How do I… with RhettWorks

Rhett is the agent you talk to. RhettWorks is the software it runs: the projects, files, approvals, and the record. You say Rhett when you mean the one in the chat; you say RhettWorks when you mean the product.

Open the project and type the question in the chat. Rhett reads the schedule, budget, drawings, RFIs, mail and files on that project and answers from them, citing what it read.

Yes. Ask from the workspace with no project selected and Rhett looks across every project you can access. Pick a project when you want the answer scoped to one job.

Each answer carries a sources trail: the records, files and messages it read. Open any source to see it in the product.

Rhett drafts the work (a change order, an RFI, a schedule update) and puts it on an approval card. It is not written to the project until a named person resolves that card. An email is different: Rhett stages a draft in your Mail composer, and you send it yourself.

Yes, when your organization allows it. Rhett can run web searches during a turn and shows the pages it used as source chips on the answer.

Yes. Rate any answer with a thumbs up or down and add a note. Rhett can also offer to file a report itself when a conversation goes badly, which you review before it is sent.

Yes. The view you are looking at and the record you have focused travel with your question, so 'what is the status of this one' means the one on your screen.

That is the point. Describe what you need in plain words. You can also say 'open the submittals' and the panel opens, no navigation required.

Yes. Press Stop on the turn and Rhett cancels the work in flight. Writes to project records only land through approval cards.

Yes. Drop a PDF, an image, a spreadsheet or a document into the chat and ask about it. Attachments are kept in cloud storage, so a page reload never loses one mid-conversation.

Yes. Click the mic in the chat box to dictate. It uses your browser's built-in speech recognition (in Chrome, Google's speech service), and the text lands in the box for you to review before sending. Firefox doesn't support it, so the mic is hidden there.

Yes. Each conversation has a Fast, Standard or Expert tier, and a border on the thread shows how full its context is getting.

Yes. Durable facts and preferences are kept as memory, and you can review and edit everything it knows in Settings → Memory.

Yes. Every conversation has its own address, so you can bookmark a thread or send its link to someone with access.

Yes. Naming a panel in the chat opens it beside the conversation, so navigation never means hunting a menu.

Rhett reads the view you have open and the record you have focused, and answers about that. A breadcrumb above the composer shows what it is reading.

Because it came from the project record, not the web. When Rhett does search the web, each claim carries a source chip you can open; when it does not need to, it says so.

Rhett tells you the request was too big and what helps: a shorter message, dropping a large attachment, or asking for a narrower slice. A New chat button sits on the message when the conversation has simply grown too long.

Yes. Files and drawing sheets open in a side tab beside the conversation, PDFs included, so you read and keep talking.

Only what you can see. Rhett runs inside your own permissions, on the projects you can access.

Rate it with the thumbs, or open the report form and send it with Cmd+Return. The report carries the debug context the team needs to reproduce it.

Yes. Ask for a new project by name and asset class, drilling into a building type if you like, and it lands through an approval card like any other write.

Yes. Each lands as its own approval card in the same conversation, so you can approve one and send another back without losing the rest.

When Rhett wants to write something to the job it creates a pending card describing exactly what will change. A named person approves, edits or rejects it. Until then nothing is stored as done.

Someone with write access on that project and the permission for that area. The person who clicks is recorded on the action.

On most cards, yes. Change the amount, the wording or the recipient, then approve the edited version. A composite card that carries a whole change order or a whole budget shows the full set so you can check it before approving or sending it back.

Yes. Every resolved card keeps the proposal, the edits, who resolved it and when. The record is the click, not a chat log.

No. There is no automatic approval. A human resolves every card.

Not the records that matter. An RFI, a submittal, a daily log, a punch item or a change order can only be deleted by a person in the product; Rhett has no delete for them, gated or otherwise. A few non-record items, such as a CRM company or an unsubmitted bid draft, can be removed on an approval card you resolve.

My Tasks shows the approvals waiting on you across every project, and each conversation shows its own pending cards inline.

A card resolves once. The second click sees that it was already resolved and by whom.

The thing being changed, by name, and the fields that will change. A change order card leads with whose money it binds and says plainly when an amount is a credit rather than a cost.

Yes. Enter confirms and Escape dismisses on approval cards and dialogs.

The resolved card carries a link that jumps to the result: the task, the RFI, the section. You are not bounced there automatically.

Only when the card is theirs to resolve: an approval step routed to their organization, a response on a record they are a party to. Cards name people at other companies when that is the case.

No. A card for a permanent deletion has to identify what it removes before Approve is available.

Most open cards expire after a few hours if nobody resolves them. Task-status proposals do not; they wait for their reviewer however long that takes.

No. A status change needs a second person: the proposal is refused when the proposer is the item's only reviewer.

One. A batch write, such as a full set of budget lines or every line of a change order, is one card carrying the whole set, so a person approves money they have actually seen summed.

A card whose write is hard to undo or reaches outside your organization: a response on a legal record, a mass removal, an out-of-office broadcast. It is labeled so you slow down on it.

Only if an admin set that tool to Always allow. Each connector tool is Always allow, Needs approval, or Blocked; Needs approval shows the same card as any other assistant action.

The schedule holds phases, tasks and their percent complete against a baseline, and the daily log records what actually happened each day. Ask Rhett 'where are we against the baseline' for the rollup.

Ask Rhett to move the task and tell the sub. It proposes the schedule change on an approval card and stages the email to that company's contacts in your Mail composer. You approve the change and press Send on the mail; nothing goes out on its own.

Yes. Save a baseline at any point and the schedule reports variance per task and per phase against it from then on.

Yes. Tasks carry predecessors, and the critical path is computed from them. Ask Rhett which tasks are driving the finish date.

Yes. Each project has its own calendar of working days, and resource capacity can be set per trade so a plan is not built on days nobody works.

Record actuals on the tasks, or tell Rhett what finished and it drafts the update for approval. The daily log carries the day-by-day detail.

Yes. Every project carries a checklist built from a template for its type, and Rhett can extract items from a document you upload and add them for approval.

Yes. The schedule panel shows phases and tasks on a timeline with dependencies drawn, and the same data drives the list view.

Yes. Snapshot all phases and the state is kept, so you can compare before and after a re-sequencing.

Rhett can propose the moves (shift these tasks, pull that one forward) as an approval card. You review the plan before any date changes.

Into the deal's stages, with nested phases under each stage for a real work-breakdown structure, and tasks and milestones inside the phases.

A milestone is a point in time with no end date. A task has a start, an end and a percent complete.

My Tasks. It lists what is on your plate across every project, with priorities, review states and grouping, and a personal to-do list beside your project work.

Yes. My Tasks has a look-ahead that rolls the coming window of work, alongside the timeline and agenda views.

Yes. Tasks carry an owner and one or more reviewers, and a status change waits on a reviewer. Rhett can assign reviewers when it creates tasks from chat.

Yes. Upload the document and ask; Rhett extracts the items and proposes them as one batch on one card.

Yes. The schedule exports to CSV, and a CSV import is proposed as an approval card before anything changes.

Yes. Ask for a quality check and Rhett reports open starts and ends, missing durations, high float and dependency issues in the style of a DCMA review.

Yes. Rhett runs a what-if simulation over the dependencies and reports what moves, without changing the schedule.

Yes. Baselines are named, and any two can be compared for variance per task and per phase.

Yes. With resource capacity set per trade, the schedule reports over-allocation warnings for the window you ask about.

Yes. Each task carries personal notes and published notes, edited in the task itself.

Yes. Settings → Schedule template defines the structure new projects start with. A new project otherwise begins empty, and you add the first task or phase yourself.

Yes. Reverting a status is a high-risk card that needs a second person, so a closed task cannot be quietly reopened by the person who closed it.

Yes. Projects carry a pipeline status, and an admin defines the stages in Settings → Pipeline stages.

Yes. The tasks move to the end of that stage rather than blocking the delete, and you re-file them where they belong.

Each project has one log per calendar day with sections for manpower, deliveries, delays, weather, work performed and photos. Fill it in on the phone from the field or on the web.

Yes. Say 'log 6 framers on Level 2 and a two-hour delay for the concrete pump' and Rhett drafts the entries for approval, picking the company, location and cost code from the project's own lists.

Yes. Live conditions are fetched for the project's location, and the recorded weather uses a closed vocabulary so it can be reported on later.

Add a delay entry with the cause from the delay-type list, the location, and the RFI or submittal it is waiting on if there is one. Your organization can add its own causes to the list.

Yes. Photos from the project library attach to a day, and a photo taken from the daily log lands in the library with that day already linked.

No. Completing a day is one-way and locks it. That is what makes the log usable as a claims record.

Manpower entries carry the company, the location and a cost code, so totals by trade, by area or by week come straight from the log. Ask Rhett for the range.

A connected sub with write access to the daily log on that project can add entries; completing the day is reserved to the people who manage it.

Yes. Copy a day and its sections carry forward as a starting point for today's entries.

Log it on a later day with a note. A completed day never reopens, which is what makes it a claims record.

Because the question the log exists to answer is how many hours went into which building, by whom, and free text never groups. A crew names a CRM company and a program location; the panel offers an inline create if the company is missing.

No. Man-hours are workers multiplied by hours per worker, and if either is unstated the figure is left blank and every total says it is partial.

From two closed lists for sky and precipitation, plus free text for wind and comments, written by the crew. The live conditions card beside it is a National Weather Service feed for reference and is never part of the record.

Yes. Current temperature, today's high and low, an hour-by-hour curve with the chance of rain, a week view, and any active warnings pinned at the top.

Yes. Visitors carry a name, company and purpose; deliveries carry the vendor from your CRM, the item, the quantity and who received it.

Yes, and it is the one removal allowed after completion, because a photo may need removing precisely because it should never have been published. The day keeps a note that a photo was there and who removed it.

Yes. There is one log per project per day, shared by everyone on the job with daily-log access, so the owner and the general contractor read and write the same record.

Someone who manages the daily log on that project. The general contractor holds that alongside the developer, since the superintendent's completion is the signature the record represents.

Yes. Photos on a day appear in the Photos panel under a Daily Log album, and a photo on a live day cannot be deleted from the library until it is removed from the day.

No. The date is always the one you supply, from your own clock, so an evening crew is never filed onto tomorrow by a server in another timezone.

Ask Rhett, or open the RFI panel and fill it in. Rhett drafts the question, references the sheet or spec section, attaches what you point at, and puts it on a card for approval.

Open the photo and choose 'create RFI', or tell Rhett 'raise an RFI for this photo'. The photo is attached and linked to the RFI.

The RFI list sorts by due date and flags overdue ones. Ask Rhett 'which RFIs are past due and who is holding them' for the short version.

Yes. A connected architect sees the RFIs on the projects shared with them and responds in the record. The response, comments and approval are kept with the RFI.

Yes. RFIs reference the sheets and spec sections they concern, so from the sheet you can see every open question against it.

From the mail panel, link the thread to the RFI. The conversation then appears with the record rather than only in someone's inbox.

Yes. Closing is a recorded action; the RFI, its response and its history stay. There is no delete for an RFI from the agent, because it is a construction record.

Yes. Ask Rhett from the workspace, or open the RFI list without a project selected.

It gets its number, moves to the assigned party, and if there is an approval chain it enters review. The official response comes back, and the RFI closes or reopens from there.

Yes. Route it through an ordered chain of approver organizations; each step is decided in the record, and the chain can be reset by the owner if an approver leaves the project.

Yes. A drawing-type RFI carries a cloud on a sheet at a specific revision. Other RFIs point at a document, a contract or a submittal and carry page highlights instead.

Yes. A forwardee can read and comment but never gates the close.

Yes. Revisions snapshot the content fields, and any past revision can be opened.

The named person now in court and the RFI manager, in the app. It is never a broadcast to the whole project.

Yes. A private RFI is visible only to the named individuals on it, and a private draft can never be attached to a delay entry by someone who could not have read it.

You pick a code from your organization's code list. Search by name or number. Two RFIs on the same trade then always use the same code.

Open any spec section and it lists the RFIs and submittals that cite it. If you lack access to those records the list says so instead of pretending there are none.

A lifecycle strip from Drafted to Closed, and when the next move is yours, a highlighted action block that spells out what the action commits you to.

Create it in the submittals panel or ask Rhett. Each submittal carries its spec section, the revisions submitted, who is reviewing, and the due date.

Every revision and every review response is a row on the submittal, with the reviewer, the stamp and the date, so the history reads top to bottom.

Yes. The log shows status and days out per item. Ask Rhett 'what has been with the engineer more than two weeks'.

Yes. A connected subcontractor with access to submittals on that project can attach the document to the item.

Yes. Review actions are available from a phone browser, and the resolution is the same recorded action as on the web.

Only by a person in the product, as a soft delete. Rhett has no delete capability for submittals; they are legal records.

Yes. Submittals are tied to the project's spec sections, so from the spec book you can see what has been submitted against each section.

Through an ordered chain of reviewer organizations you type per submittal. Each reviewer answers approved, approved as noted, revise and resubmit, rejected, or for record.

The submittal returns to draft and the chain resets, so the next revision cannot inherit the previous round's approvals. The record of what the earlier round concluded is kept.

Yes. Packages group submittals, and Rhett can propose a whole register from the spec book as one card, drafts grouped into packages.

Yes. It reads the spec manual and proposes the register in one approval card, up to several hundred items, all or nothing.

Yes. Link an existing project file to the submittal instead of uploading it again, and the file cannot be deleted while a live submittal references it.

Yes. Markups and comments are kept per submittal and survive a revise-and-resubmit round.

Yes. A private submittal is visible only to the named individuals on it, and its notifications go only to them.

The named reviewer if one was set; otherwise your organization's owners and admins, so a submittal routed to a company is never routed into silence.

Yes. Submittals and packages take a code from your organization's list beside the spec section, and the register shows it as a column.

Ask Rhett ('draft a change order for the foundation rework') or open the register. Rhett proposes the lines and pricing on a card; you approve the draft, then it moves through submission and sign-off.

The named approvers on the contract sign in the record. Executing a change order is a human step; Rhett can prepare it but cannot sign it.

The committed amount posts to the budget lines the change names, so the budget reflects the change the moment it executes. Postings are forward-only; reversing one is a voided posting, not an edit.

Yes. A change order links to the RFIs, ASIs, sheets and photos behind it, so the why travels with the what.

Yes. A change order can be marked private during pricing and made visible to the counterparty when it is ready.

No. Voiding committed money is a human action in the record. Rhett can create, update, submit, comment and close, all on approval.

The register lists every change order by status with its amount. Ask Rhett 'what is the total of pending change orders on this job'.

Not in RhettWorks yet. Today it tracks the contract sum and every executed change order against it, and Rhett can write an invoice letter or a spreadsheet into your files for you to send. Pay applications with retainage are on the roadmap, not in the product.

No. It records the change orders and the contract sum they move; payment itself happens in your accounting or banking system.

Per contract, in order, so each counterparty's change orders count up on their own contract.

Draft, submitted, in review, approved, closed, with rejected as a stop. Revise-and-resubmit returns it to draft. Void is a stamp on an executed change order, not a status.

No. The author can never approve any step of their own change order, and no one person can clear two steps of the same round.

A specific named person at the contract owner's organization, chosen by whoever submits. Earlier steps, such as an architect's review, can be routed to a whole organization so a vacation does not stall the job.

To a phase, a building, a unit or a site component picked from the project's program. A draft can start without one; nothing goes out for signature unscoped.

What is in your court, what is in review and what has executed, with the original sum, the net of previous change orders, this one, and the new contract sum, laid out the way a signed change order reads.

No. The location's name is frozen at execution, so the paper reads the way it did the day it was signed.

Its postings are reversed and it is stamped void by a steward admin. The execution itself stays on record; it is never erased.

Only a never-submitted draft. A submitted change order is a legal record and stays.

Yes. Comments live on the change order, and Rhett can add one on approval.

Yes. The author names the counterparty organization and the contract between the two is found or created; a change order can never be addressed to your own organization.

Create a bid package with its scope lines and drawings, then invite companies from your CRM. Each invitation is logged with what was sent and when.

The scope matrix shows every scope line against every bid, with each bidder's declared coverage and price, so gaps and exclusions are visible before award.

Yes. Upload the bid and ask Rhett to ingest it. It extracts the line items and proposes the coverage per scope line on one card for approval.

A company holds one base bid per package plus any number of labeled alternates, compared as separate columns. Awarding a base and an alternate sums them.

Set the bid's status to Awarded. That is the buyout; several awarded bids on one package sum, and the awarded amounts roll up against the line budget.

Yes. A bidder gets their own bid view on that package, submits their pricing, lines and takeoff there, and never sees other bidders or your budget.

Yes. Counted quantities live under each scope line as takeoff items, with labor items attached to the material they install.

Yes. Unit-price history reads across your projects from bids Rhett extracted, so 'what did we pay per square foot for drywall last year' has an answer.

The transmittal log appends a row per bidder per send with the drawing area and issuance snapshotted, so a later revision cannot rewrite what they got.

Yes. A contingency reserve can be set at the item, section or project level and shows in the budget as headroom.

Name it from the spec list so the cost code fills itself, set a due date, write instructions to bidders, assign its scope lines, and pick the drawing sets every bidder must price off.

Yes. Save the package as a revision and its whole template is frozen as r1, r2 and so on; any past revision opens read-only.

Yes. Invite bidder lets you check any number of packages, or all of them, and one-off scope lines in no package become a package for that invite.

No. Leave it alone and the bid files under a General package created the first time you need it. The bid amount is optional too, so you can log who bid before the envelope is open.

Their own bid page: the packages they were invited to, the scope lines, the spec sections you linked, the drawing sets, and their own pricing. Never your budget, never another bidder.

Yes. Each package lists the sections it and its lines point at, and the bidder opens the current text right there. Only linked sections are readable.

Yes, and you can print any bidder's bid from the bid room. Both produce the same sheet: company name and logo, project and package, every scope line with coverage, every line item, takeoff and labor rows, the total, alternates, and a signature block. Nothing of yours is on it.

They get a branded invite by email, and they appear in the bid room roster with an Invited status that says whether they still need to sign up, said they will bid, or declined.

The bid's details sit in an editable panel beside the estimate, saving as you go, and the sub's own line items open for editing underneath with the item subtotal behind each scope's number.

Focus that bidder's column and the rest step aside: every scope line, their number beside your budget, and their quantity, unit, unit price and labor in the same columns as yours, with any difference from what you counted highlighted on that number.

Yes. Every bidder's price shows its difference to your budget underneath, and a Buyout switch colors each scope line by where its buyout stands and totals what is bought out.

It lists underneath the scope, and items that fit no scope row collect in an Unmatched row, so nothing they sent gets folded in quietly.

Yes. A quantity, unit and rates, or a lump sum and a price, go straight onto the scope row, and you can still open the scope later to price item by item.

Yes. Upload from the bid room and the document lands in the project's bid uploads folder as that bid's newest version, with an optional new total and note. Rhett then reads it and proposes lines and coverage unless you uncheck that.

Yes. Each coverage cell names who last declared it, and a History link shows every declare, update and clear, including ones made through Rhett approvals.

A sub you have added to a package while assembling the bid list, before any invite goes out. It shows as a column but does not count toward invited or responded.

Yes. Hand Rhett the proposal and it adds the company, creates the column and files what the document said, all on one card.

Yes. A subcontractor invited to bid can attach their proposal and say 'put this on my bid', answer an invite, add an alternate, and edit their line items by chat. Rhett sees only their own bid.

Yes, and reviewing a bid shows whether it was priced off the package's required sets, naming any that are missing. Advisory only; no status is ever blocked.

Yes. A labor item nests under the material it installs, and a sub's quote can be labor-only or material-only.

Yes. Make a code in your organization's list for miscellaneous work. Several custom scopes can share it, each with its own row, budget and coverage column.

Build the line budget on your organization's codes, by hand or by asking Rhett to propose lines from the spec book. Each line carries budget, committed and variance.

Your organization keeps its own code list in Settings → Codes. Budgets, bids, RFIs and submittals all use that list.

Custom titles can be added to a line, and industry codes map onto the chart at commit time without overwriting a code you picked by hand.

The budget summary rolls awarded bids and executed change orders into committed per line and shows the variance. Ask Rhett for the lines that are over.

Yes. The program panel holds phases, buildings, units and components per scenario, with revisions and as-of history, and Rhett can update it on approval.

Settings → Codes lists every code your organization uses. Add, rename or retire a code there.

Yes. It proposes budget lines from the committed spec sections as one card.

Yes. The project's code roster counts codes across specs, submittals, RFIs, budget, bids and drawings, under Coverage in Specifications.

Yes. Save a scenario as a new named branch and edit it independently; every scenario keeps its own revisions and history.

Yes. The program's history lists every change with who made it and whether it was a person or Rhett, and a past state can be reconstructed.

No. The program document carries the owner's prices, so it is never shared through a connection. Subs and GCs get the location names only, through the daily log and change-order pickers.

Upload the PDF set to the project. Rhett interprets the sheets, indexes them by number and title, and records the issuance so each sheet has a revision history.

The set shows each sheet with its live revision and what superseded it. Ask Rhett 'is A-301 current' and it answers from the set, not from memory.

Yes. Ask for 'the level 3 electrical plan' and Rhett finds the sheet by number or title across the set.

Create an ASI, attach the revised sheets and add the recipients. Issuing serves it to the distribution and the served record is kept; drafts can be edited or deleted, issued ones can only be voided.

Yes. Connected design firms see the sheets on the projects shared with them, with the same revision status you see.

Yes. RFIs reference sheets, so from the sheet you see every open and closed question about it.

Yes. A bid links to the drawing areas it was priced off, stamped with the revision at the time, so a later issuance does not silently move the goalposts.

Yes. Draw polygon regions, drop text, cloud an area, and see grouped markup counts with fill controls. Markups are an overlay; the original is never altered.

They carry forward onto the new revision, and simultaneous edits merge instead of overwriting each other.

Yes. Every sheet has an activity feed of who changed what and when.

Yes, and you can control visibility by review status so unreviewed sheets stay out of the way.

Yes. Sheets belong to sets, and drawing areas group the sheets a bid or a transmittal refers to.

Yes. A revision carries a calibration so markups and annotations are placed in real units.

Yes. Markups carry a review state and comments, so a sub's markup can be accepted or answered in place.

Yes. Ask for a sheet and it appears as a card you open straight into the viewer.

An architect's supplemental instruction: a formal instruction to the field, the outbound mirror of an RFI. Draft it, name recipients, attach files and sheets, then issue to allocate a number and publish the selected revisions.

On their own row in the register, and they can file or update an impact claim while the instruction stays open.

Only a draft. An issued instruction is a served record; withdraw it by voiding.

Yes. The spec reference searches your project manual, and when it resolves, the ASI links straight to the section.

Sections are authored directly in the app, and Rhett can draft a section for you to approve. Bulk PDF import of an existing spec book is in progress.

Yes. Search by section number, title or wording, or ask Rhett 'what does the spec say about fire-stopping at penetrations'.

Yes. Rhett can draft a new section or a replacement body for an existing one, on an approval card that shows the full text before it lands.

Yes. Your organization keeps master standards, and a project adopts the ones it needs, so the standards evolve in one place.

Yes. Coverage shows, per section, what has been submitted and what has not.

Yes. The section list exports to CSV for the people who want it in a spreadsheet.

Bulk PDF import. Upload, review the parsed outline, then commit, is in progress. Today, sections are authored directly in the app, or drafted by Rhett for your approval, and each keeps its own revision history from that point on.

The standard one: PART, article, paragraph, subparagraph. Sections it writes are collapsible and edit block by block, like an imported book.

Your save is stopped and you are told the section changed, so you can copy your work, load the updated text and decide what to do.

Yes. A section number stays free-form, so you can number a book your own way, and a recognized industry code still resolves to its Rhett equivalent.

Yes. Coverage shows what has been submitted against each section, and each section lists the RFIs and submittals that cite it.

Only the sections you linked to their package or its scope lines. Nothing else is visible to a bidder.

Add items from the web or the phone, each with a location, an assignee company, a description and photos. Rhett can create one from a photo you describe.

A connected sub sees only the items assigned to them on that project, with the photos and the location, and can mark them ready for review.

Same record shape, different purpose. A punch item is closeout work that must be done; an observation is a field note (a safety concern. A quality issue) that may or may not become work.

Yes. From a photo, choose 'create punch item' or tell Rhett 'punch this: cracked tile at the level 2 lobby'. The photo is attached and linked.

In-app notifications are always on; email notifications for punch items are switched on per organization.

Draft, open, ready for review, closed, with rejected as a return to open. The sub moves it to ready for review; closing, rejecting and reopening a closed item need manage rights.

By a unit picked from the project's program, or free text. Say 'unit 302' and Rhett resolves it; if the same number exists in two buildings it asks which.

Yes. Pin it on a sheet at a specific revision, the same way an RFI does.

CRM contacts, platform users or plain email addresses, up to 25. Ask Rhett for everyone at a company on the project and it builds the list.

A closed set: quality non-conformance, quality workmanship, safety fall protection, safety electrical, safety housekeeping, with a hazard and a contributing condition.

Yes. A draft gets its number and waits; opening it is the send.

Yes. Non-photo files attach to a punch item, and a connected sub can open the files on their own items.

No. Comments are internal to the item's organization.

No. Numbers are never reused, so a reference in an email stays unambiguous.

Into the project's photo library, with albums, tags, a comment thread per photo and links to the RFI, punch item or daily log it belongs to.

Yes. Ask Rhett to interpret a photo or a set and it summarizes what it sees, which you can attach to the record as a note.

Yes. Drop a photo into the conversation and ask Rhett to add it to the project library; it lands with the tags and album you name.

Yes. Albums can be private to your organization, and moving photos between public and private albums is an approval-gated action.

Yes. Select the photos and export a PDF of them for the people who want them outside the product.

Yes. The tag vocabulary is your organization's, so 'MEP rough-in' means the same thing on every job.

A timeline grouped by the date the photo was taken, and albums. Each photo carries a caption, tags, a location from the program, and a trade.

The camera's EXIF capture date. A photo with no EXIF date sorts by upload time in an Undated group.

On upload the stored copy has its GPS and camera maker note removed for JPEG, HEIC, PNG and WebP; the pixels are never re-encoded. Nothing plots photos on a map.

Yes. HEIC is decoded on upload and a thumbnail is made.

Yes. A private photo is visible only to people who manage photos on the project, even inside a shared album.

Yes. Pick teammates on the project; they see that album and its photos while everything else private stays hidden. Sharing never adds edit rights.

Yes. Stars are personal, so your Favorites are yours and a teammate's are theirs.

Yes. Pairs are symmetric and show together.

Yes. Comments support @mentions, and the mentioned person is notified.

Yes. Markups are an overlay; the original is never changed, and rotation is applied at display time.

Yes. Any library photo can be the project's cover.

It goes to the photo recycle bin, where it can be restored or permanently purged by someone who manages photos.

Yes. Filter by up to 20 tags at once, by a program location, by trade, or by album, and change how many thumbnails sit across the grid.

Settings → Workspace → Photo tags, the list your team picks from, with counts and last-used dates. Renaming a tag updates every photo that carries it.

Yes. A photo links to the RFI, punch item, observation or daily-log day it belongs to, and the record lists its linked photos.

No, deliberately. RFI and submittal attachments are visible to fewer people than the library, so they stay with their record.

Connect your Microsoft 365 account and your own mailbox appears in the project. Rhett reads the threads you point at, drafts replies, and links conversations to the record.

Rhett never sends. It stages a draft into your Mail composer with the recipients and the text filled in; you read it, edit it if you like, and press Send from your own Microsoft mailbox.

Ask Rhett to write to the subs on the job. It pulls the contacts from the project team and CRM and stages one draft in your composer with the recipient list filled in. You check the list and send. A recipient outside the thread you were reading is flagged for you to confirm.

When your organization enables mail context, Rhett can read the relevant threads in your own mailbox to answer 'what did the architect say about the canopy'. It reads only your mailbox.

Ask. Rhett scans your sent mail and lists the recipients at that company with the dates, and tells you whether the scan reached all the way back.

Yes, on approval. Because an auto-reply goes to everyone who writes to you, it is treated as a high-risk card.

Your mailbox stays in Microsoft. RhettWorks stores the links between threads and projects, staged drafts until you use them, and a record of the messages Rhett read for you with a short preview, so a later question can find them again.

Yes. It proposes one from your profile and organization details; you edit and save it.

Yes. The Mail panel has a composer with reply, reply-all and forward, attachments, signatures and your confidentiality notice, and it saves to your real Outlook drafts.

Yes. Pick it from My Files, Org Files, a project or OneDrive and it attaches to the draft server-side, so nothing buffers in your browser. Rhett can suggest attachments too, marked as added by Rhett.

Yes. Ask Rhett to save it to My Files, Org Files or a project and it lands under its real filename.

Link it from the thread, or accept Rhett's offer when it notices the contacts belong to one of your projects. Linked threads show a project chip and can be filtered by project.

Yes. Named signatures per mailbox, with one default.

Yes. An admin can set the organization's notice, and each person can inherit it, turn it off or write their own.

No. Unsent drafts are excluded from what Rhett reads as repliable, so a revision updates the same draft instead of answering it.

No. A forward must carry its source, and the composer halts a send whose forward came back without the original attached.

Two sections: your CRM and your Microsoft directory. A Microsoft-only address offers Add to CRM.

Yes. The sidebar shows the unread count, and it updates the moment you read something.

Yes, and if the token expires you get a Reconnect button right in the chat instead of a dead end.

Yes. It downloads the attachment and extracts the text to answer from, without dumping the file into the conversation.

No. A body-less invitation or cancellation shows a synthesized line saying what it is and when.

Yes. With Microsoft 365 connected, your Outlook calendar appears in the workspace, and Rhett can create or move events on approval.

Ask Rhett to schedule it. It drafts the calendar event with the attendees from the project team and puts it on a card; approve and the invite goes out.

Yes, for a meeting your account organized, once your Microsoft tenant allows transcript access. Rhett fetches the transcript and produces a summary, the decisions and the action items. For a meeting someone else organized, upload the transcript file instead.

Yes. Upload the transcript file and Rhett summarizes it the same way. A recording itself is not processed; Teams produces the transcript.

Yes. Each action item is a record with an owner and a due date that you can resolve, and Rhett can turn one into a task or an RFI on approval, or stage an email draft for you to send.

Yes. Once the meeting is ingested, ask Rhett and it answers from the transcript with the timestamp.

Yes. It finds free times across participants and proposes the event on a card.

Yes, on approval. Adding or removing attendees is its own card, and a recurring series is refused rather than edited as one occurrence.

Agenda, week, month, and a meetings list of past meetings with their summaries.

Yes. Open the event and ask for its prep brief: the people, the project context and the open items.

Yes, for a meeting you organize that has not started. It records, not transcribe-only, and every surface says so; the person who enabled it is logged.

Only when you fetched a Microsoft transcript yourself, and only for people the transcript shows actually spoke. An uploaded transcript proposes a card instead. Rhett never writes contact history on its own.

Yes. Link the meeting and its summary and action items live on the project.

Yes, by hand. A transcript never links a participant to a contact on its own.

In the project's file store, organized by project with your organization's folders, and readable by Rhett when you ask about them.

Yes. Upload it to the project or drop it into the chat and ask. Rhett extracts the text and images and answers from them, citing the file.

Yes. Ask for the comparison, the schedule of values or the draw, and Rhett generates an .xlsx in the documents workspace on approval, which you can then edit in place.

Yes. Link the workbook and ask for the change. Rhett edits the cells you name on an approval card that shows before and after.

Yes. Decks in the documents workspace can be edited in place.

Yes. Ask for the letter, the notice or the report and Rhett writes it as a Word file or a PDF into your files once you approve the save.

No. Start with one project and the files it needs. Drawings, mail and the rest come in as the job asks for them.

Yes. In your own space it just does it; in a shared project folder the change goes on an approval card.

My Files, Org Files, a folder per project, and your OneDrive, shown as tabs in the Library so it is always clear where a file lives.

Yes. Deleted files go to a Trash view and can be restored until the trash is emptied.

Yes. An admin sets a folder template in Settings and new projects start with it. A new project otherwise starts with no folders.

Yes. Files open in a side tab, PDFs included, beside the chat.

Yes. Conditional formatting, color scales, data bars, icon sets, banding, named ranges, dropdowns and cell notes show and survive a save, as long as you change cell values. Rules that compare two cells are not shown yet.

Yes. Add, rename or delete one and it saves into your original workbook with charts, images, dropdowns and comments left alone.

Yes. It reads raw source cells to ground its edits, and an edit card shows before and after.

Into the project's folder when a project is in context, on approval; otherwise into My Files right away. Rhett asks where to save when it is ambiguous.

Yes. Opening the document from a bid card opens it as a tab in the workspace instead of downloading it.

Yes. A spreadsheet exports as an .xlsx; documents Rhett writes are Word or PDF files already.

No. Rhett reads a file when you ask about it or point at it. Nothing is indexed in the background.

From the project Team panel, invite the person or the company. A person in your organization gets the project; a company from another organization gets a connection with the slices you grant.

Send a connection invitation to that company. Their admins see it as a Pending row at the top of Projects and accept it there. Their people then see exactly the areas you shared (submittals, RFIs, their punch items) and nothing else.

In the CRM: companies with their trades and the asset classes they cover, contacts with their roles, and the communications logged against them.

Yes, on approval. 'Move Dana to the estimating role at Summit Concrete' becomes a card you approve.

Yes. The Team panel merges your people and the connected companies' people into one roster with each person's access.

Yes. Companies carry the asset classes and building types they cover, so 'who has done cold storage' is a CRM question Rhett can answer.

Transferring project ownership between organizations as a deliberate, recorded step is in development.

Yes. In-app messaging supports direct messages, groups and a room per project, with optional end-to-end encryption, and you can bring Rhett into a conversation with an @mention.

The Team panel's Add members dialog lists your whole organization as checkboxes; tick everyone, pick one access level, add them together.

Your own people pinned at the top, then every connected company and its people, each row with their access on the project. An invite from another company shows as a Pending row at the top of Projects.

By a branded email invite to one address. Nothing is shared until they accept.

The construction phases, by default. An architect also keeps the design phases they author. You can change what any company sees from its panel on the Team tab.

Yes, if the preset you share allows it. Raising new ones and seeing who else is on the project stay with your own team.

From its panel on the Team tab, under everything the removal affects. If it fails the confirmation stays open and says why.

Yes. A person's card shows your direct message and shared groups, and a contact with an account has a Messaging tab.

Its type and sub-types, the asset classes and building types it covers, addresses, emails, phones, social links, a color, its people, and its history.

Yes. Interactions are logged against the contact and show in their timeline.

Yes. A contact downloads as a vCard for your phone or mail client.

Yes. People search looks in the CRM first and falls back to your Microsoft directory.

Yes. Types and roles carry sub-types your organization builds as it goes, so 'Structural' under Engineer and under Architect are different entries.

Owner, admin and member. Owners manage everyone; admins can add admins but not owners; an organization can never be left without an owner, and every role change is recorded.

Yes. Settings → Members shows each person's projects and lets you cut them from specific ones.

Yes. Direct messages, groups, and a default room per project, with @Rhett available in a conversation.

End-to-end encryption is available as an option per conversation, with the keys held in your browser.

Yes, and the notification preview is rewritten along with it.

Yes. Reactions are supported in messaging.

Organizations partition the data, and the isolation is enforced in code on the server, never by asking the model to be careful.

A connected company sees the slices you grant on that project and nothing beyond it. Approvals and the project record stay scoped to the people on the job.

Assign a permission template (viewer, editor, admin, or one you define) and override individual actions per person when a role does not fit.

Yes. Tool permissions let an admin block agents or tools organization-wide (web search, outbound mail. A whole module) and the block applies to every conversation.

No. Rhett acts within the asking person's permissions. If you cannot read a record, neither can Rhett on your behalf.

Yes, within the ceiling you granted the company. They cannot grant their people more than you gave them.

Yes. Connectors such as market-data services use each person's own credentials, and an admin controls which connectors are available.

Every write is recorded with the person, the time and the trace of the turn that produced it, and approvals keep the proposal beside the resolution.

A named set of what someone may read, write and manage per area, with an organization-wide default. Templates apply per person, and per-person overrides adjust individual actions.

Yes. A collapsible rollup of organization and project permissions lets an admin audit access in one view.

Anything past the sharing ceiling: write access beyond the schedule, the program's prices, your organization's roster. A preset that grants too much is greyed out with the reason named when you try to share it.

Yes. Settings → Help lists everything RhettWorks can do, filtered to what your role and permissions let you use.

Access management goes read-only: no adding members, no connecting organizations, no access edits, with a notice saying why. Restoring the project brings it all back.

Yes. The resolver of every card is recorded, and consequential records keep a full audit trail of who did what and when.

No. It is enforced by the server on every read and write. The model never chooses which organization's data a query runs against.

A small operations team can access records to operate the platform, as the Privacy Notice states. Those views are separately gated and recorded.

RhettWorks runs in the browser today, on desktop and phone. A native field app is in development. Write to us if you want to be in the beta.

Ask Rhett, approve cards, take and tag photos, fill the daily log, work punch items, and read the record for any project you can access.

The in-app inbox keeps the list of approvals waiting on you, assignments and mentions. Push notifications to the phone are in development.

Yes. It is one record. A card approved from the browser on your phone is resolved everywhere.

Yes. RhettWorks can be pinned to light or dark from Settings, in the browser on desktop or phone.

Push notifications to the phone are in development. Today the in-app inbox shows a title and a short summary, which can include the start of a message or the subject of an RFI, submittal or task.

Yes. Add photo on a punch item opens your phone's or computer's camera or photo picker in the browser, and the photo lands in the library linked to the item.

No. RhettWorks runs in the browser, so updates ship to everyone the moment they're released. Nothing to install or update.

Investor reports (built from the project record, reviewed as a draft, and published to the investors on the deal, with Rhett drafting the narrative on approval) are in development.

A read-only investor portal (reports, offering terms and their own capital account for the people who funded the deal) is in development.

Modeling the waterfall and distributions from a deal's offering terms is in development.

A scoped, read-only assistant for an investor's own reports and position, inside the portal, is in development.

Your organization keeps a brand kit (logo and colors) in Settings, so documents produced for investors carry your identity rather than ours.

Draft assistance for investor reports (Rhett reads the project record and drafts the narrative for you to edit before anything is published) is in development.

The plan is an investor invite to the deal: they get the portal and nothing else in your workspace. The portal itself is in development.

An accreditation attestation with evidence upload is planned for the portal and is in development.

Yes, through the property-data connector your organization enables. Rhett answers with the vendor named as the source.

Yes, through the market and property-data connectors in Settings → Connectors. Each person uses their own key by default, or the organization's shared key if an admin allows it, and answers cite the vendor.

For jurisdictions in the records library, yes: planning and zoning minutes, agendas and staff reports can be searched and read. Ask Rhett whether a jurisdiction is covered.

Yes. When web search is enabled for your organization, Rhett runs the searches and shows the sources it used as chips on the answer.

An outside data service you plug in from Settings, with your own credentials. Rhett gains that service's tools for your conversations only.

Through the market-data connector: quotes, company financials, earnings transcripts, macro indicators, commodity prices such as lumber and steel, and finance news. Read-only, with the vendor cited.

FRED for rates and macro series, BLS for employment, Census for demographics and HUD for fair market rents and income limits, each connected from Settings.

Yes. Admins pick who may connect, whether a connector uses each person's own key, one shared organization key, or both.

Yes. Each connector has a Test connection that calls the vendor and reports what it found.

It is switched on per organization. Off, Rhett answers only from your record; on, it searches and cites.

Tell us how you build today on the contact page. We set up your organization and you sign in at app.rhettworks.com.

Yes. Passkeys are supported alongside email sign-in, and the app nudges you to enroll one on a new device.

From Settings, invite by email. Each person joins your organization and gets access to the projects you assign.

Yes. Connect your account from Settings to bring your mailbox and calendar into the workspace. Your organization's Microsoft admin may need to consent once.

Your organization's records remain yours. Deletion and erasure requests are handled through the contact address on the privacy notice.

From the feedback control in the app, or by rating a Rhett answer. Reports go to the team with the context they need, and Rhett can offer to file one itself when a conversation goes wrong.

Settings → Version lists every release and what it added.

This site publishes an llms.txt describing the product for agents. Product integrations are arranged through the contact page.

A short guided profile flow sets your name and workspace, so you land ready to work.

Yes. Switch between accounts from the account menu without signing out.

Yes. Close the browser and come back to the same tabs, document, folder, sheet and mail conversation. A chat or a mail thread reopens only within an hour, so on a shared machine yesterday's messages are never left on screen.

From Settings → Security, or when you first set up your account. Remove any you do not recognize there. Your password keeps working.

A small dot appears on your profile; update from Settings → Version whenever it suits you. The app never reloads out from under you.

Yes. Light and Dark, plus System to follow your device setting.

Yes. Enter confirms and Escape dismisses dialogs and cards, and Cmd+Return sends a report or an email.

Settings → Workspace: members and roles, permissions, connectors, checklist and schedule templates, folder templates, photo tags, pipeline stages, spec standards, brand and assets, and the email notice.

The bell and its panel, with per-tab badges. A notification resolves itself when the thing it was about gets handled, and you can swipe one away.

Yes. The public site's home page runs an interactive product demo per audience, and the contact page arranges a walkthrough with the team.