--- title: "How to use Claude and PickFu as your in-house research team - The PickFu blog" description: "Plan, design, and run real-customer product research from your AI chat, with the PickFu MCP working alongside other tools your team uses." --- # How to use Claude and PickFu as your in-house research team Most teams know they need feedback from their customers or audience before they ship a rebrand, a new product, or a major campaign. The hard part is figuring out what to ask, who to ask, and how to turn feedback into something the rest of the team can act on. The brief usually lives in Notion. The roadmap lives in Asana or Linear. The creative lives in Figma or Canva. The conversation about what to do next lives in Slack. None of those tools are the right place to design and run a research study or audience test – so the work either gets compressed into one rushed survey, kicked over to an agency, or skipped entirely and decided by gut. There’s a third option that’s gotten dramatically more practical in the last year: have your AI assistant pull information from the tools you already use, plan a testing process with you, run it on [**PickFu**](https://www.pickfu.com/how-it-works), and push the results back into the place your team will actually see them. That’s the workflow our co-founders [**Justin Chen**](https://www.linkedin.com/in/justinchen/) and **[John Li](https://www.linkedin.com/in/johnli/)** walked through in a recent webinar, using a fictional rebrand for “Daybreak Coffee” as the example. They used Claude to: 1. Pull the rebrand brief from Notion 2. Plan a research and testing process 3. Run surveys on PickFu (sent to real people in their target audience) 4. Add the results back into Notion when finished You can watch the recording below, or keep reading to get a step-by-step breakdown – with the prompts to run it on your next project. ## AI assistants + your other tools = powerful workflows The first phase of AI was about having it _synthesize_ things: summarize a doc, draft an email, explain a concept. The phase we’re in now is about having AI _do_ things for you, in the tools you already use, with full context from your existing systems of record. That second phase only works if your AI assistant can see the same things your team can: your project brief, your past customer feedback, your roadmap. The more context you give it, the better the work – just like with a human collaborator. MCPs are the standard that makes that possible: a connector between your AI chat and the tools where your work actually lives. When you wire up PickFu and Notion (or Asana, Google Drive, Slack, whatever you use) into the same chat, the AI can pull a brief, propose a study to validate it, run that study with real consumers, and push the results back into the document where the decision will get made. You stay in one conversation. The work flows back to your team automatically. ## The workflow in five steps 1. **Start where your team already works.** Pull a brief, roadmap, or PRD from Notion (or any tool you use) into the AI chat. 2. **Plan the audience.** Have Claude recommend the targeting traits that’ll get you the cleanest signal. 3. **Build the survey.** Use multiple question types in one survey — ranked, open-ended, multi-select — to get layered insights. 4. **Organize as you go.** Tag and group surveys into projects so future-you can mine them later. 5. **Close the loop.** Push the survey link (and eventually the results) back into the tool your team is using, so they can collaborate and take action easily. ## What is the PickFu MCP? MCP stands for Model Context Protocol: a (still relatively new) open standard that lets AI assistants like Claude, Cursor, and ChatGPT talk directly to other tools without you having to copy and paste anything between tabs. The [**PickFu MCP**](https://www.pickfu.com/docs/developers/mcp-server) lets your AI assistant create surveys, target audiences, generate images, organize results, and pull data on your behalf. You chat with the AI in plain language, and it handles the rest, without leaving the conversation. The MCP works with any MCP-compatible client. We recommend Claude because it’s the easiest to set up and the most reliable for multi-step work like this, but Cursor, Claude Code, and ChatGPT (with the right plan) all work too. Setup in Claude takes about a minute: 1. Open Claude, then click **Customize → Connectors**. 2. Click the **+** button → **Add custom connector**. Name it PickFu and paste in this URL: `mcp.pickfu.com/mcp`. 3. Authorize with your PickFu account. Heads up if you’ve gone looking for it: the PickFu MCP isn’t in Claude’s connector marketplace yet (we’re working on it), so you’ll need to add it as a custom connector for now. Full setup instructions are [**here**](https://www.pickfu.com/docs/developers/mcp-server). If you don’t have a PickFu account yet, [**sign up for free**](https://www.pickfu.com/ai). 💡 **A quick AI tip before we dive in:** don’t enable every connector you have at the same time. AI models have a fixed amount of context – think of it like desk space. The more connectors you load in, the more “stuff” the model has to keep track of, and the more easily it gets confused. Just enable the tools you’ll actually need during each conversation. ## Step 1: Start where your team already works Most research projects begin with something written down: a creative brief, a PRD, a campaign plan, a launch goal. Your AI assistant can pull from any of those and use it as the starting point for a research plan. In the demo, we pointed Claude at a Daybreak Coffee rebrand brief in Notion. The brief had everything you’d expect – repositioning goals, audience shift (men 30-50 → women 25-44), three new logo concepts, tagline candidates, color palettes, and a list of what needed to be validated. [![](https://www.pickfu.com/blog/wp-content/uploads/2026/05/Screenshot-2026-05-09-at-2.09.11 PM-970x886.png)](https://www.pickfu.com/blog/wp-content/uploads/2026/05/Screenshot-2026-05-09-at-2.09.11 PM.png) **We used this prompt:** > _“Pull our Daybreak Coffee rebrand brief from Notion and suggest a PickFu research plan to validate it with our target audience.”_ What you’ll get back is a structured plan – usually a few surveys covering the things that need to be validated, plus suggested formats and audiences for each. The more context you give AI, just like with humans, the better the work. A clear brief in Notion plus a clear ask in the chat tends to produce a much sharper plan than starting from scratch. If you don’t have a fully fleshed-out brief yet, that’s also fine. You can chat with Claude about a vague idea or goal, and it’ll help you build the brief itself. Some of the best research starts as a conversation, not a finished document. ## Step 2: Plan the audience PickFu has over 100 targeting traits you can layer to reach exactly the kind of person who’d buy your product. The trade-off: the more traits you stack, the smaller and slower your audience pool becomes. John’s recommendation during the demo: go broader than you think you need to. You’ll get more responses, you’ll get them faster, and you can use **demographic reporting** (a separate setting that asks panelists about themselves _after_ they answer your main questions) to slice the data by audience segment after the fact. Customers regularly find that what they thought was their target audience isn’t actually where the strongest signal comes from – and broad targeting plus demographic reporting is what surfaces those pockets. The distinction between targeting and demographic reporting comes up a lot, so worth being explicit: - **Targeting traits** filter the pool. Only people who match the traits you choose can answer the survey at all. - **Demographic reporting** asks for additional info after the fact. Everyone in your pool can answer, and you’ll see the demographic breakdown alongside their responses so you can slice and dice afterward. **Try this prompt:** > _“Recommend the 1–2 targeting traits that’ll give us the most useful signal for the Daybreak rebrand. Then add education and household income as demographic reporting questions so we can slice the responses afterward.”_ Another useful framing while you’re drafting questions: think about your target audience as if they were in a focus group with you in the room. What would you actually ask them? Claude is surprisingly good at surfacing the kinds of questions a research pro would ask: purchase motivations, switching behavior, emotional reactions to the brand, the words people would use to search for the product. A lot of those go unasked when teams default to “which logo do you like better?” ## Step 3: Build the survey with multiple question types You can ask up to 16 questions in one PickFu survey, with a mix of question formats. That’s an underused feature – most people default to a single poll, but a deeper survey will tell you a lot more about your audience in one shot. For the Daybreak rebrand, that meant one survey with several questions: a ranked poll on the logo concepts and candidate taglines, an open-ended question asking respondents to describe the brand for the logo they chose, and a multi-select question on which messaging pillars resonated most. With every question type, the audience submits a written comment explaining their choice, so you’ll always get qualitative feedback. The open-ended feedback is usually what will surprise you. Getting your audience to describe your product or brand in their own words is one of the most useful things you can do – especially when you’re rebranding, naming a new product, or figuring out how to rank in search or which words to use in ad copy. **We used this prompt:** > _“Build a PickFu survey with multiple questions to validate the Daybreak rebrand — positioning, logos, taglines, and overall tone. Include at least one open-ended question asking respondents to describe the brand in their own words.”_ If you’ve never designed a research study before, this is also where Claude shines as a thinking partner. Ask it what questions you should be asking, what order to ask them in, what to leave out. Iterate on the draft, look at what it created, and adjust until it feels right. ## Step 4: Organize as you go Tags and projects sound like housekeeping, but they pay off the moment you have more than a handful of surveys to your name. Some of our customers run thousands of polls – finding “all the Q2 image tests for ASIN B07XYZ” is a different problem at that scale. Two simple habits make a big difference: - **Tag every survey when you launch it.** Useful tag categories: campaign name (`daybreak-rebrand`), function (`logo`, `tagline`, `positioning`), product or SKU IDs, brand names, season, quarter, year. The point is to make future-you’s search easier. - **Group related surveys into a project.** A project is a folder for a single initiative. PickFu auto-generates an AI summary across the surveys in a project, so when you come back to it later you’ll see a synthesized view of what you’ve learned, not just a list of tests. [![](https://www.pickfu.com/blog/wp-content/uploads/2026/05/Screenshot-2026-05-09-at-2.09.39 PM-970x787.png)](https://www.pickfu.com/blog/wp-content/uploads/2026/05/Screenshot-2026-05-09-at-2.09.39 PM.png) **Try this prompt:** > _“Tag this survey with ‘daybreak-rebrand’, ‘logo’, and ‘Q2-2026’, and add it to the ‘Daybreak Q2’ project in PickFu.”_ You can do all of this from the PickFu web UI too – it’s just faster to delegate the housekeeping to Claude when you’re already in the chat. ## Step 5: Close the loop Once a survey is live, the last step is making sure the results end up where your team will actually look at them. In the demo, Claude wrote the live survey link back into the Daybreak rebrand page in Notion as soon as it launched, then added a section for “research in progress” that anyone on the team could check. Once results came in, we asked it to add the headline numbers, the AI summary, and recommended next steps directly into the same Notion page. That sounds small, but it’s the difference between research that drives decisions and research that gets forgotten. As Justin put it: the AI chat isn’t where you do your day-to-day work — Notion or Asana or Google Drive is. The whole point of connecting these tools is that your work flows back to where your team is. [![](https://www.pickfu.com/blog/wp-content/uploads/2026/05/Screenshot-2026-05-09-at-2.10.23 PM-970x885.png)](https://www.pickfu.com/blog/wp-content/uploads/2026/05/Screenshot-2026-05-09-at-2.10.23 PM.png) **Try this prompt:** > _“Add a link to this PickFu survey to our Daybreak rebrand Notion page so the team can check results.”_ **And when the survey finishes:** > _“Pull the survey results into the Daybreak Notion page: show us which options won and why, key trends or themes, and recommended next steps.”_ The same pattern works for any tool you’re using: Asana tasks, Linear tickets, Slack threads, Google Docs. Wherever the decision lives, the research can land there. ## FAQs **Do I need to pay extra to use the MCP?** No. The MCP is included with every PickFu account, including free accounts. You only pay when you launch a poll – pricing starts at $1 per response. **I’m an agency managing multiple client accounts. Can the AI know which client account to use?** Not yet, but it’s on the roadmap. Our agency program lets you manage multiple client accounts on the web today, but the MCP doesn’t currently support delegated client account access. **Will Claude notify me when my poll finishes?** Not in standard chat clients. You’ll get the same email notifications you’d normally get from PickFu, then check back in with Claude when you’re ready to pull results. (While a poll is running, you can ask Claude to analyze partial results that have come in so far.) **How do I send feedback or feature requests?** The PickFu MCP includes a feedback tool. Just tell Claude what didn’t work or what you wish it could do, and it’ll send the feedback to our team with full context so we can work on it. ## Beyond rebrands The same workflow applies to almost any decision your team needs to validate with real customers: - **Product launches.** Pull the launch brief, validate positioning, pricing, packaging, and main image in parallel. - **Marketing campaigns.** Test headline directions, ad creative, and audience messaging before you put paid spend behind any of it. - **Naming and copywriting.** Pressure-test names, taglines, and descriptions with the people you’re trying to reach. - **Roadmap prioritization.** Get qualitative signal on which features your customers actually care about before you build them. - **Existing product audits.** Benchmark your current creative, copy, or packaging against competitors to find the biggest opportunity to improve. If the input lives in a doc and the decision needs to land in a doc, this workflow can connect the two ends. ## Try it on your own next project You can run this workflow today. Most survey results come back the same day, and the planning step takes minutes once your AI assistant has the context. [**Start here**](https://www.pickfu.com/ai) for everything you need: free PickFu signup, MCP setup docs, prompt examples, and upcoming events. We also run live workshops fairly regularly. If you want to see what’s coming up next – or any of the other AI-powered research workflows we’re sharing – [**check out our Luma page.**](https://luma.com/pickfu-events) * * * **About PickFu** PickFu is an on-demand consumer insights platform. Build a survey, choose your target audience from our verified global panel, and get in-depth feedback within hours. Founded in 2008 and used by brands of all sizes worldwide to make confident decisions. **[Sign up for free.](https://www.pickfu.com/users/sign_up)** Surveys start at $1 per response. ![alt](https://www.pickfu.com/blog/wp-content/uploads/2026/02/cropped-Adrienne-headshot-1-1-scaled-1-96x96.jpg) #### Adrienne Van Niman Adrienne Van Niman is the Marketing Lead at PickFu. She has 8+ years of experience as a marketer and writer, specializing in content strategy and wearing many hats for growing B2B tech companies. Outside of work, she loves to read, travel, go to concerts, and spend time in the great outdoors. ![PickFu Logo](https://www.pickfu.com/blog/wp-content/themes/twobitoperation_pickfuwpbs4/assets/images/pickfu-logo/pickfu-logo-white.svg) Stop guessing what your customers want [Sign up for free](https://www.pickfu.com/users/sign_up "Sign up for free")