Most software hands you an empty workspace and wishes you luck. Offra runs the opposite way: the first three minutes are spent describing your firm, not your tender, because everything the system does later — pre-filling an answer, judging whether a requirement is already covered, estimating whether a bid is worth chasing — is grounded in what it knows about you. Here is what the first ten minutes actually contain, including the parts that are not on every plan. ## Minutes 1–3: your company, before any tender There is no self-serve sign-up form. An account reaches you one of two ways — a colleague invites you, or someone on your team buys a plan — and both land on the same activation page, where you set a password and sign in. From there you are held on **Let's set up your workspace** until it is done. Three steps, a **Step 1 of 3** counter, and a **Skip for now** on every one of them. **Step 1 — Your business.** The shortcut here is worth taking: paste your company website into **Company website** and press **Extract**. Offra reads your public pages and pre-fills your legal name, address, phone and services, then tells you plainly that it did — *We pre-filled these from your website. Please review and complete them before continuing.* Review it; extraction is a head start, not a source of truth. **Enter details manually** is right there if your site is thin. There is no region switch to find. Your market is derived from the country and province or state you enter, and it decides more than it looks: your default interface language, which credential fields you are asked for (RBQ and NEQ in Quebec, state licensing in the US), and which public sources feed your opportunities later. **Step 2 — Your areas of expertise.** Plain language is fine: *road paving, water mains, structural concrete, building renovation*. This is what a tender gets matched against. The same step carries the AI-training toggle, on by default, which keeps your documents out of model training while still letting Offra read them to run the analysis. **Step 3 — Invite your team.** Everyone is invited as a member; you promote to admin afterwards in **Settings → Members**. You can skip all three. It costs you later — a workspace that knows nothing about your firm has nothing to check a requirement against. ## Minute 3: the tour, and the crowns The guided tour offers itself once, automatically, the first time you reach the dashboard. It is seventeen steps and it drives the real application, not a video: real routes, real screens, on a demo submission for a fictional roof-replacement tender. It never touches your data and never appears in your submission list. You can leave it and pick it up later from the sparkles button in the help assistant. Four of those seventeen steps carry a crown badge marked **Business** — starting a submission from the opportunity radar, win intelligence, the AI-drafted answer inside a task, and the deep dive. They are demonstrated as working regardless of what you pay for, which is the tour's job, but it means the trial deserves a plain statement: **the 14-day free trial runs on Professional**, not Business. If that is how you arrived, four of the seventeen steps show you capabilities your workspace does not currently have. Better to read that here than to discover it on day two. ## Minutes 4–6: upload a real tender **New submission** in the sidebar opens two drop zones. The left one, **Tender documents**, is required — everything the buyer published, PDF, Word or Excel, in any number. You do not need to sort or rename them; identifying what each file is (instructions to bidders, forms, technical specifications, a bid form) is part of the analysis. The right one, **Your submission**, is optional and easy to under-use. It takes your estimate, your partly filled forms, a draft response. Adding it now means the very first analysis comes back with coverage and gaps rather than only a requirement list. There is also an instructions box for steering the AI on this specific tender — *this RFP forbids subcontracting, emphasise our in-house crews*. Then the analysis runs, and takes five to ten minutes. It reads every file, extracts requirements, key dates and evaluation criteria, builds the coverage matrix and cross-checks your knowledge base. You do not have to watch it; you are notified when it lands. Which is the honest answer to what the first ten minutes hold — a decent share of them is waiting. ## Minutes 6–10: read what came back **Overview** opens on a readiness count and what is still outstanding, with every section reachable from there. **Documents to produce** sits above the task list and is the section most worth understanding early. It regroups scattered requirements by *deliverable*, and tags each one with what it actually demands of you: **Fill in** a form the tender provided, work **From template**, **Create** a document from nothing, **Provide** something you already hold, or **Obtain** one a third party must issue — a surety, a registry, a laboratory. That last tag is the one that saves bids, because obtaining takes days you have not budgeted. Each document shows its own progress and a **Only this document** switch to narrow the task list to it. This is on every plan. Below that, tasks in three groups: **Needs correction**, **To complete**, **Completed**. On a first visit only *Needs correction* is expanded, on the reasoning that it is the one with consequences. Open any task and you get the requirement, and its source. The source drawer has two tabs — the rendered page with the passage highlighted, and the extracted text the AI actually read — plus a confidence badge that will tell you when it could only find an approximate match rather than the exact clause. A tool that shows you where it is unsure is more useful than one that hides it. **Key dates** and **Important legal notes** are the last two worth opening on day one: every deadline pulled out of the documents with the submission date flagged, and the bonding, warranty and risk-transfer clauses ranked by how much they should affect your decision, each linked back to its source page. ## Files, after the first pass File handling is built for the fact that tenders change. Submission files keep version history, download individually or as a zip, and deleting one from an already-analysed submission raises a **Re-run analysis** banner rather than silently leaving stale results behind — the previous analysis is versioned, not lost. Excel files get a per-sheet breakdown so you can see what Offra made of a workbook. The **Knowledge base** has three tabs: entries, company documents and past submissions. The middle one matters more than it sounds — company documents are the only pool the analysis can link directly to a task, so your insurance certificate, licence and proof of incorporation belong there rather than in a folder on someone's desktop. Offra reads expiry dates off them. And every past submission you import makes the next tender's answers better, which is the compounding part. ## What is Professional, and what is Business On **every plan**: upload and full analysis, documents to produce, the task board, source traceability, key dates, legal notes, the whole knowledge base including company documents, and the guided tour. **Business adds** the capabilities that read the market and your own record — the opportunity radar that surfaces open public tenders ranked for your firm, win intelligence with a win-chance verdict and pricing band, the grounded suggested answer inside a task, the deep dive that researches a single task and drafts a full response with citations, and automatic linking to the public notice, which is what makes addendum tracking automatic rather than manual. Professional is sized for three tenders a month and three users; Business for twenty and ten. The current numbers and prices live on the [pricing page](/pricing) — and worth knowing that a workspace cap can be raised without moving tiers if the only thing pinching is volume. ## After the first ten minutes You end with a workspace that knows your firm, one tender read end to end, and a specific list of what is still outstanding with a source behind every line. What you do not end with is a finished bid, and Offra is deliberate about that boundary: it reads, extracts, cross-checks and flags, and the decisions — what to bid, what to price, what to sign — stay with your team. Related reading: [making the bid/no-bid call in twenty minutes](/blog/bid-no-bid-decision-in-twenty-minutes).