Save what you learn

Save what you learn

Your team already knows a lot. The trouble is that most of it lives in scattered files, old email threads, and people’s heads — so the same questions get asked again, and good answers get re-created from scratch.

This walkthrough shows you how to turn that scattered knowledge into something durable: a Knowledge collection Chat can search and cite, paired with an assistant tuned for the job and live Web Search for what’s happening in the market. We’ll follow one realistic story end to end — building a competitive analysis that blends your own product docs with up-to-the-minute web research, then turning the result into a reusable brief you hand to Marketing.

By the end you’ll have:

• A shared collection of your proprietary product docs Chat can ground answers in
• An assistant set up for competitive research
• A synthesized competitor comparison — differentiation, risks, opportunities
• An exported, reusable competitive brief, shared with your team

The story

You’re a product marketer. A new competitor just shipped, and your VP wants a sharp read on how you stack up — using your real positioning, not generic internet summaries. You want it grounded in your product truth, refreshed with live market signals, and saved so the next person doesn’t start from zero.

Before you start

• You can create Knowledge collections and folders (most members can; if you can’t, your admin can grant it).
• You have your product docs handy — datasheets, positioning briefs, FAQs, release notes. Each file needs to be a supported type and under 100 MB.
• You have a folder (or are happy to create one) to keep this work together with the people who’ll use it.

Step 1 — Put your product docs into a Collection

Knowledge is where Chat keeps documents it can search and cite. Instead of pasting your positioning into every prompt, you load it once and let every answer draw on it.

1. Open Knowledge from the sidebar and select New collection.
2. Name it something findable — e.g. Product Positioning & Specs.
3. Choose Add Source → Upload and add your files: datasheets, the positioning brief, the pricing FAQ, recent release notes. Each shows progress as it’s processed; Chat converts Office files to text and indexes everything for semantic search behind the scenes.
4. Wait for the documents to finish processing. They’re now searchable and citable.

Tip — quality in, quality out. A tight set of current, authoritative docs beats a dumping ground. Add your real positioning, not drafts you’ve moved on from. You can always add or remove sources later.

For the full mechanics — supported types, web crawls, and how to promote an existing attachment into a collection — see Upload a file to work with.

Share the collection with your team

A collection is only as useful as the people who can reach it. Pick the scope that matches your audience:

Scope	Who can see it	Use it for
Personal	Only you — private, even from admins	A working draft you’re not ready to share
Folder	Members of a specific project folder	A team’s shared source of truth
Tenant	Everyone in your organization	Company-wide reference material

For this work, share the collection at Folder scope with your competitive-analysis folder (you’ll create it in Step 3), or grant access to specific colleagues. Widening scope is deliberate and one-way, so start where it belongs.

Step 2 — (Optional) Set up a Competitive Analysis Assistant

An Assistant is a reusable setup Chat runs against — a system prompt, a persona, and a linked Knowledge collection. Building one means you don’t have to re-explain the job every time.

1. Go to Assistants and create a new one — call it Competitive Analysis Assistant.

2. Give it a system prompt that frames the task, for example:

   You are a competitive analysis researcher. Compare our product to named competitors using our internal Knowledge as the authority on our own capabilities, and current web sources for competitors and market context. Always distinguish our differentiators, risks, and opportunities. Cite your sources and flag anything you’re unsure about.

3. Link Knowledge — attach the Product Positioning & Specs collection from Step 1 so the assistant grounds every answer in your real product truth.

4. Save it.

If you’d rather not build one, you can do this entire walkthrough by linking the collection directly to a conversation and writing your instructions in the prompt. A dedicated assistant just makes it repeatable. See Choosing an assistant for when each option fits.

Step 3 — Create a shared Competitive Analysis folder

A Folder keeps related conversations, knowledge, and people in one place — the home for this project.

1. From the sidebar, create a folder named Competitive Analysis.
2. Add the colleagues who’ll contribute — your fellow marketers, maybe a product manager. Members can collaborate in the same conversations in real time.
3. Grant the folder access to your Product Positioning & Specs collection (or confirm the Folder-scope sharing you set in Step 1).

Now start a new conversation inside the folder. Because it lives in the folder, teammates with access can follow along, jump in, and pick up where you left off.

For more on organizing this way, see Organize your conversations.

Step 4 — Turn on Web Search and link your Collection

You want the analysis grounded in your docs and refreshed with live market signals. Set both before you ask.

1. In the composer, open the + menu and turn on Web Search (or use the always-visible Web pill). The setting sticks for this conversation.
2. Confirm your Product Positioning & Specs collection is in play — via the Competitive Analysis Assistant from Step 2, or by referencing it directly in the conversation.

With both on, Chat consults your internal Knowledge for what’s true about you and searches the web for current information about competitors and the market — blending the two and citing each.

For the full picture of how web results are retrieved and cited, see Web Search.

Step 5 — Ask for the competitive analysis

Now make the request. Be concrete about competitors, criteria, and what “good” looks like.

Compare our product against [Competitor A] and [Competitor B] across pricing, deployment options, security and governance, and ease of adoption. Use our Knowledge for our own capabilities and current web sources for the competitors. For each criterion, give a short verdict and cite your sources.

Chat works through the request and replies with a structured comparison. You’ll see:

• Inline citations — numbered [N] markers you can click to see the exact source, with a Sources Used card listing them. Internal points trace back to your docs; market points trace back to live web pages, so you can tell at a glance where each claim came from.
• A grounding badge on the answer, showing it leaned on real sources rather than guessing.

If a point isn’t covered, Chat tells you it’s answering from general knowledge rather than inventing a citation. That’s your cue to add a source or narrow the question.

Step 6 — Iterate on scope, competitors, and criteria

The first pass is a starting point, not the finish line. Refine it conversationally — the collection and web search stay on, so each follow-up builds on the last.

Adjust the competitor set:

Add [Competitor C] to the comparison, and drop [Competitor A] — they’re not in our segment.

Sharpen the criteria:

Re-do the security and governance row in more depth: data residency, audit trails, and access controls. Quote specifics where you can.

Tighten the scope:

Focus only on mid-market buyers. Re-frame each verdict for that audience.

Each reply re-grounds against your Knowledge and fresh web results. If a row looks thin, the Refine this button under the answer asks Chat to improve it without retyping your request.

Step 7 — Ask for a synthesized comparison

Once the pieces are right, ask for the analysis that actually helps you decide.

Now synthesize this into a competitive brief: our clearest points of differentiation, the biggest risks where competitors are ahead of us, and the top opportunities to win. Keep it tight and decision-ready, and cite the evidence behind each point.

Chat pulls the comparison together into a narrative — what sets you apart, where you’re exposed, and where the openings are — with citations carried through so every conclusion is traceable.

Tip. If you need a deeper, more thorough pass, switch the conversation to a higher reasoning tier before this step — the analysis will do more digging across your sources and the web. You’ll find the tier selector in the composer.

Step 8 — Export it as a reusable competitive brief

When the brief is solid, get it out of the chat and into a shareable document.

1. Open the conversation or message menu and choose Export.
2. Pick a format — PDF for a polished handout, Markdown for editing in your own tools, or Text / JSON for further processing.
3. Save the file. It captures the synthesized brief, ready to circulate.

See Export for all the formats and where to find the option.

Step 9 — Share it with Marketing

Get the brief in front of the people who’ll act on it:

• Share the folder or conversation with your Marketing teammates so they can read the full thread, see the citations, and ask their own follow-ups against the same grounded sources.
• Or drop the exported brief into a Channel — a persistent, multi-person space — where the team can discuss it, pin it, and keep it close at hand.

Because the underlying Product Positioning & Specs collection is shared, the next person who needs a competitive read starts from your work, not from scratch.

The outcome

You started with scattered product docs and a vague ask. You finished with:

• A shared, citable Knowledge collection of your real product truth
• A repeatable assistant for competitive research
• A synthesized, evidence-backed brief blending internal knowledge and live market signals
• An exported document circulating with Marketing — and a collection the whole team can build on next time

That’s the heart of saving what you learn: good work stops being a one-off and becomes something your organization can find, trust, and reuse.

Related

• Upload a file to work with — build the Knowledge collection at the center of this flow
• Web Search — blend live market signals into grounded answers
• Choosing an assistant — set up a reusable research assistant
• Make decisions with confidence — turn analysis into a clear call