Project Knowledge

Project Knowledge is the free-text background context Harmony's AI is given on every insight run for a project. It is not part of the conversation transcript — it is concatenated alongside the prompt for each definition, so the AI interprets the transcript with the right business context in mind.

Project Knowledge is project-scoped: it applies only to the project it lives on, not the whole workspace. The workspace has its own organisation-wide context field — see the Handbook on the General Settings page.

Why context matters

A generic AI can be tripped up by your team's nicknames, abbreviations, or product names. With a few paragraphs of context, Harmony can correctly interpret your competitors, your packaging tiers, your sales methodology, or anything else that matters for the meetings in this project.

If your Insights feel generic, adding two or three paragraphs of project-specific knowledge usually sharpens results noticeably.

How to add or edit Project Knowledge

Project Knowledge lives on the Basics tab of the project (and is also editable from the Configure Insight page when you create a new definition).

1. In the sidebar, open Data Lake and pick the project.
2. Open the Basics tab (?tab=edit).
3. Find the Project Knowledge section.
4. Type or paste your context.
5. Save the project. Harmony uses the new knowledge on every subsequent insight run.

The project header includes a Knowledge Base indicator that shows whether Project Knowledge has been added (a green book icon means yes).

You can also add or edit Project Knowledge inline from the Configure Insight flow when creating a new definition — there is a dedicated Project Knowledge panel in the right column of that page.

Editing Project Knowledge applies to future insight runs. To apply the new knowledge to past meetings, run a Revision from the Revisions tab — see Reports and exports.

What to include

When adding content to Project Knowledge, it’s important to think about the kinds of information that make your project unique, and the details that would help an AI “think like” someone on your team. Project Knowledge is most effective when it provides a rich, narrative explanation of the language, culture, and nuances of your business as they pertain to the project at hand.

Start by describing any terminology, abbreviations, or jargon that are specific to your company or industry. For example, if your team frequently refers to “MQL,” clarify that this stands for Marketing Qualified Lead and explain any criteria that must be met for a lead to qualify—such as needing a confirmed budget over $10,000. This helps the AI reliably interpret shorthand or internal lingo that might mean something different elsewhere.

Next, give a clear summary of your core products and any feature names, especially those that might sound generic out of context. It’s helpful to explicitly state, for instance, that your product “Flow” is a workflow automation tool, so the AI doesn’t confuse mentions of “flow” with the general concept of conversation flow or process steps often discussed in meetings.

Information about your competitive landscape is also valuable. Listing your key competitors and noting how your team and customers commonly refer to them allows the AI to accurately distinguish between your offerings and those of others. This detail is especially useful for insights like Competitors Analysis, where recognizing named competitors impacts the precision of results.

Providing a brief explanation of your pricing structure, packaging tiers, or any exceptional deal arrangements is another way to ensure the AI evaluates meeting discussions with accurate commercial context. Clarify what each tier means (e.g., Solo, Plus, Max), and mention if your team handles custom deals, discounts, or bundles in a non-standard way.

Finally, if your commercial or sales team operates within a known methodology—such as SPIN, MEDDIC, or another process—make sure to mention it and give a short overview. Naming your methodology and describing the core ideas behind it enables the AI to apply the relevant framework, vocabulary, and reasoning style when reviewing transcripts.

It is also important to understand the distinction between Project Knowledge and Additional Context. Project Knowledge is intended to provide foundational context that applies across all insight definitions within the project. This information is stored on the Project Basics tab and serves as the broad background for interpreting every meeting in the project. In contrast, Additional Context is specified individually for each insight definition during its configuration. This field should be used for specific instructions or clarifications that relate to a single insight definition, rather than the entire project. Use Project Knowledge to supply overarching, persistent context that touches every analysis within the project, and use Additional Context when you need to guide the AI in a more targeted way for a particular definition.

Use Project Knowledge for product names and methodology; use Additional Context to nudge a single definition's behaviour.

Project Knowledge vs the workspace Handbook

These are different fields:

• Project Knowledge lives on a single project under Data Lake → {project} → Basics. It applies only to insight runs for that project.
• Workspace Handbook lives on Settings → General. It is workspace-wide context Companion and other AI surfaces can use across the org.

If a piece of context matters for one initiative only, put it in Project Knowledge. If it matters across the entire organisation, put it in the workspace Handbook.

How the AI uses this data

When a meeting is processed for a project, Harmony reads the project's Knowledge first and uses it as the lens for interpreting the transcript. For a Yes/No or Checklist definition, the AI checks the transcript against your definitions to keep the answer aligned with your business.

For privacy and training questions, see Models and training data.