What's new.
New features and improvements in Brieful, newest first.
- New
⬇ Backlog export for Azure DevOps and Jira
The document view has a Backlog menu: export Process → Use case → Requirement as an Azure DevOps CSV (Epic → Feature → User Story in one import), a Jira CSV (Epic → Story with process as Component and phase as Fix Version), or a traceability matrix workbook with screens and test cases per requirement. MoSCoW priorities map across; every item keeps its Brieful id.
Learn more → - New
📄 Portals section in the requirements document
Dual-portal projects now get a Portals section in the generated requirements document: the role-to-portal table, deployment topology, sign-in for each portal, design system per portal, and field-exposure counts. It sits just before the Navigation and Shell section and appears on the next regenerate. The field-by-field exposure list downloads from the Portals panel as an Excel workbook (Roles, Field exposure, Settings, Findings).
Learn more → - Improved
🌐 Portals: merge roles, fix all findings at once, or move the one role behind them
Each role row has 'merge' to fold it into another role on every screen and use case, and one click merges every bracketed variant such as 'Analyst (read-only)' into its base role. In-place fixes now know which roles are public and which are staff, the panel says when a finding is still open, a hint offers to move a mis-filed role, and 'Fix all' applies every open finding in one AI pass per use case.
Learn more → - Improved
❓ Open questions are asked, not filed as requirements
When an answer leaves something for you to decide, the interview now asks it in the next round instead of adding an 'Open Question' card to the requirements. Existing open-question cards have an 'Answer this' action: your answer goes through the interview, becomes requirements, and the card is resolved.
Learn more → - Fixed
🛠 Interview rounds with long answers no longer fail
Answering a full round of questions with detailed, multi-option answers could fail with 'Requirement extraction was cut off', which lost the round and asked the same areas again. Each answer is now read on its own, so long rounds complete and yield more requirements.
Learn more → - Fixed
🪙 A paid seat's credits are spent in the workspace that paid for it
If you hold seats in two paid workspaces, you now get a full monthly allowance in each, instead of one allowance counted across both. Each workspace pays for its own seat, so each seat funds its own credits. The free trial is unchanged: one allowance per account. Brieful staff invited into a workspace for support no longer count as a seat; an affected subscription is corrected the next time Billing is opened.
Learn more → - Improved
⚡ Auto run fills your response forms
A new Fill response forms stage, off by default, fills every uploaded agency form in one go: price forms from the price sheet, requirement matrices against the scope. A form whose input is missing is skipped with a note. Confirming each line stays with you on the Response forms tab.
Learn more → - New
💲 The agency's price form, filled from your price sheet
Upload the agency's pricing workbook on the Response forms tab as a price form. BidWin finds its lines and year columns, funds each line from your price sheet (labor by phase, non-labor, annual subscription by year, totals) and computes the cells, never typing a number itself. Review the mapping, override a source or a figure, confirm each line, and export the agency's own workbook with the figures in place. Lines nothing on your sheet covers are flagged for you.
Learn more → - Improved
📋 Response forms get their own tab
The agency's fill-in workbooks moved out of Compliance into a Response forms tab between Proposal and Demo, where they read as a deliverable in their own right. Nothing about filling or exporting them changed.
Learn more → - Improved
💲 Multi-year pricing, and summaries written last
The price sheet has a Recurring (annual) block for subscription, hosting and support, with the years taken from the contract term, an escalation rate, a year-by-year table and the total cost of ownership, all exported to Excel and quoted in pricing tabs. The cover letter and executive summary are now drafted after everything they summarize, and the Auto run can redraft existing sections to bring an older proposal onto the current price and facts.
Learn more → - Improved
✍️ RFP tabs drafted item by item, with a Needs-your-input list
A proposal tab that must answer numbered scope items or requirements is now planned from the RFP and drafted a few items at a time under the RFP's own numbering, so nothing is cut off or skipped. Where your requirements, architecture or Company Library do not cover an item, the draft leaves a clear 'Respondent to add' slot instead of a guess. The Proposal tab lists every slot with a jump link, Word export warns while any remain, and an applied red-team fix now says so until you re-score.
Learn more → - New
🏢 Company profile, and proposals that quote your price and facts
The Company Library now holds a company profile: legal entity, signatory, insurance, registrations, certifications and audits. Cover letters, forms and compliance answers draft from it, and Qualify can clear insurance and registration gates against it. The Auto run prices before it drafts so pricing tabs carry real figures, long technical tabs no longer stop mid-sentence, every numbered RFP item gets a response, and a new Red team fix stage applies the evaluator's instructions and re-scores.
Learn more → - New
⚡ BidWin Auto run: from Parse to Red team in one go
Upload the package as usual, then press Auto on the bid. Tick the stages to run, from Parse through Qualify, Scope, Requirements, Map screens, Strategy, Architecture, Proposal, Delivery plan, Price sheet and Red team, and choose per stage whether the run stops for your review or goes ahead with the recommendation. Everything lands in the normal tabs; what the run accepted for you is marked AI-proposed until you make it yours. Prototypes and approval stay manual.
Learn more → - New
🧰 Build kits run on OpenAI Codex as well as Claude Code
The BuildAI kit now ships its project brief as AGENTS.md, which Codex, Cursor and most coding agents read natively, with a one-line CLAUDE.md so Claude Code reads the same file. The start-here guide carries a quick start for each agent, and newly written playbooks address the agent neutrally.
Learn more → - Improved
🔐 Workspace access tightened
Only workspace owners can grant the owner or admin role, and invitations from BidWin count against your seats. Someone removed from a workspace no longer keeps the projects they created there, cannot export them, and deleting their own account passes team projects to an owner instead of removing them. Replacing an existing authenticator now asks for your password and a current code.
- New
🗺️ One-page Business Analyst cheat sheet
The Run-a-project guide now opens with a printable poster of the whole journey: ten stations from brief to tested requirements, the button for each, what to do, and what to check before moving on. Download it as PDF or PNG from the guide.
Learn more → - Improved
🧪 Free trial is one-time per account
The free trial's AI allowance is now a single allowance for your account rather than a monthly one, and it does not refill by creating another workspace. People whose access comes through their organisation's pilot or agreement work in that workspace; the trial is not available alongside it. Everything you built stays, and upgrading to Brieful keeps you going.
- Improved
✍️ House style: no em dashes
Everything Brieful writes for you now follows one punctuation rule: no em dashes. The site, the app, emails, and every AI-generated document, specification, test case, prototype and proposal use commas, colons and full stops instead. Documents generated earlier keep their original text until you regenerate them.
- New
📜 Subscription Terms
Brieful's Subscription Terms are now published at /terms. New accounts confirm them when registering, existing accounts confirm once on their next visit, and Stripe asks for the same confirmation at checkout. Billing and Pricing link to the Fees and Payment section.
Learn more → - Improved
🧠 Opus 5.5 for prototypes, documents and readiness; Sonnet 5.5 for extraction
Prototype builds, final documents, readiness reviews and the heavy BidWin steps now run on Claude Opus 5.5; process and requirement extraction, screen maps and flows on Claude Sonnet 5.5. Same workflows, lower per-call prices (cache hits are now billed at 5% of input). Usage reports and seat allowances reflect the new prices.
- Improved
⚡ Faster, cheaper everyday AI: Haiku 5.5 and Sonnet 5
Interview questions, per-answer requirement capture and document classification now run on Claude Haiku 5.5; process and use-case extraction, requirement reading, screen maps, process flows and most BidWin generators run on Claude Sonnet 5. Same workflows and outputs, lower per-call cost (your usage reports and seat allowances reflect the new prices). Final documents, prototype builds and readiness reviews stay on Opus.
- New
🖥️ Screens show which layout pattern the next build will follow
Every screen in the Screens view now carries a layout-archetype badge: List, Form, Upload, Dashboard or Detail, and says whether the next prototype build follows a house pattern from your Design Library or Brieful's default layout. The archetype is read from the screen's name and purpose; pick a different one on the badge to change which pattern applies. The project Design panel lists the patterns in effect per screen type with how many screens use each, and Design Library cards now say whether a screenshot's structure was captured or only its description.
Learn more → - Fixed
🎬 Movie mode: the recorded voice plays as soon as it is ready
After you publish a movie the narration is recorded in the background. The storyboard used to keep playing with the browser's built-in voice, which some browsers keep silent, until you closed and reopened it. It now waits for the recording, tells you which voice ▶ Play will use, and switches to the recorded voice by itself. The player also says so when a voice cannot play, instead of running silently.
Learn more → - Fixed
🔎 Readiness: reviews stay on what you picked, and a repeat pass stops piling up
A review now runs only on the use cases you selected, it no longer picks up unfinished reviews of other use cases left over from an earlier attempt. Re-reviewing a use case that has not changed is now a blockers-only check, because each full pass was returning the next tier of smaller points and the list never stopped growing. And the same gap described in different words on a later pass is recognised instead of being added again.
Learn more → - Improved
📄 Requirements documents: exact tables, much faster on large projects
The requirements table and each screen's fields, actions and rules now come straight from your project data instead of being rewritten by the AI, so every row matches what you captured, word for word. The AI writes the overview, actors and main flow. Sections are also generated four at a time. A large document that took hours now takes minutes, and costs far less to produce.
Learn more → - Fixed
📄 Requirements documents hold up on very large projects
Three fixes for projects with hundreds of screens and over a thousand requirements. A use case section that outgrows the length limit is now continued instead of being cut off mid-table, and is clearly marked if it still cannot be completed. The closing sections (non-functional requirements, business rules, constraints, open questions) are written per category in batches when the list is long. And delivery use cases are no longer given screen maps while the document is prepared.
Learn more → - New
🔎 Readiness: accept blockers as assumptions, and export them for the proposal
On a bid you cannot get hundreds of answers from an agency before the deadline. Filter the Readiness list to 'block build' and you can now accept every blocker that has a recommended default as an assumption in one confirmed step, each records its own default and your name. Items with no possible default (a value the client owes) stay open. The new '⬇ Assumptions' export produces a Word document with the assumptions the project is priced on and the questions still open for the agency, ready for the proposal.
Learn more → - Improved
🔎 Readiness Review no longer needs test cases first
A use case can now be reviewed as soon as it is complete with screens and a process flow. Test cases are generated from the specification, so requiring them first meant paying for tests the review then made out of date. Run Readiness, answer the clarifications, then generate test cases and prototypes from the corrected spec. If tests already exist they are read as context only, a value that appears only in a test is still raised for you to confirm.
Learn more → - Fixed
📦 Process flows are no longer drawn for delivery work
A process made up only of delivery use cases, a project plan, training, data migration, documentation, has no system flow, but 'Map all flows' drew a swimlane for it anyway. Those processes are now skipped and marked 'No flow needed, delivery work', and in a mixed process only the feature use cases feed the flow. Flows drawn earlier for delivery-only processes are kept but left out of the view, the document and the build kit.
Learn more → - Fixed
📥 Large requirement documents import reliably; requirements finalize in batches
Importing a long requirements document (a full Business Requirements Document, a big user-story workbook) used to time out with 'Network error'. It is now read in parts with a 'Parsing part N of M' counter, up to about 300 pages per file. Separately, the step that consolidates auto-generated requirements and tags their complexity now runs per use case in batches, so large RFP scopes get every requirement finalized and sized instead of falling back to the raw list, and the progress line shows which batch is running.
Learn more → - Improved
📖 BidWin: large RFP packages are read in full, and 'Read the rest' for existing projects
Auto-scope now reads far larger packages (roughly 1,400 pages for processes and 1,500 for requirements), and when a cut is still unavoidable it reads the scope-bearing documents first, requirements, statement of work, main RFP, before plans and terms. A project scoped earlier with documents only partially read gets a 'Read the rest' button on the Scope tab: it extracts requirements from the unread parts only and adds them to the existing project, keeping every use case, requirement and edit.
Learn more → - Improved
📚 BidWin auto-scope reads every package document in full
Auto-scope used to join the whole RFP package into one text and cut it at about 60 pages, so on a large package the Q&A, Addenda and Price Sheet could drop off the end without a word. Each document is now read file by file (up to roughly 500 pages for processes, 250 for requirements), a budget cut trims the tail of every long file evenly, and the Scope tab says 'All N documents read in full' or names what was only partially read.
Learn more → - New
📊 BidWin: Excel and CSV in the RFP package
Pricing schedules, requirements matrices and Q&A logs often arrive as workbooks. The RFP package upload now accepts .xlsx (each sheet read as a table) and .csv alongside PDF, Word, text and images, so they feed the parse, compliance scan and auto-scope like any other document. Legacy .xls files should be saved as .xlsx first.
Learn more → - Improved
🔍 Readiness Review: 'not reviewed' is now visible
A bid whose scope was never put through Readiness Review used to look identical to one reviewed and clean. Now the bid Overview shows an amber 'Readiness review not run' criterion, the Phases card shows coverage per phase, the BuildAI page warns above the kit download, and the kit's CLARIFICATIONS.md opens with a coverage notice. It is a warning, not a gate, nothing new is blocked.
Learn more → - New
🏗️ BidWin Architecture: decision register + deeper synthesis
The synthesis overview now comes in three depths: Executive, Standard, or Detailed with headed subsections tied to the named decisions, and a collapsible Decision Register under it lists every accepted decision exactly as recorded (why, alternatives set aside, impact, client's option and date). Both are in the Solution Architecture Word export, the register right after the overview. The 3-year TCO figure on option cards is gone; pricing lives in the Price tab.
Learn more → - Improved
📋 Dashboard: projects linked to a bid stay visible
Linking a project to a BidWin Project-mode bid used to remove it from the workspace dashboard, so a workspace owner lost sight of a project they could still open. The dashboard now lists these under 'Linked to a Build' with an Open button (and Open build for BidWin users). RFP bids stay inside BidWin.
Learn more → - Improved
🧭 Shell panel: a guided, step-by-step mode
The Shell panel now opens in a guided mode: five steps: Propose, Check the menu, Who sees what, First page per role, Preview & approve, each showing only what it needs, with a one-sentence explanation, a 'What's left' list with one-click fixes, and a tick when done. The landing step is simpler: choose Dashboard or Marketing from two pictures, keep the cards you want, and one 'Build all landing pages' button generates every role's page (and picks a featured list where there is none); the finer settings sit under 'Customize this page'. Nothing was removed, click Advanced for the full editor, and the panel remembers your choice.
Learn more → - New
🏠 Landing pages are now generated, not templated
The page every role sees first used to be a fixed template that ignored your Design Library. Now each role's landing page is built through the same pipeline as every other screen: pick Marketing (a public front page) or Dashboard (a signed-in work surface), choose the sections, set a headline and tagline, embed a featured list from one of the role's screens, add a hero image the client supplies, and click Build. It follows your Design Library, including a new 'Landing / Home page' pattern you can teach from a screenshot, passes the fidelity scan, and takes one-line refinements. Links can only go where that role's menu goes. A stale badge tells you when to rebuild, the build kit describes each landing page for the developer, and projects that never build one keep the simple template.
Learn more → - New
🌙 Dark mode in the Design Library
A new Mode switch (Light / Dark) on the workspace library, on project and portal design forks, and on admin templates. Dark gives a near-black page with the brand color as a glowing accent, and everything else is derived from that one switch: raised surfaces for cards, inputs and menus, reversed color ramps, dark status colors, flat elevation, and text on the action color chosen to pass contrast (a neon accent gets dark text). Contrast stays WCAG AA by construction. Prototypes, the navigation shell and landing page, the live preview, the fidelity scanner and the build kit all follow the mode, so a built system can be restyled dark through a change kit. A dark reference screenshot sets the mode for you, and 'Dark: Consumer app' is a one-click template. Existing designs are unchanged: no mode means light, exactly as before.
Learn more → - Fixed
🏛️ Design templates no longer wipe your screen patterns
Clicking Use on a Design Library template used to replace the whole library, even the template already in use, so List, Form and other screen patterns you had built from screenshots disappeared. Use now asks how to apply it: 'Apply look' takes the template's colours, fonts, corners and type scale and keeps your screen patterns, screenshots, navigation choices and logo; 'Replace everything' starts over. The template you applied last is marked In use, and the page warns before you leave or apply a template with unsaved changes.
Learn more → - Fixed
🎨 Project design forks no longer vanish after saving
A per-project design fork saved from the 🎨 Design panel (for example, starting from the Modern citizen app template) could disappear the next time the project page autosaved, because the page wrote the project back without it. Design forks, portal design forks and the portal configuration are now owned by the server and can never be overwritten by a page save. If you lost a fork, apply it again once and it will stay.
Learn more → - Improved
🧭 Shell: any workspace member can approve
The navigation & shell is the analysts' working artefact, so approving it no longer needs a workspace owner or admin, any member of the workspace (or the creator of a personal project) can approve. Phases keep their stricter owner/admin gate.
Learn more → - Improved
🧭 Shell: 'off-menu on purpose', and why approval is disabled
Sign-in, reset password, own profile and actions opened from a button on another screen have no menu item by design, until now the Shell panel warned about each one forever. You can now mark a use case 'off-menu on purpose' (one click per warning, or all at once): the warning and the readiness finding go away, the document lists it under 'Reached without a menu item', and NAVIGATION.md tells the builder not to invent navigation for it. ⇪ Import as-built marks them automatically, since a built system is the truth, and drops any landing card a role could not see instead of leaving a blocking error. Blocking issues now sort first in red with a count, and the disabled Approve button says exactly what blocks it. Roles are reconciled too: after an import the notice names roles the built app uses that the project does not know, and project roles the built app gives no menu; each is a warning with a one-click fix (rename to a project role, remove the role, or copy another role's menu).
Learn more → - Improved
🧭 Project header: outputs on top, tools below
The project header's top row now carries the outputs: Training, Draft doc (or Document), History, and the second row keeps the working tools: Phases, Process, Screens, Prototype, Tests, Meeting, Import requirements, Readiness, Portals, Shell and the requirements count. Buttons are a little tighter so every tool fits on one row. 'Review readiness' is now simply 'Readiness'.
Learn more → - New
🧭 Navigation & application shell, approve the menu before the build
Until now every use case was approved on its own and the global menu, the landing page for each role and who sees what were decided at build time. The new 🧭 Shell panel derives all of that from your processes, use cases, screen maps and portals, lets you rename, regroup, reorder and set a landing dashboard per role, shows a live preview per role, and records an approval as a gate. Once approved, the shell is everywhere the client sees the system: every prototype renders inside it (in the app, in Present mode, in shared links and in the offline file) and its menu opens other use cases; each role gets a landing page; the shell follows the Design Library, its Navigation pattern's house choices (layout, accordion or open groups, a rail that collapses to icons, a “More” menu for long bars, icon modes) come first, its colours, font, logo and icon set paint the bar, and every choice shows where it came from and can be changed per project; 'Play all' opens on the landing page before the first use case; the requirements document gains a 'Navigation and application shell' section; the Readiness Review gains a Navigation lens; and the build kit ships NAVIGATION.md, design/navigation.json, design/shell.html and a gated shell step that generates the navigation allowlist test, with change kits and the reconcile pass tracking navigation drift. Nothing changes for projects without an approved shell.
Learn more → - Fixed
💾 Saves on large projects, and a warning when a save fails
On the biggest projects, changes made on the project page could stop reaching the server without any warning: the page sent the entire project on every save, and past about 4.5 MB the save was rejected silently. Readiness clarifications, screen maps and test cases now stay on the server instead of travelling with every save, which cuts the save to a fraction of its size. If a save does fail you now see a red message telling you the change was not saved. Phases can only be changed from the Scope Plan, so an out-of-date tab can no longer undo a phase change.
Learn more → - Fixed
🔒 Completed phases are locked
Nothing can enter a phase you have marked completed, not a new use case, not a new requirement, and not an item whose phase was changed. Anything that would land there goes back where it was, or into your working phase if it is new. This also fixes a bug where answering interview questions could silently move Phase-2 use cases into a completed Phase 1.
Learn more → - Improved
💳 Build kits are paid by card at purchase
Build-kit licences are now paid by card at the point of purchase, and the kit unlocks as soon as payment is confirmed, replacing the earlier invoice-with-terms arrangement announced in June. Pricing is unchanged: once per bid, by the number of approved requirements, with a top-up for the difference if scope grows later. Purchase orders and invoicing remain available for larger engagements, arranged as part of the engagement.
Learn more → - New
🖼️ Design templates grounded in open design systems
The template gallery gains Federal (exact U.S. Web Design System values) and Modern citizen app (Material 3 values), and the Enterprise / Data-dense and Clean SaaS templates have been re-authored on exact IBM Carbon and Radix values, all openly licensed, all deriving a full accessibility-safe design system. Pick one for your workspace, or use them as seeds for per-project and per-portal design forks.
Learn more → - New
🎨 Design Library v3, professional screens from the same two colors
Prototypes now look designed, not generated. Your brand colors expand automatically into a complete design system, real typographic hierarchy, consistent spacing, status colors that pass accessibility contrast by construction, a proper icon set instead of emoji, and page layouts that follow per-screen-type structure. Nothing changes in how you author: pick a template, set two colors, done. Existing projects pick the new look up on their next prototype build. Bonus: import an agency's brand file directly (Figma variables, W3C design tokens, or Tokens Studio JSON) and the whole system derives from it.
Learn more → - New
🌐 Dual portals, public and agency staff, from one spec
Agencies that need a separate public portal and a staff portal (signed in with their Active Directory) can now capture that in one click: assign every role a portal, and screens, field exposure, validation and the build kit all follow. A 16-question agency questionnaire lands in Clarifications with recommended defaults, screens get portal badges with in-place fixes for any that span both, and build kits ship portal-aware prompts, enterprise SSO, deployment topology, and generated isolation tests included.
Learn more → - New
🔎 Readiness review, find the gaps before the build
One button reviews each finished use case the way a builder would read it, flow, screens, fields & data, business rules, and returns numbered clarifications: the question to ask your client plus the default we'd otherwise assume. Record decisions (or turn them straight into requirements), export the open questions as a client-ready Word document, and see "N clarifications block build" on the phase gate before a build kit ever ships a guess.
Learn more → - New
🔍 Guided search (add-on module)
Give the system you deliver a search box that answers from the client's own data, never from a model. The searchable catalogue derives from your approved screens, sensitive fields stay out by default, every query is permission-scoped and cost-bounded, and the generated SRCH requirements import into your project so search is tested and traceable like everything else.
Learn more → - New
🏗️ Ask us to build it for you
BidWin (bids & proposals) and BuildAI (the built system) are now service engagements rather than plans you buy yourself, our team does the work and you review it at every gate. Tell us what you need from the Contact page and we'll come back with how we'd approach it. Brieful itself is unchanged: still $49 per seat, still yours to run.
Learn more → - Improved
👥 A seat added mid-month is credited for what it paid
Adding a member part-way through a billing period is charged pro rata for the days remaining, and their AI-usage allowance is now prorated to match, so the credits track the money, the same way a mid-cycle plan upgrade already did. Their full monthly allowance starts at your next renewal. Members who joined before the period began are unaffected.
Learn more → - Improved
🔁 Change plan from Billing, with the price shown before you pay
Switching between Brieful and Brieful + BidWin now happens in Billing rather than the Stripe portal, and always shows the exact amount first. Moving up takes effect immediately, charges the prorated difference, and adds credits in the same proportion, so upgrading late in a cycle costs less and adds proportionally fewer credits, with the full allowance arriving at your next renewal. Moving down takes effect at your next renewal instead: you keep the higher plan and its full allowance for the period you have already paid for, and nothing is charged or refunded. Cancelled a subscription and changed your mind? A “Keep my plan” button brings it back without starting a second one. Any billing credit on your account is now shown on the page too.
Learn more → - Improved
💳 Billing now leads with what you have left
The Billing page has been rebuilt around the question people actually arrive with. It opens with your remaining credits rather than your spend, and one bar shows what you've used, what's left in this month's allowance, and any purchased credits held in reserve behind it. Credit packs now say what they buy before you click, and a new Credit activity list shows exactly where purchased credits went.
Learn more → - New
⚡ Buy extra credits when you run out mid-month
Hit your monthly AI budget in the middle of a deadline? Owners and admins can now buy a credit pack ($25 / $50 / $100 / $200) from Billing, charged once to your card and usable immediately. Purchased credits never expire and are used only after your monthly allowance, so nothing you pay for disappears at the end of a cycle. The usage meter shows your reserve alongside the monthly budget.
Learn more → - New
💳 Plans, billing and BuildAI kit licensing
Paid plans can now be started and managed from Billing in your workspace, monthly per seat, cancel anytime, with your AI-usage budget following your billing cycle. BuildAI build kits are licensed per bid and priced by the number of approved requirements they contain, fixed when the kit is first generated; if scope grows later, only the difference is invoiced. Kit licences are invoiced with payment terms rather than charged to a card, so agency buyers can pay by PO or transfer.
Learn more → - New
📐 Build kits now ship the technical documentation package
Every BuildAI kit now includes delivery-grade technical documentation, assembled from work you already did: a Solution Architecture Document with C4 diagrams, an Architecture Decision Record per workshop decision (options-not-taken included), interface control sheets, a mandated-vs-designed service-level document, and the intended data model. Two new build-session prompts complete it against the real code: as-built API/schema/operations docs, and a nine-area, review-gated security assessment with a government-readiness summary.
Learn more → - New
🔄 Living delivery, change kits, reconcile & drift
Delivered systems now stay synchronized with your specification instead of drifting from it. Every build kit carries a version manifest; when requirements change after delivery, download a Δ change kit containing only what moved, with CHANGES.md, plain-English release notes for end users, and a regression gate. A reconcile prompt in every kit makes the delivered code report its own drift back for your review, and a drift card on the BuildAI page (plus stale flags on training tasks) tells you at a glance whether what you shipped still matches what you specified.
Learn more → - New
📕 Role guides & quick-reference cards
Approved training tasks now assemble into deliverables: a per-role Word guide (getting-started with your site facts, every procedure with its traceability line, the status glossary) and a 1–2 page quick-reference card. The build kit also ships the approved tasks with instructions for expanding them into full guides against the real code, so delivered training can honestly claim to reflect the deployed system.
Learn more → - New
🎓 Training view, the task library
A new Training view turns each use case’s positive test cases into user procedures, plain-language steps bound to real screens, traceable to requirements and test cases. Every screen and button a task names is automatically verified against the screen map before you review it, and only tasks you approve feed the role guides and quick-reference cards coming next.
Learn more → - New
🎓 Training deliverables: SCORM modules, printable walkthroughs, roles & statuses
Your published narrated walkthrough is now three deliverables from one approved source: play it, package it as a SCORM 1.2 module for the agency’s LMS, or export it as a Word step-by-step procedure with a field reference. Plus a roles & statuses glossary generated from your domain pack, with anything used-but-undefined flagged instead of invented.
Learn more → - Improved
🏛️ Design Library v2, enforced brand fidelity, per-project design & government templates
Prototypes now carry your design system as real CSS baked into every build, and each build is verified, off-brand colors and fonts are detected and auto-corrected. Upload one screenshot and Brieful reads your whole theme from the pixels (including the neutrals humans get wrong). New: logo/seal upload with a consistent branded header on every screen, per-project design forks for client work, WCAG AA contrast warnings, and built-in Government templates with the official-banner and accessibility conventions ready to go.
Learn more → - New
🎬 Movie mode, narrated, self-running prototype demos
Every prototype can now play like a product film: generate a screen-by-screen storyboard (review and edit every line, then publish), and ▶ Play walks the prototype by itself, screens advance, a voice narrates, a spotlight highlights what's being discussed, captions always on. 🎬 Play all chains every use case into one continuous demo, published movies play on share links, and ⬇ Demo HTML exports a standalone self-narrating file that works offline.
Learn more → - Improved
Generate all test cases in one click
The Test Cases view now has a “🧪 Generate all” button that covers every use case whose interview is finished, no more generating them one at a time. They run in the background and appear as each finishes, and the run stops cleanly if you hit your usage limit.
Learn more → - New
Phases, trim a big spec into a buildable Phase 1, grow with Phase 2+
The new 🧩 Phases panel organizes any project into phases: ✂️ AI trim proposes the minimum coherent Phase 1 and moves the rest to Phase 2 (nothing deleted, later phases show greyed as “Next phase”). Prototypes, screens, tests, and the document are phase-aware: build lazily per phase, later-phase prototypes EXTEND the approved earlier ones instead of rebuilding, and the document presents Phase 1 as the spec with later phases clearly marked. Completed phases lock; deleting a phase asks merge-vs-remove.
Learn more → - Improved
Map all screens & flows now runs in the background
“Map screens for all” and “Map all flows” no longer need the page kept open, they run in the background, so you can leave and come back. Progress shows on the button, finished maps appear as they land, and if anything is interrupted your completed maps are saved (just click again to finish the rest).
Learn more → - New
Step-by-step “Run a project” guide
A new 10-stage walkthrough (Guide in the top nav) takes you from a document or an idea to a build-ready spec, prototypes, tests, and an exported document, each stage shows what you do, what Brieful does, and how you know you’re ready for the next. The feature-by-feature reference is still there as “Help”.
Learn more → - New
Confirm scope, then choose: interview or auto-generate
After your processes and use cases are extracted, you now review and confirm them first, then choose your path: dive into the guided interview, or (for projects with a source document) auto-generate requirements for every confirmed use case in one step. The same unified scope engine powers core projects and BidWin bids.
Learn more → - New
Delivery use cases, work products, priced but not prototyped
Use cases now carry a kind: Feature (software behavior) or Delivery (work products and services like project plans, design documentation, training, or migration). Delivery use cases are tagged automatically when scope is generated from documents, flip any with the 📦/🖥 chip. They keep their requirements and count in estimates, but prototype building skips them: no more prototyping a project plan.
Learn more → - New
Watch your prototypes build live
Click Build all and a full-screen live view opens: every use case's screens appear on the canvas as the AI builds them, with a running screen count and timer, finishing with a build summary. You can close it anytime and reopen with the 🎬 Watch button.
Learn more → - New
Napkin mode, sketch to prototype
Snap a photo of a whiteboard, napkin sketch, or hand-drawn wireframe and upload it when starting a project. Brieful now reads the design, boxes become screens, arrows become flows, margin notes become rules, and turns your sketch into structured requirements and a clickable prototype.
Learn more → - New
Share prototypes with a link
Send anyone a clickable prototype, no login needed. Click 🔗 Share in the prototype view to create a time-limited public link (24 hours, 7 or 30 days), see how many times it was viewed, and turn it off anytime. The shared page carries your copyright and a confidentiality notice.
Learn more → - Improved
Edit screens without starting over
The Screens & Flows view now lets you make a targeted change, add a field, change a screen, adjust a rule, with a plain-English instruction, keeping the rest of the spec intact. Each use case expands/collapses, shows its requirement coverage, and screen mapping is available once a use case's interview is finished.
Learn more → - Improved
Much better prototypes
Prototype screens are now generated by our most capable AI model, with an automatic quality check that catches and repairs broken screens before you see them. Informational pages (FAQ, landing, dashboards) render as real content instead of forms, each screen shows which requirements it covers, and key actions in a prototype now actually work in demos, submit a form and watch the list update.
Learn more → - Improved
Better Word document exports
Your requirements document now exports to Word with process and screen-flow diagrams embedded as clean images (they used to sometimes appear as code). Large documents, hundreds of pages with lots of diagrams, export much faster now, with a live progress indicator so you can see it working.
Learn more → - New
Two-factor authentication (2FA)
Protect your account with an authenticator app: set up 2FA from Account → Two-factor authentication (QR enrollment + one-time recovery codes). Workspace owners can require 2FA for all members, and platform admins are always required to use it.
Learn more → - New
Build all prototypes in one click
In the Prototype view, use '🎨 Build all' to generate prototypes for every use case that doesn't have one yet, instead of building them one at a time. They build in the background and fill in as they finish.
Learn more → - New
Bring your existing requirements into Brieful
Joining a project midway? Use '⬆ Import requirements' to paste or upload requirements you already have (Word, PDF, Excel, CSV, or Markdown): Brieful maps them into processes and use cases (preserving your IDs and priorities), you confirm where each lands or merge duplicates, and it then only interviews the gaps.
Learn more → - New
Draft requirements for a use case in one click
Any use case that has no requirements yet now shows a '✨ Draft requirements' button: Brieful drafts its core requirements from your uploaded document or notes so you don't start that use case from scratch, then you refine them in the interview. The use-case list also collapses/expands per use case to stay scannable.
Learn more → - New
Browse the domain pack library
A new Domains page lists every available domain knowledge pack, grouped by industry with search and filtering, so you can see which fields Brieful already covers as a built-in subject-matter expert. The library is growing across industries.
Learn more → - New
Domain packs: expert-verified vs AI-generated
Every domain pack now shows a trust level, '✅ Expert-verified' (reviewed by a person) or '🤖 AI-generated · beta' (a fast AI-built kickstart, not yet expert-validated). You see it when picking a pack, with a reminder to confirm AI-generated suggestions.
Learn more → - Improved
Domain packs: seeded requirements + per-pack release
Starting a project from a domain pack now offers the pack's default requirements and domain assumptions as one-click confirm items (the '💡 Domain' button). And each pack can be released to users individually, kept Internal for testing, then made Public when ready.
Learn more → - New
Domain knowledge packs
Start a project from a curated domain pack (e.g. Professional & Occupational Licensing) and Brieful asks expert questions and proposes the typical processes and use cases for that field, you confirm everything. Packs are authored and approved in the new Knowledge Studio.
Learn more → - Improved
Zoom and pan process diagrams
Process and flow diagrams now have an Expand button that opens a full-screen view at readable size, scroll or drag to move around, Ctrl/⌘+scroll to zoom, so large swimlanes are easy to follow instead of being squeezed to fit.
Learn more → - New
Project history & baselines
Every project now has a History panel: an activity timeline of who did what, and Baselines, named snapshots of your requirements you can compare against the live project and restore at any time. Great for client sign-off and tracking change.
Learn more → - New
See your AI usage as you work
A live usage meter now sits in your dashboard and project headers, showing how much of your monthly budget you've used. It warns you as you approach the limit, and each project shows its own approximate credit cost.
Learn more → - New
Choose how much detail you need
Start a project in Presales demo, MVP, or Full spec mode, the interview adapts its depth so a quick demo isn't buried in the same questions as a full build-ready spec. Change the scope anytime from the project header.
Learn more → - Improved
Discard & rebuild a prototype
Regenerate a prototype from scratch in one click when you'd rather start over than refine, your requirements stay untouched.
Learn more → - New
See how to navigate a prototype
Presentation mode now shows what's clickable: toggle highlighted hotspots on every screen, follow a Next steps bar to jump between screens, and hover a step to flash the link that takes you there.
Learn more → - New
Sync requirements from a meeting
Paste or upload a Teams/Zoom transcript and Brieful reconciles it against your requirements, proposing confirmations, edits, additions, and removals for you to review, each backed by a quote from the discussion.
Learn more → - Improved
Large projects build reliably
Prototypes, test cases, and the requirements document now build in the background section-by-section, so even very large use cases finish without timing out, however many requirements you add.
Learn more → - New
Start from a design template
Apply a ready-made design template, colors, fonts, and layouts, in one click, then customize. New workspaces start from a default template automatically.
Learn more → - New
Generate QA test cases
Turn each use case into traceable test cases (happy path, negative, edge, and permission), in step-based or Gherkin format, with requirement coverage and CSV export for Jira/Xray, TestRail, or Azure DevOps.
Learn more → - New
Team workspaces
Invite teammates into a shared workspace, collaborate on the same projects with roles and live presence, and share one design library across the team.
Learn more → - New
Clickable prototypes & flow canvas
Generate a working, clickable HTML prototype per use case, view it as screens or a Figma-style flow canvas, refine it by chatting, and export it as a standalone file.
Learn more →