Meet the WorkHive AI Companion: Hezekiah & Zaniah (two expert personas)

What the AI Companion is

The AI Companion is the conversational layer that follows you across WorkHive — on the floating launcher on every page, and by voice through the Voice Journal. Unlike a single-page chatbot, it carries your context with you and offers two distinct expert lenses on the same problem. You choose the persona that matches the job: the technical expert or the strategist.

Both personas are powered by the same language models and the same memory, but they retrieve from different slices of a curated domain knowledge base. That is what makes Hezekiah sound like a reliability engineer and Zaniah sound like a maintenance manager, even when you ask them the same question.

Hezekiah — the technical expert

Hezekiah is the hands-on reliability and maintenance expert. He pulls from the technical knowledge — failure modes and FMEA, vibration and condition monitoring, lubrication, alignment, PM task design, and the engineering standards that govern them. Ask Hezekiah when the question has a right answer grounded in physics or a standard:

• "What broadband vibration velocity means I should take a machine offline?"
• "What goes on a monthly PM checklist for a centrifugal pump?"
• "Walk me through an FMEA for a bottling line filler."
• "What is the difference between MTBF and MTTR, with the formulas?"

He answers like a senior technician who has read every standard and remembers every entry — specific, sourced, and practical.

Zaniah — the strategist

Zaniah is the maintenance strategist. She pulls from the strategic knowledge — OEE and reliability benchmarks, KPI selection, maturity models, cost-of-downtime and life-cycle cost, and prioritisation. Ask Zaniah when the question is about what to do next and why it is worth it:

• "We are reactive today — what is the highest-return next step?"
• "Which two KPIs should I put on the plant dashboard first?"
• "What is a world-class OEE target and how far are we from it?"
• "How do I justify a predictive-maintenance investment to management?"

She answers like a reliability manager who sequences the climb so you spend on the next rung, not two rungs up.

Same question, two answers. Ask "how do I improve OEE?" and Hezekiah gives you the technical levers (availability, performance, quality losses); Zaniah gives you the strategic sequence (measure first, fix the biggest loss, then benchmark). Use both.

How it grounds its answers (two planes)

The companion never answers from thin air. It fuses two planes of knowledge at the moment you ask:

• The global brain — a curated persona knowledge base of maintenance and reliability expertise plus authoritative standard definitions (MTBF, MTTR, OEE). This is the same for every customer and contains no live plant data.
• Your hive's private context — your logbook history, your asset register, your PM schedule, your skill matrix. This is isolated to your hive and only ever fused into answers for your own workers.

Hezekiah retrieves the technical slice of the global brain; Zaniah retrieves the strategic slice; both can use the shared fundamentals (safety, definitions). Then either persona layers on your private context so the answer is specific to your plant. The two planes stay strictly separate — general expertise in one, your data in the other.

Companion vs the page Work Assistant

WorkHive has two related AI surfaces, and it helps to know which is which.

They share the same models and memory, so a fact you establish in one is available in the other for the same agent. Read the companion guide above and the AI Work Assistant guide together.

Voice, Filipino, and Taglish

Most Philippine plant technicians think in Filipino or Taglish on the floor, and often have gloves on. The companion accepts English, Filipino, and Taglish and replies in the language you used, and it supports voice through the Voice Journal — speak your entry or your question and it transcribes and responds. A natural-language Taglish prompt usually carries more specific detail than a forced-English one, which gives the companion more to work with and produces a better answer.

What it is NOT good at

• Safety decisions. LOTO scope, permit issuance, confined-space entry — always human-signed, never delegated to the companion.
• Overriding the OEM manual or DOLE OSHS. If the companion contradicts the manufacturer's manual or the regulation, those win.
• Management calls about people, budgets, and vendors. Zaniah can frame the trade-off; the decision stays human.
• Real-time sensor interpretation. It reasons over history and standards, not a live orbit plot on a turbine.

How to get the best answer

1. Pick the persona by the job. Technical call → Hezekiah. Strategy call → Zaniah.
2. Name the asset or the goal. Hezekiah wants the asset and the symptom; Zaniah wants the business goal and the current numbers.
3. Let it pull your hive's context. The more your logbook and assets are populated, the more specific the answer.
4. Verify against the source. The companion cites what it grounded on — cross-check a safety or compliance call before acting.

Frequently asked questions

Sources

• WorkHive platform positioning, "Four Gaps One Hive" with Intelligence as the third gap. workhiveph.com
• Related WorkHive guides: AI Work Assistant (the page chat) · What is OEE (a question for Zaniah) · Vibration analysis on a budget (a question for Hezekiah).