ends

Apollo 13 is remembered for the adapter because the materials and constraints were visible. Mission control knew what was available, what could fit, and what needed to happen next. Data controls deserve the same clarity.

With Nkomo, conversation history is kept on the device under your control. You can turn history off, and it is purged immediately. You can also export your data or delete your account with one tap. If something goes wrong, Nkomo shows the error rather than quietly swallowing it.

Those details change the questions to ask before you settle on an assistant. Can you see whether history is on? Can you turn it off when you want? Can you take a copy of your information? Can you delete the account without an unclear process? When voice fails, do you get an error you can act on?

Run that check with a harmless test conversation first. Turn history off, confirm the result, and locate export and deletion controls before the day you need them. The Apollo 13 team could work because the constraint was visible. Your AI companion should make its own boundaries visible too.