AI Adoption Moves at the Speed of Trust: How Nectar Builds

I talk with energy managers, controllers, and sustainability leads every week. Most of them have been told that AI will change their work. Very few doubt what the technology can do. What they doubt is whether they can rely on it when their name is on the budget, the payment, or the disclosure.

That doubt is reasonable, and it shapes everything we build. Our biggest learning so far: AI adoption moves at the speed of trust, and trust comes from human relationships. You come to rely on a colleague because they show up, explain their work, and get things right again and again. Software earns trust the same way, and loses it the same way too.

Here's how we think about trust at Nectar, what goes into it, and how it shapes the product, the service around it, and the way we work as a team.

Trust is the bottleneck, not capability

AI is capable of a lot more than most people think. It can read a utility bill, find an unusual charge, and draft an explanation in seconds. The gap isn't capability. The gap is whether a team can count on the output enough to act on it.

You can't develop conviction if you're operating in the blind. An energy manager who can't see why a number is what it is won't act on it. They'll check it by hand, and the AI will have saved no time at all. A controller who doesn't trust an anomaly alert will pay the bill anyway. A sustainability lead who can't trace a figure back to its source won't put it in a disclosure.

So we treat AI adoption as a human-computer interface problem as much as a model problem. The question isn't only "is the answer right?" It's also "can a person see that it's right, quickly, in the place where they already work?"

What goes into trust

We think about trust as the sum of six things. None of them are new. All of them are easy to underinvest in when the technology itself is exciting.

1 Ease of use

Ease of use influences trust more than most software teams admit. Software that's hard to use feels hard to rely on. We hold the product to a high bar: bills arrive without anyone chasing them, accounts connect through a simple invitation instead of a shared password, and answers appear where the work happens. The experience includes the service around the product. Our team handles onboarding, account setup, and the unusual cases, so customers aren't left alone with a tool.

2 Intelligence built into the system, not a chatbot bolted on

Adding a chat window to a dashboard doesn't make a product intelligent. It moves the work onto the user, who now has to know what to ask and then check the answer. We put intelligence where decisions are made: anomaly checks on every bill before it's paid, tariff checks on every account, budget variances explained next to the numbers. You can ask questions in your own words, and we answer them with sources, but most of the value should reach you without having to ask.

3 Data as a narrative

How data is displayed and communicated decides whether anyone acts on it. Data is a story: what changed, why it changed, and what to do next. A red number on a dashboard starts a week of email. A sentence like "demand charges at this site rose after a new July peak, and the account moved to a different rate schedule" starts a decision. We design every view so that a finding can be explained to a CFO in one sentence.

4 Customizability

Every organization has its own sites, cost centers, GL codes, approval chains, and reporting calendar. People trust tools that fit the way they already work, and resist tools that ask them to start over. Custom reports, dashboards, approval rules, and exports into the systems a team already uses aren't extras. They're how trust gets built inside an organization.

5 Transparency and sourcing

Can you understand why the AI is doing what it's doing? In Nectar, every number traces back to a line item on the source bill, and every answer cites where it came from. When we flag a bill, we show which charge, compared with what, under which rule. Transparency turns a black box into a colleague whose work you can check.

6 Accuracy

Accuracy of outputs still matters most. Our AI parses bills at the line-item level with 99.2% accuracy, and we measure it continuously. Accuracy also means restraint. Most tools flag too much noise, so teams stop trusting the alerts. We'd rather raise fewer issues that are real than many that aren't, because one false alarm costs more trust than ten correct ones earn.

Trust is built in phases

Nobody hands over their accounts payable process to software on the first day, and nobody should. Trust is built through phased, thoughtful rollouts where people stay in control and never feel confined by the AI. Each phase gives a team evidence before it asks for more responsibility. The path runs from "just chat" to action.

1. See. Nectar collects and parses every bill and shows the data with its source. The team compares it with its own records. Confidence comes from source links on every number and measured accuracy.
2. Ask. Nectar answers questions in plain language, with citations. The team asks in its own words and spot-checks the answers. Confidence comes from cited answers that hold up when checked.
3. Flag. Nectar runs anomaly and tariff checks and explains each finding. The team decides what to do about each one. Confidence comes from low noise and a clear reason for every flag.
4. Act. Nectar prepares payments, disputes, and exports under the team's approval rules. The team sets the rules and approves exceptions. Confidence comes from a parallel billing cycle, bills with errors held rather than paid, and a full audit trail.

Customers move through these phases at their own pace. Some stay in the first two for months, and that's fine. The product is designed so that each step is visible and reversible. Our 30-day pilot runs on a team's own bills with no contract. Before Nectar pays a single bill, we run a parallel billing cycle so the team can compare our results with theirs. Bills with errors or disputed charges are held for review, never paid automatically.

Building confidence this way is slower on day one and much faster by month six, because nobody has to go back and recheck the work.

Trust comes from people

Software alone doesn't create trust. Relationships do. Behind the Nectar platform is a team that answers questions, explains findings, and fixes what is wrong. When a customer asks why a bill was flagged, a person can walk them through it. When we make a mistake, we say so and show the fix.

This is why we pair product with services. Account opening and closing, historical bill audits, and help with unusual billing situations are part of how we work with customers, not add-ons. Every interaction is a chance to strengthen the relationship, and every strong relationship makes the next step in adoption easier. Teams rely more on software when they know the people behind it.

How we work as a team

We apply the same principle inside Nectar. A team moves as fast as its members can trust one another's work.

• We use AI throughout our own work, in engineering, operations, bill review, and sales. Our sales team builds its own decks and case studies with AI. We adopt it the same way we ask customers to: AI drafts, people review, and autonomy grows only where the record is strong.
• We write decisions down. Designs and plans are written and reviewed before they're built, so the reasoning behind a change is visible to everyone, including the next person who touches it.
• We ship in small steps. Small changes are easy to check, easy to explain, and easy to reverse. That keeps the cost of trusting each other low.
• We hold ourselves to the customer's standard. The dependability of the data is the product. Nobody on the team treats "close enough" as done.

Early on, Katherine and I spoke with more than 200 sustainability teams, and the same theme came up again and again: they didn't trust their own utility data. That's still the problem we work on every day. Today we're a small team across engineering, operations, and go-to-market, and what we share is a belief that energy decisions should be made on data people can count on. The fastest way to get there is to earn trust one verified step at a time.

Questions to ask any AI tool you're evaluating

If you're deciding whether to bring AI into your energy, finance, or sustainability work, these questions separate tools you can rely on from tools you'll end up checking by hand:

• Can I see the source for every number and every answer?
• Can I start with read-only access and expand as I gain confidence?
• How does the tool tell me when it's unsure, and what happens when it's wrong?
• How many of its alerts turn out to be real?
• Does it fit the way my team already works, or do we have to start over?
• Who do I talk to when something looks off?

We built Nectar to have good answers to all six. If that's how you want to work, as a customer or as a teammate, we'd love to hear from you. For more on where AI fits in day-to-day bill management, read Utility Bill Management and AI.