Frequently asked questions

Common questions about how Preflight works, what it analyzes, and how the AI is used. Something not covered here? Ask us directly.

What Preflight does

What is Preflight?
Preflight is a project readiness platform. It uses AI to analyze your intake documents — SOWs, proposals, briefs, discovery notes — and surface the gaps, ambiguities, assumptions, and conflicts that create delivery problems downstream. The output is a structured gap analysis, risk log, and scope baseline, along with a readiness assessment that tells you what is established, what is partial, and what is missing before work begins.

A useful way to think about it: CI/CD pipelines run automated checks before code ships to catch problems early. Preflight does the same thing for projects — running structured readiness checks on your documents before work starts, so that gaps and contradictions surface when they're cheap to fix, not after six weeks of design work built on them.
Is this just an SOW summarizer?
No. A summarizer tells you what's in a document. Preflight tells you what's wrong with it. It acts as an expert analyst — identifying specific, actionable risks, gaps, and contradictions that can jeopardize your project's timeline and budget. The difference between a summary and a gap analysis is the difference between knowing what a document says and knowing whether it's safe to build on.
Who is Preflight for?
Preflight is built for delivery consultants, digital agencies, and internal product teams running fixed-price or milestone-based software projects. It is most useful for project managers, delivery leads, and business analysts — the people responsible for intake quality who need to know what a document actually establishes before the team commits to building on it.
How is this different from a project management tool like Jira or Asana?
Jira and Asana are great for managing tasks. Preflight is for ensuring the work itself is well-defined and safe to start. Think of it this way: Preflight is what you use before you create the tickets in Jira, to make sure you're building the right thing and that the scope is actually clear enough to build from.
Does Preflight force a waterfall process?
No. Preflight doesn't prescribe how your team works — it surfaces what you know and what you don't, regardless of your delivery approach. Whether you're running agile sprints, a phased engagement, or a hybrid process, the underlying question is the same: is the knowledge required to do the next piece of work actually established? Preflight helps you answer that question. What you do with the answer is up to your team.
What does Preflight not do?
Preflight does not manage your backlog, run your sprints, or replace your project management tool. It does not auto-advance gates or make decisions on your behalf. It analyzes documents and surfaces issues — your delivery team reviews everything and decides what to do about it.

Documents and analysis

What types of documents can I upload?
Preflight accepts PDFs, Word documents, and plain text files. You can upload a single document or multiple documents for the same project — for example, a Statement of Work plus a discovery summary. The system analyzes them together as the project's founding intake material.
What does the analysis produce?
Each analysis produces a structured set of outputs: an intake gap analysis identifying what conditions are established, partial, or missing; a risk log surfacing delivery risks flagged in the document; a scope baseline statement summarizing what is in and out of scope; and a readiness assessment indicating which delivery stages are safe, conditional, or blocked based on current document state.
What formats can I export outputs in?
Every deliverable — gap analysis, risk log, scope baseline, entity enumeration, and others — can be exported as PDF for client-ready reports, Markdown for use in documentation systems or internal wikis, and JSON for programmatic use or integration with other tools. All three formats are available on every plan.
How is this different from just reading the document myself?
A careful manual review by an experienced PM can catch the same issues — but it takes time most reviews don't get. Preflight evaluates documents against a structured set of delivery conditions systematically and consistently, every time. It identifies issues that are easy to miss under time pressure: implicit scope assumptions, unstated constraints, contradictions between sections, and missing domain definitions that only become visible when you know exactly what to look for.
Does Preflight understand our specific domain or industry?
Preflight is domain-agnostic. It evaluates documents against delivery conditions that apply across software projects — problem definition, scope boundary, decision ownership, user identification, domain modeling, risk inventory, and data classification. It does not require industry-specific configuration to find gaps in these areas.

AI and decisions

How does Preflight use AI agents?
Preflight runs discrete AI agents — each one is a specialized analysis pass against your documents. When you run the Intake Gap Analysis, a dedicated agent evaluates your documents against each delivery condition and populates the results as structured, typed output. Agents run on demand: you trigger them when you're ready, review the output, and decide what to act on. Nothing runs automatically in the background without your initiation. Each agent produces structured data — not a free-text summary — so your team works with specific, actionable findings rather than generated prose to interpret.
Is this just a wrapper around ChatGPT?
No. Preflight uses AI to populate a structured delivery data model — it does not generate a narrative summary and call it analysis. Each run produces discrete, typed objects: truth conditions with defined states (defined, partial, or undefined), issues classified by type (gap, ambiguity, assumption, or conflict) and severity, and readiness assessments computed from those conditions. The AI's job is to evaluate evidence in your documents and populate that model. Your team works with structured data they can act on, not a block of generated text to interpret.
What are the limits of the AI analysis?
Preflight analyzes what is in your documents. It cannot see what isn't written down — prior client conversations, team knowledge, relationship context, or undocumented decisions made before the document was produced. If your team already knows that a flagged risk is low because you've handled the same integration twenty times, that context doesn't exist in the document. The right response is to confirm or resolve the issue in Preflight with a rationale, so the decision is on record. The analysis is a starting point for your team's review, not a replacement for it.
The AI flagged something that isn't actually a risk. What do I do?
Confirm or resolve it with a note. If your team knows an issue is low-risk — because of prior experience, an existing relationship, or context that predates the document — you can mark it as confirmed with a brief rationale. That decision is logged in the project record, so it's auditable if it ever comes up later. Preflight is designed for this: the AI surfaces candidates for review, and your team makes the final call on each one.
Can I trust the AI's analysis?
The AI provides expert-level analysis to augment your professional judgment, not replace it. Preflight is built around a human-in-the-loop principle: the AI surfaces potential issues, but you are always the final arbiter. You can confirm, override, or resolve every issue the AI finds. No gate advances, and no output counts, until your team approves it.
Can I override the AI's assessment?
Yes. If a condition appears unsatisfied in the document but your team knows the situation has been resolved — through a client conversation, a prior decision, or context that exists outside the document — you can mark it as confirmed with a rationale. The override is recorded in the project's decision log so the reasoning is auditable if it is ever revisited.
How accurate is the analysis?
Preflight uses structured prompting against defined delivery conditions, which produces consistent results across document types. Like any AI-assisted analysis, it can miss nuance or misread ambiguous language — which is why every output is reviewed by your team before acting on it. The goal is to make comprehensive review tractable, not to replace professional judgment.

Data and security

Is my confidential data used to train AI models?
Never. The content you upload is sent to our AI partners (currently Anthropic) via their secure business APIs under strict data processing agreements that contractually prohibit them from using your data for model training. Your data is your data.
Who owns the content I upload?
You do. As stated in our Terms of Service, you retain all ownership rights to the content you upload. We are granted a limited license only to process that content in order to provide the service to you.
Is it safe to upload client SOWs and confidential documents?
Yes. All data is encrypted at rest and in transit. Access is restricted to members of your workspace — no other organization can access your documents. We use industry-leading infrastructure (Supabase and Vercel) with strict access controls and multi-tenant isolation. If your organization has specific data residency or confidentiality requirements, contact us at help@preflightops.com to discuss Enterprise options.

Pricing and trial

How does the free trial work?
When you create an account, your first workspace starts on a 14-day free trial at the Pro plan level — no credit card required. After 14 days, the workspace drops to read-only access: you can view everything that was produced, but analysis and write actions are paused until you subscribe. The trial is once per account; you can't extend it by creating additional workspaces.
What happens after the trial ends?
Your workspace switches to read-only. All your project data, outputs, and reports remain accessible — nothing is deleted. To resume analysis and editing, subscribe to the Pro or Team plan from your workspace settings.
What is a workspace, and how does billing work?
A workspace is an isolated environment for your team and projects. Plans are attached to a workspace, not to a user account. You can have multiple workspaces — for example, one for each client organization or business unit — and each workspace has its own plan and billing. A single user account can belong to or own multiple workspaces.
What exactly is an AI analysis credit?
A credit is consumed each time you run one of Preflight's core AI agents on a project artifact — for example, running the Intake Gap Analysis, Risk Identification, or Scope Baseline on an uploaded document. Viewing reports, resolving issues, and other management tasks do not consume credits.
What happens if I run out of credits?
Analysis features will be paused until your next billing cycle begins. You'll be prompted to upgrade to a higher plan if you need more capacity immediately.
What counts as a seat?
A seat is a user within your workspace. Team members who need to view project outputs, review issues, or approve gate decisions each require a seat. The Pro plan includes up to 5 seats. The Team plan supports up to 25 seats. Enterprise plans include unlimited seats.
Can I cancel my subscription at any time?
Yes. You can cancel at any time from your workspace settings. Your plan remains active until the end of the current billing period, and you will not be charged again after that.
Can I upgrade or downgrade my plan?
Yes. Upgrades take effect immediately. Downgrades apply at the next billing cycle.

Working with existing tools

Does Preflight integrate with Jira, Linear, or other project management tools?
Not yet. Preflight currently operates as a standalone intake and gate analysis tool. Outputs can be exported as PDF, Markdown, or JSON for use in other systems. Native integrations are on the roadmap — contact us if this is a requirement for your team.
Do I need to change how my team works?
No. Preflight fits into the work your team already does at project intake. You upload the document you already have — the SOW, brief, or discovery output — and run the analysis before kickoff. It adds a structured review step to your existing process rather than replacing it.

Still have questions? Email us or visit the Delivery Methodology series to understand the thinking behind how Preflight works.