A comprehensive vision for what Connected is building – across proposal intelligence, digital asset management, client-facing proposals, Salesforce integration, and the legal framework to protect it all. Informed by the Discovery Service Matrix, the Connected Proposal Playbook, and the OpenAsset competitive context.
PreparedJune 2026
StatusDraft for review
AudienceConnected leadership and delivery team
Version3.0 – consolidated
01
What we are building and why
The OpenAsset context
OpenAsset is a purpose-built AEC SaaS platform that does two things: a centralised digital asset library – project photos, capability statements, whitepapers, client quotes, team references and testimonials, all searchable and tagged – and an AI-powered proposal module that ingests RFPs, runs go/no-go analysis, and assists with content assembly. It is, in essence, the infrastructure layer that would sit under everything Connected is building across AU and the US.
We should not buy it. We should build our own. The difference is what gets baked in. OpenAsset is a generic AEC tool. What Connected builds will have our ECI methodology, our sector verticals, our Playbook, our IP, and our data embedded at the foundation. That is not a feature difference – it is a structural competitive advantage that a SaaS subscription can never replicate.
The Connected Proposal Intelligence Platform is the internal system that makes Connected faster, more consistent, and more strategically rigorous in how it pursues and wins work. It starts with the discovery proposal generator – built and ready for deployment pending API configuration – and builds toward a full platform that spans proposal intelligence, digital asset management, client-facing proposal delivery, Salesforce integration, and AU/US operational split views.
Every layer of the platform compounds on the last. The data generated by proposals informs the asset library. The asset library feeds the proposal assembly engine. The proposal assembly engine powers the client-facing delivery layer. The analytics from that layer feed back into win strategy. Built correctly, this becomes a system that gets sharper with every bid Connected runs through it – and that is the version competitors cannot buy.
02
The four-layer platform
The platform is built in four layers, each extending the last. The first is built and ready to deploy. The rest are the roadmap.
Layer 1 – Ready to deploy
Proposal Intelligence
AI-assisted proposal narrative generation via the Discovery Generator. Built on the Connected Discovery Service Matrix and Proposal Playbook. Consultants select services, provide client context, upload a project brief, and receive a tailored proposal narrative in under two minutes. Prompt architecture drawn from the Playbook’s ten Proposal Intelligence Prompts.
Delivers
Discovery proposals generated in under 2 minutes once deployed. Playbook methodology enforced by default. Consistent quality regardless of author.
Layer 2 – Next
Digital Asset Library (DAM)
A lightweight, Connected-built digital asset management layer with Salesforce as the data backbone. Project photos, capability statements, whitepapers, case studies, team profiles, client quotes, and testimonials and referee contacts – all tagged, searchable, and linked to project records. Client quotes and team references are stored against the project and client they relate to, making them instantly retrievable when building a proposal for a similar sector, environment, or client type. Clean frontend for internal access. This is Connected’s version of OpenAsset’s asset library, built on infrastructure Connected already owns.
Delivers
A single source of truth for all proposal assets. No more hunting through shared drives. Assets linked to projects and sectors, ready to pull into proposals.
Layer 3 – Planned
Proposal Assembly Engine
Pulling project data, assets, and Playbook prompts into a single proposal output – Connected’s version of Shred.ai. The consultant provides the opportunity context. The platform assembles a structured first-draft proposal: executive summary, methodology, relevant case studies, team profiles, and service narrative. Output is a branded, editable Word document or PDF.
Delivers
A complete first-draft proposal in under five minutes. Human effort focused on strategy and relationships, not document production.
Layer 4 – Future
Full Platform
AU/US split views with regional content and team data. HubSpot/Salesforce sync for opportunity and client intelligence. Client-facing proposal delivery with engagement analytics. Win/loss feedback loop feeding back into prompt optimisation. Mailchimp integration for automated client-facing phase reports triggered by Salesforce project milestones. The complete intelligence system – from first RFT to post-award debrief.
Delivers
Connected’s proposal capability becomes a compounding strategic asset. The platform improves with every bid. Competitors can replicate individual proposals – they cannot replicate this system.
03
What it does now
The discovery proposal generator is the Layer 1 tool. It is purpose-built around Connected’s discovery methodology and Playbook. The interface is complete – deployment requires an Anthropic API key, a server-side proxy function to handle API requests securely without exposing credentials in the browser, and an access token to gate the site. Once configured, it is accessible to the full team from a URL.
01
Client details
Name, sector, project reference, author, date, and project context
02
Document upload
Optional PDF – brief, feasibility study, site report – read and used to tailor output
03
Service selection
Tick-box from the full discovery matrix, organised by phase
04
Generate
Tailored narrative in Connected’s voice, informed by the Playbook
05
Review and use
Consultant refines and pastes into the branded proposal template
1
Phase 1 – Understand and define
7 services
Foundation. Building a rigorous picture of the current state, people, and operational requirements before design thinking begins. Applicable across all six sectors.
Current state assessmentPast projects lessons workshopBusiness leader interviewsDepartmental functional briefingEquipment list and procurementServices and utility strategyReturn brief / URB / PPR
2
Phase 2 – Explore and assess
6 services
Options. Stress-testing the opportunity across property, location, funding, and scale. Not all services apply to every sector – the platform surfaces relevant ones based on sector selected.
Property / land due diligenceProperty scenarios comparisonPlanning pathwayFunding supportLocation / transit and distribution impactRight sizing scenarios
3
Phase 3 – Validate and de-risk
7 services
De-risking. Change readiness, business continuity, sustainability, redundancy, and business case – ensuring the project is structurally sound before design begins.
Change management assessmentRedundancy strategyBusiness continuity planningSustainability objective mappingBenchmarking toursRisk register and mitigationBusiness case development
04
Master prompt architecture
The platform’s AI output is determined by its master prompts – structured instructions drawn directly from the Connected Proposal Playbook. The Playbook defines a six-stage bid framework and ten AI prompt modules. The platform operationalises these in a structured interface so consultants don’t construct them manually.
The Playbook’s six-stage bid framework
Go/No-Go, Kick-Off, Win Strategy, Proposal Build, AI Prompts, Pre-Submission. The platform currently operationalises the AI Prompt stage (Stage 5) for discovery proposals. The full six-stage digital workflow is the Layer 2 and 3 build.
Stage 1Go / No-GoShould we even bid?
Strategic fit, delivery viability, and commercial risk assessed before committing resources. Platform (Layer 2): digital checklist with decision recorded against the opportunity in Salesforce.
Stage 2Kick-OffAlign the team fast
Submission logistics, format, team ownership, programme locked. Content owner per section. Draft date locked, not just the final. Platform (Layer 2): structured kick-off form generating a team briefing with roles and deadlines.
Stage 3Win StrategyWhat is the client really buying?
Real decision-maker identified. 3-5 win themes: client-facing, evidenced, differentiated, board-level readable. Risk intelligence mapped. Platform (Layer 1 on deployment, Layer 2 full): AI narrative generator uses win theme logic. Full win strategy module planned.
Stage 4Proposal BuildStructure and language standards
No generic cliches. No unprovable claims. Executive summary answers “why Connected?” specifically. Every section reinforces win themes. Platform (Layer 1 on deployment): the AI is explicitly instructed per the Playbook’s language standards.
Ten master prompts. Context provided once. Prompts run selectively. Prompt 10 first if time is short. Prompt 6 in the first win team meeting.Platform (Layer 1 on deployment, Layer 3 full): generator uses Playbook prompt architecture once deployed. Full ten-prompt suite is Layer 3.
Stage 6Pre-Submission ReviewRed-team before anything goes out
Sceptical evaluator review. Delivery team alignment confirmed. A win that leads to a poor project is not a win. Platform (Layer 2): digital red-team checklist using Prompt 10 to stress-test before submission.
The AI voice – Step 2 of the Playbook prompt framework
How the AI is instructed to think
The Playbook defines the AI’s role precisely. It acts simultaneously as four perspectives: the client’s CEO – outcomes, risk, and organisational consequences of getting this wrong; the client’s Head of Procurement – evaluation criteria, value for money, commercial risk; a bid director with 20 years winning complex capital works tenders in life sciences, healthcare, and laboratory environments; and a construction specialist who understands what actually goes wrong on projects like this.
The job of the AI is not to make Connected sound good. Its job is to tell Connected what they need to hear to win – and what they need to avoid to not lose. Every response must be specific to this project, this client, this environment. No construction cliches. No generic statements any contractor could make.
Step 1 – Context inputs required before any prompt runs
The Playbook specifies eight inputs. Quality of inputs determines quality of outputs. The platform pre-loads these from the project context form.
RFT / RFP document or summary – the brief or a detailed summary
Site constraints and programme milestones – contract type, key dates, conditions
Client organisation – who they are, what they care about, known intelligence
Connected’s pre-engagement work – site knowledge, prior relationship, any advantage built
Known or suspected competitors – who else is likely bidding
Connected’s proposed delivery model – how Connected intends to approach this project
Client meeting notes or call transcripts – direct intelligence from the client
Decision-maker personality profiles – what the person signing the appointment cares about
01
Client Decision Drivers
What will the person signing the appointment lose sleep over? Real decision drivers – not what the RFT says matters. Flags where stated criteria differ from real criteria. Identifies what typical bidders will get wrong.
Output: 3-5 real decision drivers and competitor failure modes
02
Risk Intelligence
Real risks across four categories: operational, programme, commercial, reputational. What risks are absent or underplayed in the RFT? What is most likely to emerge post-award?
Output: Risk register by category with post-award forecast
03
Risk to Advantage Conversion
For each major risk: how to mitigate in delivery and proposal language. Which risks, if owned early, become competitive advantages? Where does Connected’s pre-engagement work convert a risk into a differentiator?
Output: Mitigation language and differentiation opportunities per risk
04
Smart RFIs
Five to eight RFIs that reduce ambiguity, protect Connected’s ability to deliver, and demonstrate Connected has thought more deeply about this project than anyone else. No generic clarification questions.
Output: Drafted RFIs that signal intelligence, not compliance
05
Value and Programme Opportunities
Opportunities to improve programme certainty, reduce client-side disruption, or improve outcomes that a less-prepared contractor would miss. Specific action, benefit, and phase for each.
Output: Opportunity register with programme and cost benefit
06
Win Themes
Three to five win themes: specific, defensible, board-level readable, differentiated. One-line headline and two-sentence supporting statement each. Use this in the first win team meeting.
Output: Win theme set ready for proposal integration
07
Proposal Structure
Structure sequenced to score highest against likely evaluation criteria. Key message, evidence type, and common mistake to avoid per section. Where to lead with outcomes vs methodology.
Output: Section-by-section structure with key message and evidence type
08
Executive Summary Blueprint
Paragraph-by-paragraph blueprint: opening hook, project understanding, delivery approach, why Connected, closing commitment. Tone, key message, and one thing to avoid per paragraph. Language guidance specific to this client.
Output: Executive summary draft ready for consultant review
09
Competitive Intelligence
Likely competitors by project type, scale, location, and client profile. What each competitor’s proposal will look like. Where each is vulnerable. What Connected must do to make each competitor’s strength irrelevant.
Output: Competitor profiles with specific counter-positioning
10
How We Lose
The most important prompt. Set aside everything Connected does well. If Connected loses this project – what is the reason? What assumption turns out wrong? What will the winning competitor do that Connected underestimates? What section will the panel score lowest and why? Identifies the one or two things that must be fixed before submission.
Run this first if time is short. Non-negotiable before every major submission.
The Playbook’s own caveat on AI
AI cannot replace your relationships and your intimate knowledge of a client and their project specifics. This model and approach is a support, not a solution. Every output must be thoroughly reviewed. The platform’s job is to eliminate the blank-page problem, enforce structural discipline, and surface intelligence the team might miss under time pressure – not to substitute for the human judgment that wins work.
05
Client-facing proposals and engagement analytics
Layer 4 of the platform introduces a client-facing proposal delivery layer – proposals sent as interactive, tracked web documents rather than static PDFs. This turns every proposal submission into an intelligence-gathering exercise.
The idea
Instead of sending a PDF that disappears into a client’s inbox with no signal of what happens next, Connected sends a secure, branded proposal link. The client reads it in their browser. Connected sees exactly what they read, how long they spent on each section, what they skipped, what they returned to, and who they forwarded it to. That intelligence feeds directly back into the follow-up conversation – and over time, into win strategy.
Who
Reads it – individual viewer tracking per link
How long
Time spent on each section of the proposal
What
Sections read, skipped, and returned to
When
Open timestamps, re-reads, and forwarding events
Heatmap analytics
Section-by-section engagement
A visual heatmap of the proposal showing which sections received the most reading time, which were skimmed, and which were ignored entirely. If the evaluator spent 12 minutes on the methodology section and 30 seconds on the exec summary, that is a signal – and it should change how Connected follows up.
Viewer tracking
Who actually read it
Individual links can be generated per recipient. When the proposal is forwarded to someone Connected didn’t know was involved, that person appears in the viewer log. This surfaces hidden decision-makers and evaluation panel members that were not declared in the RFT process.
Re-read signals
Interest and concern indicators
A section read three times in one session typically means one of two things: the evaluator is impressed and wants to understand it better, or they have a concern. Combined with section time data, this gives Connected a view of where to focus the follow-up conversation.
Competitive intelligence
Informing future proposals
Aggregate engagement data across multiple proposals reveals which sections of Connected’s proposals consistently receive high engagement and which consistently get skimmed. This feeds directly back into Playbook prompt refinement and proposal structure decisions.
How this is built
Client-facing proposal analytics can be implemented via purpose-built tools (Qwilr, Proposify, Pandadoc – all offer section-level analytics and viewer tracking) or built natively as part of the Connected platform. The native build is the right long-term answer – it keeps the data inside Connected’s systems, integrates directly with Salesforce, and allows the analytics to feed back into the AI prompt layer. The SaaS tool approach is faster to deploy and appropriate for the Layer 4 phase while the native build is scoped.
Note: client-facing proposal links require careful handling – see Section 08 on legal and data obligations before deploying tracking to any recipient.
06
Salesforce integration
Salesforce is the data backbone of the platform. Every piece of client intelligence, opportunity data, and proposal outcome that Connected captures should live in Salesforce and be accessible to the platform. The integration makes the platform smarter with every engagement.
The principle
The platform should never ask a consultant for information that Salesforce already has. Client name, sector, project history, key contacts, and prior engagement notes should pre-populate the proposal form from the CRM record. Conversely, everything the platform generates – proposal content, win/loss outcomes, client engagement data – should write back to Salesforce automatically. The goal is a single source of truth that gets richer with every bid.
Data from Salesforce to platform
Pre-population and intelligence
Client name, contact details, account history, prior project data, sector classification, and any engagement notes pre-populate the proposal form from the Salesforce Opportunity record. Consultants confirm and augment – they do not re-enter. Decision-maker profiles from CRM contacts inform the AI’s client intelligence layer.
Data from platform to Salesforce
Outcome recording and learning
Proposal content, selected services, generated narratives, win/loss outcome, and client engagement analytics all write back to the Salesforce Opportunity record. Over time, this builds a dataset of what Connected proposed, how the client engaged with it, and whether it won – the foundation for the platform’s intelligence layer.
Digital Asset Library
Project data as the asset backbone
The DAM (Layer 2) uses Salesforce project records as the underlying data structure. Project photos, capability statements, case studies, client quotes, testimonials, and referee contacts are tagged to Salesforce project and client records. When the proposal assembly engine (Layer 3) pulls relevant case studies or supporting references for a proposal, it queries Salesforce for projects with matching sector, scale, and environment type – surfacing the right quote or testimonial for the right opportunity automatically.
AU / US split
Regional views and data separation
Salesforce supports multi-region data structures natively. AU and US operations each maintain their own Opportunity and Account data with a unified reporting layer above. The platform surfaces the relevant regional data based on the logged-in user’s region, with cross-region visibility for leadership.
Client quotes, testimonials, and team references as proposal assets
Client quotes and testimonials are among the highest-value assets in any proposal, yet they are consistently the hardest to find under time pressure. The DAM treats them as first-class assets, stored and tagged with the same rigour as project photography or capability statements.
Each quote or testimonial is tagged to: the client who provided it, the project it relates to, the sector and environment type, the specific outcome or capability it evidences, and the team member or role it references. When the proposal assembly engine runs a prompt against a new opportunity, it queries these tags to surface the most relevant quote or reference automatically – a healthcare client quote for a healthcare proposal, an operational continuity testimonial for a live environment bid.
Referee contacts are stored as a separate asset type linked to both the client record and the project record in Salesforce. When a tender requires nominated referees, the platform surfaces the relevant contacts based on project similarity – no more hunting through emails or relying on memory for who gave a good reference on which project. This alone is worth building the DAM for.
Integration point
Direction
What it enables
Build phase
Opportunity record pre-population
Salesforce to platform
Client and project details auto-fill the proposal form. No re-entry of data already in CRM.
Layer 2
Contact and decision-maker profiles
Salesforce to platform
Key contact intelligence informs the AI’s client decision driver analysis.
Layer 2
Proposal content write-back
Platform to Salesforce
Generated proposal narratives and selected services recorded against the Opportunity.
Layer 2
Win/loss outcome recording
Platform to Salesforce
Bid outcomes recorded and linked to proposal content – enabling win/loss pattern analysis over time.
Layer 3
DAM project record linkage
Bidirectional
Assets tagged to Salesforce project records, including client quotes, testimonials, referee contacts, and team references. Proposal assembly engine queries project data to pull relevant case studies and supporting evidence.
Layer 3
Client proposal engagement analytics
Platform to Salesforce
Heatmap data, viewer tracking, and re-read signals written to the Opportunity record for follow-up context.
Layer 4
HubSpot sync
Bidirectional
Marketing and business development data from HubSpot synchronised to Salesforce for unified client view.
Layer 4
Mailchimp integration
Platform to Mailchimp
Automated client-facing reports triggered by project phase milestones. When a project moves through a defined stage in Salesforce, a consolidated report is automatically generated and dispatched to the nominated client contacts via Mailchimp. See below for detail.
Layer 4
AU/US regional split views
Salesforce configuration
Region-specific Opportunity and Account data with unified leadership reporting layer.
Layer 4
Mailchimp integration – automated client reporting by project phase
Connected’s Mailchimp account connects to the platform as a client communications layer. When a project moves through a defined milestone in Salesforce – for example, Phase 1 discovery complete, Phase 2 options report issued, or Phase 3 validated brief signed off – the integration automatically compiles a consolidated project report and dispatches it to the nominated client contacts.
The report is generated from the project data already in the platform: services completed in that phase, key findings and outputs, next phase scope, and any relevant assets from the DAM. It is branded, consistent, and arrives without the consultant having to manually prepare a client update under project pressure. The trigger is the phase milestone in Salesforce, not a manual send.
This serves two purposes simultaneously. For the client it is a professional, timely, and informative touchpoint that reinforces Connected’s delivery standards. For Connected it is a data source – Mailchimp’s open and engagement data feeds back to Salesforce, showing which clients are actively engaged with the project narrative and which are not. A client who has not opened a Phase 2 options report before the Phase 3 kick-off is a relationship risk that the platform surfaces before it becomes a delivery problem.
07
What this achieves
Speed
Minutes, not hours
A discovery proposal narrative produced in under two minutes. A complete first-draft proposal in under five minutes by Layer 3. Senior consultants freed to focus on strategy and relationships.
Win rate
More rigorous, more tailored
Every bid passes through the Playbook’s strategic disciplines. Proposals that feel specific to the client’s situation and demonstrate deep project understanding consistently outperform generic submissions.
Intelligence
Proposals become data
Client engagement analytics, win/loss patterns, and section-level performance data compound over time. The platform gets sharper with every bid. Decisions are informed by evidence, not instinct.
Consistency
One standard of quality
The Playbook methodology is embedded in the tool, not in the individual. Junior and senior consultants produce proposals of the same structural quality. New hires operate to Connected’s standard from day one.
Proprietary IP
Built, not bought
OpenAsset can be purchased by any AEC firm. What Connected builds – with ECI methodology, sector verticals, Playbook prompts, and project data – is proprietary. It cannot be replicated by buying a SaaS subscription.
Scale
AU and US on one platform
The platform scales to Connected’s US expansion without duplicating infrastructure. Regional split views, separate team data, and unified leadership reporting – built on the Salesforce backbone from the start.
08
Legal and data obligations
Important
The following is a summary of legal areas that require professional advice before the platform is deployed in client-facing or data-intensive modes. This document does not constitute legal advice. Connected should engage an IT and commercial lawyer with experience in Australian privacy law, US data regulations, and AI/SaaS contracting before proceeding with Layers 3 and 4. The areas below define the scope of advice required.
Tracking how a client reads a proposal – heatmaps, viewer identification, time-on-section data – involves the collection and processing of personal information under the Australian Privacy Act 1988 (and equivalent US state laws for the US operation). This likely requires: disclosure in the proposal itself that the document is tracked, a privacy policy that covers the data collected, and potentially explicit consent from the recipient before tracking is activated. The method of individual viewer tracking (unique links per person) raises additional obligations around personal data collection.
Action: Obtain legal advice on disclosure requirements and consent mechanics before deploying any client-facing proposal tracking. Draft compliant disclosure language for inclusion in proposal cover pages.
02
AI-generated content and professional liability
Proposals generated by the platform may contain AI-drafted content that a client relies on in their decision to engage Connected. If that content contains inaccuracies, omissions, or commitments that cannot be fulfilled, there is potential professional liability exposure. Australian consumer law prohibits misleading or deceptive conduct – AI-generated content is not exempt from this standard. The platform’s status as a drafting tool that requires human review must be documented and enforced, not merely stated.
Action: Establish and document a review and sign-off protocol for all AI-generated proposal content before submission. Retain records of who reviewed and approved each proposal. Consider whether professional indemnity insurance covers AI-assisted content.
03
Client confidential information and AI processing
Client RFT documents, project briefs, site intelligence, and meeting notes uploaded to the platform are processed by Anthropic’s API. Connected likely has confidentiality obligations to clients under NDA, engagement terms, or implied duty of confidence. Sending client information to a third-party AI service may constitute a disclosure of confidential information. The client name minimisation feature partially addresses this, but does not cover all information types – particularly detailed project context, financial data, or site-specific intelligence.
Action: Review all active client NDAs and engagement terms for provisions that restrict third-party processing of confidential information. Consider adding AI processing disclosure to Connected’s standard engagement terms. Obtain Anthropic’s data processing agreement and review against client obligations. Upgrade to Anthropic’s enterprise tier for stronger contractual data protections as platform usage grows.
04
Intellectual property in AI-generated content
The ownership of AI-generated content is not yet definitively settled under Australian law. While Connected’s prompts and inputs are protected, the AI-generated output may have ambiguous IP status. Additionally, if the AI model was trained on third-party content that it reproduces in output, there is potential copyright exposure. For a platform that generates client-facing proposal content, clarity on IP ownership matters – particularly as the platform scales and the content becomes more central to Connected’s commercial proposition.
Action: Review Anthropic’s terms of service for IP ownership provisions (current terms assign output IP to the user, but this should be verified and documented). Consider IP assignment clauses in Connected’s engagement terms that cover AI-generated deliverables. Monitor Australian AI copyright developments – this area is evolving rapidly.
05
Data sovereignty and cross-border data transfer
Anthropic’s API processes data on US-based infrastructure. Under the Australian Privacy Act, sending personal information offshore requires that Connected take reasonable steps to ensure the recipient provides equivalent protections. The US operation introduces additional complexity – data about US clients may be subject to US state privacy laws (CCPA in California, for example) and federal regulations depending on the client’s industry. The Salesforce backbone has its own data residency configuration requirements.
Action: Confirm Anthropic’s data processing locations and obtain their DPA. Configure Salesforce for Australian data residency for AU operations. Review cross-border obligations for the US operation with a US-qualified attorney. Establish a data map showing where each category of information is processed and stored.
06
Platform as proprietary IP – protection strategy
The Connected platform – its prompt architecture, Playbook integration, discovery matrix structure, and the data it accumulates over time – represents significant proprietary IP. This IP needs to be protected: from staff who leave and take knowledge of the system, from competitors who might attempt to replicate it, and from inadvertent disclosure through demo environments, screenshots, or platform documentation shared externally.
Action: Ensure employment contracts include IP assignment clauses covering contributions to the platform. Implement confidentiality obligations specific to the platform in staff agreements. Register the platform’s distinctive elements as trade secrets where applicable. Do not share the Playbook, prompt architecture, or platform documentation externally without NDA. Consider whether core methodological elements are patentable – unlikely but worth a brief assessment.
07
Salesforce data governance and access control
Salesforce as the data backbone means that the platform’s value – client intelligence, project history, win/loss data, proposal content – is concentrated in a single system. This creates obligations around access control (who can see what), data retention (how long is data kept and who can delete it), and breach response (what happens if Salesforce is compromised or a staff member accesses data they should not). AU and US data must be appropriately separated to avoid inadvertent cross-jurisdictional data flows.
Action: Implement role-based access control in Salesforce aligned to need-to-know principles. Document data retention policies per data category. Establish a data breach response plan that covers Salesforce incidents. Configure Salesforce Shield or equivalent for audit logging of sensitive record access.
08a
Return on investment — cost vs time saved
The platform’s running cost is negligible. The time it saves is not. This section makes the financial case for building rather than waiting, and for prioritising deployment over further scoping.
The core argument
A senior consultant’s time is one of Connected’s most valuable and finite resources. A discovery proposal currently takes an estimated 2-4 hours of senior consultant time to write from scratch — time spent on drafting rather than strategy, relationships, or delivery. At a realistic volume of 2-4 proposals per month, that is 4-16 hours of senior time every month consumed by writing rather than winning. The platform produces an equivalent first draft in under two minutes at an API cost of roughly $0.02. Regardless of what that consultant time is worth in dollar terms, the platform pays for itself on the first proposal it generates.
Current cost per proposal
2-4 hrs
Estimated senior consultant time per discovery proposal written from scratch. At 4 proposals per month — Connected’s realistic maximum — that is up to 16 hours of senior time per month on drafting alone.
Platform cost per proposal
~$0.02
Anthropic API cost per generation at current Claude Sonnet pricing. Infrastructure hosting adds approximately $0.003 per request. Total: under $0.03.
Time saving per proposal
2-4 hrs
First draft produced in under 2 minutes. Consultant time shifts from writing from scratch to reviewing, refining, and applying judgment — the work that actually wins bids.
Proposals per month
Senior time currently spent
Platform cost
Senior time saved per month
Senior time saved per year
2
~4-8 hrs
<$1
~4-8 hrs
~50-90 hrs
3
~6-12 hrs
<$1
~6-12 hrs
~70-140 hrs
4 (max)
~8-16 hrs
<$1
~8-16 hrs
~95-190 hrs
What those hours actually represent
The table above assumes 2-4 hours per proposal — a reasonable estimate for a narrative written from scratch. The real figure is likely higher. Proposal writing typically pulls a senior consultant away from billable or client-facing work. It is rarely done in one sitting — meetings, interruptions, and review cycles add time that is hard to track but real. A more honest estimate puts the total time cost at 3-5 hours once coordination and review rounds are included.
At maximum volume — 4 proposals per month — the platform saves an estimated 8-16 senior consultant hours per month. That is one to two full working days of senior capacity returned to the team every month, every month, at a platform cost of under a dollar. Apply Connected’s own view of what that time is worth, and the ROI case writes itself.
Platform running cost — full scale
What the full platform costs to operate
Vercel hosting
~$20/mo
Anthropic API (25 proposals)
~$15/mo
Cloudinary (DAM storage)
~$50/mo
Domain and SSL
~$5/mo
Total platform cost
~$90/mo
Annual platform cost
~$1,080/yr
OpenAsset comparison
What not building costs
OpenAsset annual licence
~$20,000/yr
No proprietary methodology
Zero IP value
No ECI logic or Playbook
Generic AEC tool
No AU/US split view
Not built for scale
vs Connected platform
~$1,080/yr
Annual cost difference
$18,920 saved
The numbers that matter most
Running cost of the full platform: approximately $1,080 per year. Senior time saved at maximum volume — 4 proposals per month — approximately 95-190 hours per year — roughly 2-5 working weeks of senior capacity returned to the team annually. Annual cost saving vs buying OpenAsset: approximately $18,920. Apply Connected’s own view of what a senior consultant hour is worth, and the ROI case is unambiguous.
These figures use conservative assumptions on time and do not account for the win rate improvement that comes from more rigorous, more tailored proposals. That is the real return on investment and the figure that cannot be easily quantified but is the most important one. A single additional ECI award attributed to a sharper proposal covers the platform’s entire annual running cost many times over.
The case for building is not marginal. The platform pays for itself on the first proposal it generates. Everything after that is recovered senior capacity, compounding IP, and a widening gap between Connected and competitors still writing from a blank page.
09
Risks and mitigations
Over-reliance on AI output
The Playbook is explicit: AI cannot replace the relationships and intimate client knowledge that wins work. Risk is that consultants accept generated content without sufficient review, submitting proposals that are structurally correct but lack the human insight evaluators respond to.
Mitigation: all output positioned as a first draft requiring review. Senior sign-off on every submission. Onboarding reinforces this. The Playbook’s own caveat should be part of every team briefing on the platform.
Client data exposure via AI processing
Client RFT documents, project intelligence, and meeting notes uploaded to the platform are processed by Anthropic’s API. This may breach confidentiality obligations owed to clients, particularly under NDAs or implied duty of confidence in engagement terms.
Mitigation: review all client NDAs before uploading any client-specific documents. Add AI processing disclosure to Connected’s standard engagement terms. Obtain and review Anthropic’s DPA. Consider data minimisation – only upload what the AI genuinely needs to improve the output.
Client engagement tracking without consent
Deploying proposal heatmaps and viewer tracking without appropriate disclosure to recipients may breach the Australian Privacy Act and equivalent US regulations. Individual viewer tracking via unique links constitutes personal data collection.
Mitigation: do not deploy client-facing tracking before obtaining legal advice on disclosure and consent requirements. Draft compliant disclosure language. Consider opt-in rather than passive tracking for initial deployment.
Platform IP leakage
The Playbook, prompt architecture, and methodology baked into the platform represent Connected’s most valuable proprietary IP. Risk of disclosure through departing staff, external demos, or inadvertent sharing of platform documentation.
Mitigation: IP assignment clauses in all employment contracts. Confidentiality obligations specific to the platform. No external sharing of Playbook or prompt architecture without NDA. Demo environments should not contain real client data or expose full prompt logic.
Prompt quality degradation
If master prompts are not maintained as Connected’s methodology evolves, output quality will diverge from current best practice. Stale prompts produce stale proposals.
Mitigation: assign a prompt owner. Quarterly review against recent winning proposals. Update service descriptions, sector context, and Playbook-derived instructions accordingly.
Salesforce data concentration risk
Concentrating client intelligence, project history, and proposal data in Salesforce creates a high-value target. A data breach or staff member with inappropriate access could expose commercially sensitive client information.
Mitigation: role-based access control aligned to need-to-know. Salesforce Shield audit logging. Data breach response plan. Separation of AU and US data. Regular access reviews as team grows.
10
Immediate next steps
To deploy Layer 1 – five steps
1. Obtain an Anthropic API key from console.anthropic.com
2. Deploy the site to a hosting provider of choice and configure two environment variables on the server – API key and access token
3. Share the URL and access code with the team
4. Brief the team on what the tool does, what it does not do, and how to review the output – include the Playbook’s own guidance that AI is a support, not a solution
5. Collect feedback from the first ten proposals generated and use it to refine the master prompts
Layer 2 priorities
Sector-specific prompt variants – cleanroom, data centre, food manufacturing, healthcare. The current master prompt produces strong general output; sector-specific variants will materially improve quality for specialist pursuits
Salesforce integration scoping – map the Opportunity record fields to the proposal form inputs. Define what writes back to Salesforce and when. This shapes all subsequent layers
Digital asset library architecture – define the tagging taxonomy before building. Asset types include: project photos, capability statements, case studies, client quotes, testimonials, referee contacts, team profiles, whitepapers, and award citations. Tags should map to Salesforce project record fields – sector, environment type, scale, client – so assets link to projects automatically and surface at the right moment in proposal assembly
Word document output – branded .docx output so the generated narrative drops directly into the Connected proposal template without copy-paste
Mailchimp connection scoping – map the project phase milestones in Salesforce that will trigger automated client reports. Define the report template structure and the recipient logic before the integration is built. This shapes how client contacts are stored and tagged in Salesforce from Layer 2 onwards
Procore schema requirement — action required now – Procore holds Connected’s post-award project execution data: programme milestones, site photos, RFIs, and subcontractor records. The full Procore integration is a Layer 3/4 build, but a Procore_Project_ID field must be added to the Salesforce Opportunity object before Layer 2 ships. Without it, the link between the proposal record and the live project cannot be made at award — and the harvest loop (site photos into DAM, milestone triggers to Mailchimp, actual vs planned programme back into AI context) cannot close. One field, added now, unlocks the entire delivery intelligence layer later
Before Layers 3 and 4 – legal advice required
Engage an IT and commercial lawyer with Australian privacy law experience to review client tracking, AI processing obligations, and IP protection strategy before any client-facing features are deployed
Review all active client NDAs for AI processing restrictions before uploading client documents to the platform
Obtain Anthropic’s data processing agreement and assess against Connected’s client confidentiality obligations
Update Connected’s standard engagement terms to include AI processing disclosure and IP assignment clauses covering platform-generated content
Confirm Salesforce data residency configuration for AU and US operations before the DAM and proposal assembly layers go live
The longer-term vision – recording win and loss outcomes, feeding project history back into context, connecting proposal engagement analytics to Salesforce, and building a client-facing delivery layer – positions this as a strategic asset rather than a productivity tool. OpenAsset costs money and gives Connected nothing proprietary. What we build compounds with every bid, every project, and every client engagement Connected runs through it.
The platform that improves with use is the one competitors cannot buy.
11
Process maps — before and after
Four BPMN-style process diagrams showing how the platform changes Connected’s proposal and BD workflows — from the current manual state through to the full platform operating model across AU and US.
Current stateProposal workflow — manual
Platform statePlatform workflow — connected
DAM Layer 2Asset lifecycle — upload to reuse
US MarketBD pipeline — lead intel to ECI award
12
Tech stack — full platform
The platform is built entirely on infrastructure Connected already owns or can access at low cost. No new enterprise contracts. No vendor lock-in. Every layer is replaceable and independently deployable.
Tool
Purpose
Layer
Cost
Already have?
Vercel
Hosting + serverless functions for all tools
L1–4
Free → ~$20/mo
No — 5 min setup
Anthropic API
Claude AI for proposals, tagging, assembly
L1–3
~$20–80/mo usage
No — API key needed
Cloudinary
Asset file storage, CDN, image resizing
L2–4
Free → ~$50/mo
No — free tier sufficient
Salesforce
Asset metadata, project links, opportunity data
L2–4
Already paying
Yes
HubSpot
Contact sync, BD intel, US market
L4
Already paying
Yes
Mailchimp
Automated client reports, SF-triggered
L4
Already paying
Yes
Procore
Project execution data, milestones, site photos, programme
L3–4
Already paying
Yes — add Procore_Project_ID to SF now
OpenAsset (benchmark)
What we’re replacing / not buying
—
~$20,000/yr
No — and not buying
Total platform cost vs OpenAsset
At full operating scale across all four layers: ~$1,500/yr in platform costs (Vercel, Cloudinary, API usage) plus tooling Connected already pays for. OpenAsset runs ~$20,000/yr and gives Connected a generic AEC SaaS tool with no proprietary methodology, no ECI logic, and no AU/US operational split. The cost difference is not the main argument — the IP difference is.
13
End-to-end workflow
How a connected workflow runs across all four platform layers — from the moment a brief lands to post-award asset capture. Every step either currently exists, is in active build, or is on the near-term roadmap.
What the platform eliminates
Hunting for assets across shared drives and inboxes — replaced by a DAM search that takes seconds
Starting every proposal from a blank page — replaced by AI-generated narrative in under two minutes
Inconsistent go/no-go decisions — replaced by a scored Playbook checklist recorded against the Salesforce opportunity
Win/loss data that disappears after debrief — replaced by structured logging that feeds back into future proposals
What the platform compounds
Every asset uploaded enriches the library — proposals get better assets over time automatically
Every win and loss refines the AI prompts — output quality improves with use, not with more effort
Every project delivered generates new case studies — harvest loop closes itself as Connected builds
Every US engagement adds to a market intelligence base competitors cannot replicate
14
Department workflows
How each department interacts with the platform — their specific process flow, the tools they use, and the actions required from them. These sit beneath the master end-to-end workflow and can be extracted as standalone operational guides for each team.
Business Development
Lead intel → ECI award
BD’s primary interaction with the platform is pre-award: identifying targets, logging opportunities in Salesforce, sharing capability via DAM collections, and using the Discovery Generator to produce proposal content fast. The platform reduces the time between a warm conversation and a credible proposal from days to hours.
Team
SamTomJaimeRoyce
BD actions required
1.Log every qualifying opportunity in Salesforce at first contact — not after the pitch
2.Use DAM collection links for capability sharing — not emailed PDFs or shared drive links
3.Run the go/no-go checklist before briefing Marketing — not after
4.Log win/loss outcome with sector, competitor, and key loss reason — this feeds the AI
Platform value for BD
→Same-day capability sharing via a branded, trackable DAM link
→Discovery proposal narrative in under 2 minutes once context is entered
→Consistent go/no-go discipline across AU and US without BD manager oversight
→US market intel accumulates in SF — every engagement makes the next sharper
Marketing
Content → proposals → campaigns → analytics
Marketing has the most complex relationship with the platform — they are both the primary builder and the primary user. They own the DAM, govern the asset library, generate proposals, run campaigns, and manage the content pipeline across AU and US. The platform was designed around this function.
Team
ElleRoxJess
Marketing actions required
1.Own and govern the DAM — set the tagging taxonomy before Layer 2 ships
2.Upload all existing capability assets into DAM in the first sprint — no legacy shared drives
3.Review and approve all AI-generated proposal outputs before submission
5.Map Mailchimp campaign triggers to Salesforce + Procore milestones before Layer 4
Platform value for Marketing
→Eliminates shared drive chaos — one searchable asset library for AU and US
→Proposal generation time: days → under 2 minutes for first draft
→Campaign performance data flows back into Salesforce — no manual reporting
→Proposal engagement analytics show what clients actually read — informs follow-up
Pre-Construction
ECI engagement → design-ready brief
Pre-Con is where Connected’s ECI methodology is delivered. The Discovery Generator is built directly around this team’s workflow — the service matrix maps exactly to what Pre-Con does in a Discovery engagement. Their interaction with the platform is operationally detailed: briefing, functional reviews, risk identification, and brief development.
1.Provide project context to Marketing before any proposal is generated — not after
2.Use Playbook Prompt 10 (How We Lose) before every major ECI pitch submission
3.Upload discovery deliverables (URB, risk registers, functional briefs) to DAM post-engagement
4.Log ECI award and link Procore project to SF opportunity at award
Platform value for Pre-Con
→The platform’s proposal prompts are built around Pre-Con’s methodology — output is sector and phase-specific, not generic
→Past project lessons workshops and risk registers in the DAM surface automatically in future proposals for similar environments
→Playbook red-team prompts enforce pre-submission discipline without a bid manager
Construction
Project delivery → harvest → next bid
Construction’s interaction with the platform is post-award. They are the primary source of the asset library’s content — site photos, milestone data, and project evidence that feeds every future proposal. Their Procore workflow is the harvest engine. The ask from Construction is minimal: upload photos with correct project tags, and the platform handles the rest.
Key integration
Procore — primary toolProcore → DAM photo harvestProcore → Mailchimp triggersProcore_Project_ID in Salesforce
Construction actions required
1.At award: link Procore project to Salesforce opportunity via Procore_Project_ID field
2.Tag site photos in Procore with project reference and sector — enables auto-harvest to DAM
3.Upload project close case study to DAM within 4 weeks of PC — while detail is fresh
4.Contribute to the lessons workshop — findings go directly into the AI prompt library
Platform value for Construction
→Minimal ask: tag photos correctly in Procore and the harvest loop runs automatically
→Client milestone reports send automatically from Procore triggers — no manual reporting
→Every project delivered builds Connected’s reference library — directly improves the next bid for the same sector
→Actual vs promised programme data feeds back into AI — proposals become more accurate over time
