Ten AI agents for general contractors, from the first bid to closeout. They find and read bids, check contracts, level sub bids and take off plans, then run the project, procurement, daily reports, finance, every message and your pipeline.
Every finding cites its source, and you make the call.
Estimators check portals by hand, open postings that don’t fit and still miss the ones that do.
A chat tool can’t watch every portal for you, and it doesn’t know your trades, region or bid range.
ClaraWatches the portals for you, reads every tender and brings you the ones that fit, with the reason.
Hundreds of pages read cover to cover for bonds, LDs, prevailing wage and DBE goals, then the bid documents filled in by hand.
A 300-page package is too long to work through in a chat window, answers come back without the page, and it can’t fill your bid documents from your company profile.
ClaraReads every page, cites each finding, drafts the Go/No-Go and fills in your bid documents, ready for your review.
Owner edits to AIA A201 or ConsensusDocs compared line by line, often by counsel at the very end.
It doesn’t know the standard text or your playbook, so it can’t tell you what changed or whether you accept it.
DoraCompares every clause to the standard and your playbook, and drafts the redline and questions.
Nine proposals in nine formats copied into a spreadsheet; scope gaps and exclusions show up after award.
It can’t line up nine PDFs against one scope and price the gaps back in.
NovaLevels every bid to one scope, flags gaps and unbalanced prices, and drafts the award memo.
Walls, rooms, doors and windows measured by hand from PDF plan sets, then checked against the bid documents.
A chat tool can describe a drawing, but it can’t measure one.
TheoTakes off quantities from 2D PDF plan sets, each marked on its sheet.
Preconstruction eats weeks of reading. The agents do the first pass in minutes, not days (illustrative). Each one owns one phase, works from the US forms you sign, and leaves a trail you can audit. Six more agents carry the pursuit and the job: Patrick, Dario, Lex, Fina, Colin and Pixer.
Bid Agent
Reads the whole bid package, flags risks with page cites, drafts your Go/No-Go and fills in the bid forms.
Quantity takeoff
Takes quantities off 2D PDF plan sets, each one marked on its sheet.
Contract & Specification Agent
Checks prime contracts and subcontracts against AIA and ConsensusDocs, and flags the clauses that shift risk to you.
Bid Evaluation Agent
Levels sub bids to one scope and shows the gaps and unbalanced prices before you award.
The schedule and three-week look-ahead, buyout, action items and the punch list to closeout.
More about Patrick →DarioProcurement and logisticsPurchase orders, delivery booking and tracking, tickets read from a photo, cranes and equipment.
More about Dario →LexField reportingDaily reports from photos and voice notes, manpower, weather, inspections and safety.
More about Lex →FinaFinance and budgetBudgets and change orders, AIA pay applications, payables, lien waivers and a cash forecast.
More about Fina →ColinCommunication hubOne inbox for email, WhatsApp and text, RFIs and submittals tracked to an answer, minutes.
More about Colin →PixerPipeline and proposalsEvery opportunity by stage, the bid calendar, proposals from your past projects and win rate.
More about Pixer →Preconstruction is just the start. BAU AI is building the complete operating system for construction, from the first bid to final handover. One record runs through it: Clara flags an uncapped LD, Dora reviews the clause, Nova checks sub scope against it. Pixer tracks the pursuit, and from award to closeout Patrick, Dario, Lex, Fina and Colin take the project forward on the same record.
BAU AIWorkspace · agents by phase9 agentsAI Brain, the knowledge base
Today your documents sit in folders, inboxes and people’s heads. The AI Brain organizes them once, and every agent draws on them right away.
Move the sliders to estimate the hours your team gets back each month. The time-reduction rates are illustrative, not yet validated with US data.
Time your team spends finding and reading bids.
Contracts your team reviews each month.
Bid packages your team levels each month.
Takeoff work each month.
Clara watches the portals for you, reads every tender and checks it against your trades, region, project size and project types. What fits lands in one inbox with the reasons it fits. What doesn’t is dropped, and you can see why.
Summit Medical Center, Phase 2
Channel dredging, Phase 4
Sample postings, distances and values are illustrative. Clara surfaces matches for review; your team decides what to bid.
Federal and state agencies, counties, cities, school districts, universities, transit and DOT, water and utility districts, airports and ports, housing authorities, hospital districts and private owners.
Everyone reads the same package for a different reason. Pick a seat to see what the agents pull out of the Summit Medical sample for that role, each finding cited to its page.
“Do we bid this, and what does it cost us if we do?”
7 of 34 findings matter at this seat. The three that decide the bid are open on the right.
ClaraDoraNova“What is in my scope, and what flows down to me?”
4 of 34 findings matter at this seat. Scope and flow-down terms first, so you price the right work.
ClaraDoraDora“Are these bids priced on the same scope?”
5 of 34 findings matter at this seat. Each one changes what a bid includes, so leveling starts here.
NovaNovaNova“Will this award hold up in review?”
4 of 34 findings matter at this seat. These are the checks an award review asks about.
ClaraClaraNova“What will bidders ask about?”
4 of 34 findings matter at this seat. Clear these before the questions arrive.
TheoClaraClara“Do the specs, drawings and reports agree?”
4 of 34 findings matter at this seat. Where the specs, drawings and reports part ways.
ClaraClaraClaraBAU AI runs live in Europe and is rebuilt for US standards. Founding teams get early access and a direct line to the people building it.
Apply for a founding spotNo invented logos or testimonials. Each control below is either in place or named as in progress.
Start with the agents you need and add more as you grow. Each higher plan includes everything in the one before. Prices are on request while the founding cohort is open.
The entry point: one agent, ready to go.
Two agents of your choice, bundled.
All three live agents on one record.
Custom deployment, your infrastructure.
Full access to all three agents with no monthly fee. You pay only when a contract is actually awarded to you, with an annual cap. Terms are agreed in writing before activation.
Discuss Pay-When-You-WinOn US infrastructure, in an isolated tenant, never used to train models. Where a control is still being audited, we say so.
Independent audit underway across security, availability, and confidentiality controls.
Processing and storage kept on US infrastructure. Your bid data stays in the country.
We never use customer documents to train or improve models. Export or delete your data at any time.
Every customer's documents are processed in a logically isolated environment, encrypted in transit and at rest.
SAML 2.0 and OIDC single sign-on, SCIM provisioning, on higher plans.
Every action documented, timestamped, exportable for owner and compliance review.
* The technical and organizational controls are in place; the independent SOC 2 Type II attestation is underway.
If your question is missing, write to tushar.adsul@bauai.co.
BAU AI is the operating system for construction preconstruction: a platform of specialized AI agents that read every bid, contract, and proposal, cite every finding to the source, and hand work to each other across the lifecycle. In practice it is a team of AI agents for US contractors, starting with preconstruction. Clara finds and qualifies bid opportunities, reads the package, drafts a Go/No-Go and fills in bid forms. Dora checks contracts against AIA and ConsensusDocs, and Nova levels sub bids. Theo takes quantities off 2D PDF plan sets.
No. BuildingConnected and Procore send invitations to bid, track bidders and run your projects, and BAU AI does not do that. It works alongside them: it reads the solicitations, contracts and sub proposals your team handles and cites every finding to the page.
BAU AI is built around US preconstruction documents. It reads the whole package, checks clauses against the AIA A201, ConsensusDocs and EJCDC base forms, levels bids by CSI MasterFormat and cites the page and clause for every finding. A general chat tool can summarize a document, but it can guess and usually gives you no page to check.
Bid packages and specifications, AIA, ConsensusDocs and EJCDC contracts, subcontracts and sub proposals, plus 2D PDF plan sets for Theo. Supported formats are PDF (including scanned, with OCR), DOCX and XLSX. DWG and DXF are on the roadmap; today, export drawings to PDF.
Your data is processed and stored on US infrastructure, in a logically isolated tenant, encrypted in transit and at rest. SOC 2 Type II: the independent audit is underway. SSO (SAML, OIDC) and SCIM are available on higher plans, and every analysis is logged and exportable.
No. We do not use customer data to train or improve models. You can export or delete your data at any time.
No. The agents do the reading and the first pass; your team makes the call. Every finding is cited so an estimator, PM or attorney can check it in seconds. Dora does not give legal advice; counsel decides.
BAU AI runs live in Europe and is built for US standards, and the founding cohort gets early access with a direct line to the team. You can watch Clara work a sample bid package on her page, every finding cited, before you talk to us.
Bring one real bid package to a 15-minute call and see what the agents find, each finding cited to the page.
Founding cohort · US data residency · No training on your data
