If you already have a SaaS product, start with what exists: its website, current pricing, source repository, product documents and the decisions you have made. Collect that material into a sourced product brief, review the gaps, then use the saved context for your next marketing or product task.
1PC's Collect Product Data skill supports this starting point. It can use the files, URLs, accessible source code or notes you supply, researches useful gaps when your connected AI client has the tools, and saves the findings into your product's knowledge. You can bring an existing product into 1PC without restarting from idea validation.
Prepare the sources that describe your product today
Start with a small source set that can change the next recommendation. A homepage and pricing page may explain your offer; a product guide or the relevant source code may explain what the product is designed to do. Your own notes can add constraints and choices that are absent from public material.
- Website: the homepage, relevant feature pages, pricing and setup instructions.
- Source code or repository: a README, routes, manifests, configuration, release notes or the parts of the codebase that document current capabilities and the technical stack.
- Product documents: a current overview, release notes, a scoped roadmap or an onboarding guide. Older decks, briefs and docs are useful too when their date and current status are kept visible.
- Founder notes: your intended customer, accepted choices, available time and the question you want to work on next.
- Evidence: relevant observations you are authorized to share, with their date and limits.
Identify which sources are current. If an older deck says the product serves agencies but your current offer serves individual freelancers, tell the assistant which direction you accepted and when. If a repository branch or an old document disagrees with the live product, preserve that conflict rather than treating either source as current by default.
Each source has limits. A public page can establish what you promise; source code can document intended or implemented behavior; neither establishes that a feature is deployed, that customers use it or that they value the promise. Keep unknowns such as willingness to pay open until you have evidence.
If you want to organize the information manually first, use the free founder context template. It also works without a 1PC account.
Connect, choose the product and set the scope
Create and verify your 1PC account, then connect your AI agent through OAuth or a personal access token. Both connections act with your User permissions, including owner decisions. Use the existing account and connection for Products already saved in 1PC.
Your AI client needs access to your 1PC account and tools for reading the supplied sources. Website account creation alone does not connect ChatGPT.
Collect Product Data is available to every User. Accepted private MCP calls use prepaid Wallet credit; your AI client has its own capabilities and billing. See 1PC pricing.
Ask to use an existing accessible product or confirm a new one before saving. If you are refreshing a product already in 1PC, specify what should be updated. Existing knowledge and accepted choices should inform the refresh.
The following prompt is an example to adapt to your sources and client:
@1pc Bring my existing product into 1PC using the website,
repository/source code, current and older product documents, and notes I supplied.
Use Collect Product Data.
First summarize the saved product state and confirm the destination
if it is unclear. Collect the current product, customer, offer,
pricing and onboarding context. Research useful gaps if your tools allow.
Keep source dates, published claims, source-code observations, my statements
and your inferences distinct. Show conflicts and unknowns. Do not treat code
as proof that a feature is deployed or used. Preserve accepted decisions unless
I explicitly ask to change them.
Save the sourced context within this scope and read it back.
Then tell me what is ready to use for a landing-page review.
Product Owner checks the current state before the collection work. The connected agent reads sources and performs research using its available tools; 1PC stores the resulting text knowledge. Access to attachments, local files, repositories and browsing depends on the host and the permission you grant. If it cannot access a source, provide a supported format or record the gap rather than treating it as inspected.
A worked example: an existing follow-up product
This is a fictional teaching example, not a customer case study or a report of a completed import. The product, price and source labels below are illustrative.
Suppose you run Follow-up Board, a small SaaS that helps freelance designers track unanswered proposals. You have a working product, a public offer and ten hours this week to improve its landing page. You supply three documents:
| Source | Useful context | What remains open |
|---|---|---|
| Current landing-page copy | The offer describes proposal tracking for freelance designers. | Whether this is the best customer segment or message. |
| Current pricing-page copy | A published plan at $12 per month. | Whether anyone has paid or whether a new price would work. |
| Founder product notes | The founder reports a working proposal list and reminders. Agency features are outside the accepted scope. | Whether the features have been independently tested or used successfully by customers. |
A useful collected brief would describe the existing product and published offer, attribute the implementation details to the founder notes, preserve the scope choice and state that customer demand has not been established by these sources. It would also retain the ten-hour limit for the next task.
For example, a compact summary could read:
# Follow-up Board — illustrative collected context
Product: proposal tracking for freelance designers.
Published offer: one plan at $12/month, from supplied pricing copy.
Implementation: proposal list and reminders, reported in founder notes.
Accepted scope: no agency features, as stated by the founder.
Capacity: ten hours this week for a landing-page review.
Evidence gaps:
No payment records or customer interviews supplied.
No independent product-use check performed.
Next task:
Review the landing page against the current offer and product scope.
Do not describe demand or customer outcomes as validated.
The full source report should retain the document names or URLs, observation dates and unresolved conflicts. The summary is a starting point for later work; it does not replace those references.
Review what was saved before using it
Collect Product Data saves a source report alongside relevant product knowledge. The report records inspected material, dates, coverage, conflicts and gaps. Relevant findings can update the overview, customer context, positioning, messaging, business model, pricing, market, competitors, documented stack and onboarding files.
The scope determines which files need updates. Missing information should remain an explicit gap. An import does not accept new strategic choices, establish customer outcomes or move the product to a new lifecycle milestone.
- Check the destination. Make sure the saved records belong to the intended product.
- Check the important claims. Verify the offer, pricing, supported capabilities and customer description against their sources.
- Check conflicts. Resolve stale or contradictory material before it becomes the basis for a recommendation.
- Check persistence. Ask the client to read back the saved source report and relevant knowledge, and inspect the files in your workspace.
Large collection work can be saved in parts. A later session should read the actual saved report and current files to see what is complete and what remains. A confident chat reply alone is not evidence that the result was saved.
Use the context for one concrete next task
If several improvements compete for your time, use the weekly product review to choose one action from the collected state and available evidence. A source import gives the review current inputs; the review decides what to do with them.
Once the collected context is reviewed, choose a task that uses it. Product Owner reads the saved state, explains what is ready or missing, and selects the relevant built-in or published private skill. For Follow-up Board, that could be checking whether the landing page promises only supported capabilities and speaks consistently to the intended buyer.
@1pc Review my landing page using this product's saved overview,
customer context, messaging and source report.
First tell me which records you loaded and what remains unknown.
Recommend changes that fit my ten hours this week.
Keep proposed wording separate from accepted product decisions.
Save the reviewed deliverable; publishing is outside this request.
Open a fresh connected conversation and check that it loads the same product, offer and constraints. If the previous work is incomplete, ask it to continue from the saved report. The context continuity guide explains how to verify that loop.
Record changed direction in a product decision log, preserving earlier rationale. Collecting evidence does not accept a decision. Related marketing and research suggestions stay proposed until authorized. A published private skill can apply your own review method.
The custom product skills tutorial shows how to create that checklist, describe its input and output hints and publish a version. Keep product facts in the collected knowledge, while the private skill supplies your repeatable method.
Know what collection covers
Collection is a requested research and saving workflow. It does not continuously synchronize your website or repository, import every past chat or automatically keep pricing current. Refresh the relevant sources after a material change, and review the proposed updates against existing knowledge.
Use only material you are authorized to share. Review both 1PC's privacy policy and your AI host's data handling before providing private documents or customer information. Access to public MCP documentation does not expose your saved product data.
The steps here describe a workflow to try, rather than a verified run in every client. Check source access, saving and fresh-session reuse in your connected host. The MCP reference lists the current workflow access and connection requirements.
You can keep a manually maintained context file if that meets your needs. If repeated product work needs a shared record, bring one existing product into 1PC, review its collected context and use it for the next bounded task in your founder workflow.
Start with your product. Choose the next move.
Collect current context, then use it for Product decisions and related marketing work. The available skills use prepaid MCP credit.
Compare collection access and plans ↗