AI is now everywhere
AI is now a feature of almost every business software product. Some uses are genuinely valuable. Many are decorative.
It is easy for a SaaS vendor to add a chatbot, attach an AI-generated label to an existing feature, and call the product "AI-powered".
It is far harder to use AI responsibly — in a way that respects customer data, communicates honestly about limitations, and treats AI as a tool rather than a marketing prop.
This article is about how we approach it.
What we use AI for
Inside KIMISUITE, AI is used to support specific, concrete tasks: drafting communications, generating page content for landing pages, suggesting next actions, summarising notes, and similar productivity-focused work.
The principle is simple. AI is an assistant. Humans decide.
We do not position AI as a replacement for judgment, for professional advice, for compliance review, or for customer relationships. We are sceptical of vendors who do.
What we do not do with your data
This is usually the more useful list.
We do not train external AI models on your private workspace content. When AI is used inside the platform, customer data is processed for the immediate purpose of generating a response — not pushed into a pool of training data for future model improvements.
We do not feed your customer information into general-purpose chatbots that you did not choose to use.
We do not share AI inputs and outputs with marketing partners, advertising networks, or unrelated SaaS vendors.
We do not treat AI as a quiet feature you have to discover. AI use is visible, configurable, and avoidable if you prefer.
These are not aspirational statements. They are operational choices.
Where AI fits — and where it doesn't
AI is a fit when the task is generative, repetitive, or pattern-based: drafting a first version of a message, summarising a long document, suggesting a tag, proposing a label.
AI is not a fit — and we explicitly do not use it that way — when:
- the task requires professional advice (legal, tax, medical, regulatory);
- the task requires reliable certainty (financial calculations, compliance verdicts);
- the task requires accountability (decisions that affect employees, customers, money).
In those cases, AI may suggest, but a human is responsible.
Our AI Usage Policy spells this out in plain language and is linked from the footer of every page.
AI is not perfect — and we say so
A growing problem in SaaS marketing is the implicit promise that AI is a kind of automated oracle. It is not.
AI can produce inaccurate, incomplete, outdated, misleading or fabricated information. That is a property of the technology itself. No amount of vendor reassurance changes it.
Customers should review, validate and approve AI-generated output before relying on it. We will say this in our policies, inside the product, and in any AI-powered feature where it matters.
We would rather be honest about the limitations than market our way around them.
Where AI lives technically
Some AI tasks are processed by external AI providers — specialised model vendors who have built and trained the underlying language models. We use them carefully and selectively, for the same reason businesses use a specialised payment processor: the alternative would be to build a global AI lab ourselves, which would not be in our customers' interest.
What we do control is how that AI is used. Inputs are scoped to what the feature needs. Outputs are returned to your workspace, not to a third-party content pool. The relationship between your data and any external AI provider is governed by contracts we maintain — including obligations to keep your content private.
We never list specific AI vendor names in policy documents that age quickly. The current set of sub-processors is documented and updated as part of the Privacy Policy.
What "responsible" actually means here
"Responsible AI" is a phrase that gets used loosely. We use it to mean a small set of concrete commitments.
Use AI only where it genuinely helps. Disclose where it is used. Keep customers' content out of training pipelines. Reject use cases where AI is dressed up as authority. Be honest about what the technology cannot do. Let customers turn AI features off when they prefer human-only workflows.
None of those commitments are dramatic. Together, they are how we keep the platform trustworthy as AI keeps changing.
A note on the future
AI is changing quickly. Some of the things AI can do in 2026 will look quaint by 2030. Some of the things it cannot do today may be possible tomorrow.
That is a reason to be careful about marketing, not a reason to be reckless with customer data.
Our commitments — what AI is used for, what we do not do with customer content, how it is disclosed — are designed to remain valid even as the underlying technology evolves.
Final thoughts
AI is a tool. Tools are useful when they are used for the right job by people who know how to use them.
We use AI to help our customers do their jobs faster. We do not use AI to build profiles, to monetise content, or to substitute for judgement.
That is not a marketing position. It is the foundation we have chosen to build on.
Continue reading the Trust Series:
← Previous: Why business software should be predictable
→ Next: Why we don't chase every trend