Agents in the project inbox
Ground Outlook workflows in project documents and prior RFIs. Draft the next action for the responsible person to review.
You receiveGrounded tool path
Delivered through Agentic AIProjects run on answers.
Grounded in the record.
Connect Outlook, project documents and Primavera P6 so RFIs, schedules and commitments reach the people who can act.
RFIs grounded in documents
Draft replies for the engineer to review.
Schedules checked against standards
Read scope and logic before the programme is accepted.
Commitments kept in view
Connect messages to the project and its next action.
Weeks → days planning time · A project-controls software company
Make the project record
work for the team.
An RFI lives in one inbox, its evidence in a document and its consequence in the schedule. Connect those records while keeping approval with the responsible person.
Ground the draft
Read project documents and prior RFIs before drafting a reply.
Connect the programme
Bring schedule and project-control data into a common model.
Keep approval explicit
Engineers review replies; operators approve commitment closures.
This is the right partner when
- RFIs require repeated searches across project documents.
- Portfolio reporting depends on copied schedule data.
- Owner’s representatives need an evidence-linked view of commitments.
The engineer reviews the draft before it is sent.
Data and AI work
for project delivery.
Start with the decision, follow the data, and leave each engagement with something your team can use.
Project controls on one model
Map Primavera P6 and project-control feeds into a validated model. Make portfolio reporting use the same checked records.
You receiveOnboarding runbook
Delivered through Data IntegrationPortfolio reporting owners can trust
Agree the definitions behind schedule and commitment reporting. Give owners a common model across projects.
You receiveSemantic model
Delivered through Analytics & BIControls on each AI action
Set the identity and permissions of the project agent. Keep consequential actions behind an explicit human sign-off.
You receiveSign-off design
Delivered through Trust LayerA costed plan for project AI
Read the inbox, document and schedule dependencies behind one workflow. Price the gaps before funding the build.
You receiveCosted modernization map
Delivered through Data Strategy Assessment
The Mandelbulb Kit Construction & AEC editionBuilt and maintained in-house
Engineered in-house.
The Mandelbulb Kit.
Accelerators, agents and products we build and maintain ourselves, brought into an engagement wherever they fit.
Construction agents · Work in the inbox
AEC Platform
RFIs, schedules, proposals and CVs, drafted from the project record.
In the project record
Explore more about AEC PlatformAlso in this edition
- Atlas AIAcross the portfolioTurns a project inbox into an operating picture: every message routed to its project, today’s priorities surfaced, commitments tracked., explore Atlas AI
- BlumcoreIn project workflowsThe runtime every Kit agent runs on: it reads your data, acts in your systems and audits every step., explore Blumcore
Proof from
project delivery.
A project-controls software company
Software · Project controls
An AI Planner That Reads Scope and Produces P6-Ready Schedules
Planning in Primavera P6 meant slow, manual scope extraction and schedule review. We built an AI planner that reads scope and produces a baseline-ready schedule that opens in Primavera P6, where planners already work. Planning time is reported to fall from weeks to days.
- Planning time
- Weeks → days
An Australian engineering firm with a pipeline-engineering practice
Engineering · Pipelines
Safety Management Study Reports in 2 to 3 Days Instead of a Week
The firm wanted AI without siloed data, shadow AI or sensitive data leaving Australia. We built one agentic AI platform in its own Azure tenant, with the firm’s own records behind the answers. Its Safety Management Study reports now take 2 to 3 days instead of about a week.
- One reusable agentic AI platform in the firm’s own Azure tenant
- Purpose-built agents on it, with no shadow AI
- Sensitive data kept in Australia
Questions buyers ask.
How is AI used in construction and AEC?
In construction and AEC, AI is most useful where answers depend on the project record: RFIs, schedules, proposals and commitments buried in email. We ground Outlook workflows in project documents and prior RFIs, bring Primavera P6 and project-control feeds into a validated model, and draft the next action for the responsible person to review. Engineers approve replies before anything is sent.
Which data platform suits a construction or AEC team?
Construction teams need a platform that can join documents, schedules and project-control data, and the choice is open: Microsoft Fabric, Databricks or BigQuery for data, with Azure OpenAI, OpenAI, Anthropic, Gemini or a self-hosted model for AI. AEC Platform and Atlas AI, from the Mandelbulb Kit, work inside Outlook and Primavera P6 on the Blumcore runtime, so teams keep the software they already use.
How should a construction firm start with AI?
Start with the inbox, schedule or commitment trail that needs attention, often an RFI waiting for an answer. Bring it to a 30-minute working session and we scope a practical next step. If the case for project AI is still open, a Data Strategy Assessment reads the inbox, document and schedule dependencies behind one workflow and prices the gaps before you fund the build.
What governance do AI agents on construction projects need?
Set the identity and permissions of every project agent before it touches live correspondence. The agent should read only the documents it is permitted to, cite its sources in each draft and leave consequential actions behind an explicit human sign-off: engineers review RFI replies, and operators approve commitment closures. Owner’s representatives then get an evidence-linked view of what was decided and by whom.
What does a first construction engagement deliver?
A first engagement delivers one artifact scoped to a single workflow. Depending on the starting point, that is a grounded agent path for project-inbox work, an onboarding runbook for Primavera P6 and project-control feeds, a sign-off design for AI actions or a costed modernization map. It is scoped in the first working session, fixed price, and the findings are yours.
What we do.
For Construction & AEC, the four practices run as one loop.
Strategy & Advisory
Judgment before the build. We map what your data can actually do for the business, price the path, and hand you a plan you keep, whichever way you decide.
Modernization
We move your data out of aging systems onto a foundation built for what comes next, without losing a single number along the way.
Data & AI
The work your teams feel every day: live operational intelligence, reporting people trust, and AI that acts inside the systems you already run.
Data Estate Optimization
A running estate quietly gathers waste. We find it, tune it, and turn the savings into the budget for whatever you build next.
Partners and recognition.
Bring the RFI
waiting for an answer.
Bring the inbox, schedule or commitment trail that needs attention. We will scope a practical next step in a 30-minute working session.