Docsy

Most questions at Hardal have already been answered somewhere. The difficult part is remembering whether the answer lives in a document, a Slack discussion, a support ticket, or a customer record.

Docsy is our fastest first stop for finding that context.

What Docsy is

Docsy is Hardal's internal knowledge assistant. Hardalians use it in Slack to search connected company sources and bring the most relevant information into one answer.

It is useful when someone needs context about Hardal, our products, a customer, a partner, or a previous decision but does not know where to begin.

We ask Docsy first. Before sending a repeated question to a teammate, we start with Docsy. This reduces interruptions, makes existing knowledge easier to reuse, and gives the person we eventually ask a better starting point.

Docsy helps us find context. It does not make decisions or replace the owner of a process.

What Docsy searches

Docsy can search several types of Hardal knowledge:

These sources do not all have the same authority or freshness. A current product document is usually stronger evidence than an old Slack message. CRM information can also become outdated if the original record has not been maintained.

When timing matters, we ask Docsy where the answer came from and how recent the source is.

How we ask questions

Docsy works best when the question includes:

• the exact product, customer, partner, or process,
• the fact or decision needed,
• relevant environment or date context,
• what the person plans to do with the answer.

We ask one clear question at a time. Broad prompts produce broad answers.

Too broad: "What do we know about this customer?"

Better: "Has this customer reported any problems with consent configuration in the last three months, and what was the latest resolution?"

Too vague: "How does Meta work?"

Better: "What is the current Hardal setup flow for Meta Conversions API, and which document explains it?"

Example questions

Docsy is a useful starting point for questions such as:

• "What is the current setup process for this product or destination?"
• "Has this customer reported the same problem before?"
• "What did we decide about this feature, and when?"
• "Which document is the source of truth for this topic?"
• "Who owns the next action for this issue?"
• "What are the known limitations of this feature?"
• "Which customer or partner records mention this requirement?"
• "What guidance have we previously given for this implementation?"
• "Is this request already covered by an existing Product decision?"

If the first answer is incomplete, we narrow the question. We can ask for the most recent source, a specific timeframe, or the missing part directly.

Good usage practices

Provide enough context. Names such as "the dashboard" or "that customer" are ambiguous. Use the real product area, company, feature, environment, and timeframe.

Ask for facts before recommendations. First establish what happened, what is documented, and what was previously decided. Then decide what to do.

Check freshness. Product behavior, customer status, ownership, and commercial terms can change. Ask for the most recent available context when acting on time-sensitive information.

Open the original source when the risk is high. Do not act on a summary alone when the answer affects money, security, privacy, legal terms, customer commitments, or production access.

Confirm with the responsible owner. Docsy can retrieve a previous decision, but it cannot approve a new exception, promise a delivery date, or commit Hardal to a customer.

Fix the source, not only the answer. If Docsy finds outdated information, correct the original document, CRM record, support ticket, or Product decision. Otherwise the same wrong answer may appear again.

Protect sensitive information. Do not paste credentials, secrets, access tokens, payment details, or unnecessary personal data into a question.

When Docsy cannot answer

A missing answer is useful information. It usually means that the knowledge was never written down, is stored in an unconnected system, or is too ambiguous to retrieve reliably.

When this happens:

1. Ask a narrower question.
2. Check whether the relevant source is current.
3. Contact the person who owns the context and include what Docsy already found.
4. Record the confirmed answer in the right source of truth.

The goal is not to make Docsy answer every question. The goal is to make Hardal's knowledge easier to find and reduce the number of times the same question has to be answered from scratch.