[ TL;DR ]
A tender price commits the business to a particular job. When a requirement changes, the team needs to understand what it means for scope, supplier coverage, cost and delivery. That takes more than a revised paragraph in the response. It takes evidence that the right people can assess together.
CSM’s demonstration follows Harbor Construction, a fictional 80-person main contractor preparing several bids. AI agents assemble a 20-page review pack for the tender meeting, bringing scope, pricing, programme, risks and responsibilities into one view. The wider design connects those records as the tender develops.
At Riverside Specialist Hospital, a late ventilation addendum reveals that required independent validation is excluded from a supplier’s quotation. Agents bring the requirement, exclusion and estimate allowance together. The estimator can review the pricing basis, the planner can assess dependencies and the executive can see the commercial implications.
The opportunity is to return time to experienced people while making the basis of the bid easier to challenge. Harbor and the financial examples are fictional; the pack is demonstrated, while wider controls require implementation and testing.
• Less searching, copying and rebuilding before a review.
• Clearer connections between required work and priced work.
• Human ownership of scope, programme, price and release.
[ The Challenge ]
The estimator has three quotations, each covering slightly different work. The planner is working from an earlier specification. The bid manager is assembling the return. A late addendum arrives while the same people are supporting another tender.
Everyone needs to know what changed, where it appears elsewhere and whose decision it affects. Finding that out can mean reopening documents, comparing passages, chasing colleagues and rebuilding the meeting pack. Experienced people spend time reconstructing information before they can apply their expertise.
The commercial consequences cross roles. An excluded test may need another specialist and an allowance. That specialist may introduce a commissioning dependency. The programme may need mitigation, with a cost that must return to the estimate. A revised scope can also change an earlier approval.
When those connections are unclear, the meeting spends its time establishing the facts. A polished tender response can still carry an unresolved cost or delivery assumption.
For a contractor handling several bids, the pressure compounds. Shared people face competing deadlines, and the business needs a view of their workload as well as the position of each tender.
[ What CSM Built ]
CSM created a tender preparation demonstration showing agents assembling a review pack from uploaded information. The original specification, late addendum, supplier return and estimate allowance are brought together, with exact versions available for checking. Preparation can reach people through Slack, Teams or WhatsApp, with the detailed evidence behind it.
The wider operating design gives each requirement a connection to its source, package, pricing treatment and owner. A supplier exclusion connects to the scope question it creates. A programme assumption connects to the delivery promise being reviewed. Each tender retains its own record.
This gives the estimator a basis for comparison, the planner the changed constraints and the executive the decisions requiring authority. The proposed company view also makes shared workloads and competing deadlines visible.
Agents support retrieval, comparison, summaries and drafts. Explicit software rules must check arithmetic, permissions and approval status. People retain responsibility for the commercial position.
The design works around existing estimating, document and programme tools. Systems shown in the film are illustrative; live connections need permission, mapping and testing. The practical aim is to reduce preparation across the tools the team already uses.
[ Before the Tender Meeting ]
In the film, an overnight run prepares Monday’s briefing for Wednesday’s tender meeting. The wider proposed workflow starts with the opportunity and keeps preparation current as information arrives.
01
Set up the tender and its owners. Record the procurement route, submission deadline, clarification cut-off, required returns and internal reviews. Assign responsibility for the response, estimate, programme and commercial assessment. A person decides whether to bid. Agents prepare the supporting evidence.
02
Establish the controlling information. Maintain document and drawing registers, distinguish published revisions from upload versions and flag uncertain precedence. The newest file in a folder is not automatically the specification the team should follow.
03
Connect requirements with the work. Prepare the technical obligations, mandatory returns and evaluation criteria for verification. Link each item to a package, owner and response or pricing treatment. Retrieve relevant approved answers and project evidence, with dates and sources available for checking.
04
Build the review around open decisions. Gather quote gaps, changed requirements, programme questions and unresolved responsibilities into role-specific pre-reads. Prepare an agenda that shows what needs a decision and the evidence behind it.
Across live tenders, the proposed coordinator view exposes clashes in people’s availability. In the film, Cian sees Siobhan’s pressure across two bids. The agent flags the conflict; Cian decides what changes. Each bid keeps its own commercial evidence and approval position.
[ A Late Addendum ]
At Riverside Specialist Hospital, a late addendum changes the ventilation requirement. Independent validation is required, but the supplier’s quotation excludes it. The team needs to establish what the missing scope means before accepting the pricing basis.
01
Bring the change into view. Agents assemble the original and revised passages with source references. Reviewers can check the requirement against the documents, rather than rely on a summary detached from the specification.
02
Expose the gap in the quotation. The requirement sits beside the supplier’s exclusion and the existing estimate allowance. The estimator can see where required work and quoted work differ, then establish whether another package already carries the validation.
03
Prepare the comparable pricing basis. Keep the raw supplier return separate from adjustments for missing scope. In the synthetic example, a €2.25 million quotation excludes validation. A reviewed €0.35 million adjustment produces a €2.60 million comparison basis. The adjustment does not change the supplier’s offer.
04
Put the programme question in front of the planner. Assemble the relevant design, delivery, installation and commissioning dependencies. The planner assesses whether the change affects completion, available float or the need for mitigation.
05
Prepare the decision and affected reviews. Bring the cost movement, programme implications and approval questions together. The wider design calls for affected approvals to be revisited against the changed basis.
Nessa, the estimator, receives the requirement, exclusion and allowance together. Eoin, the planner, receives the dependency chain. Erin receives the commercial options. The demonstrated value is preparation for their assessment; the broader impact links and approval controls remain proposed features to validate in implementation.
[ Pricing and Programme ]
The synthetic worked example makes the commercial question concrete. Its deliberately large figures illustrate the mechanism; they are not Harbor’s results or a typical tender value for an 80-person contractor.
01
Compare equivalent scope. Supplier A’s €2.25 million return excludes validation. The reviewed adjustment brings its comparison basis to €2.60 million. Supplier B offers €2.70 million including validation. Warranty, technical and delivery conditions still need assessment.
02
Reconcile the estimate movement. The original package allowance was €1.80 million. Selecting A as the estimating basis raises that package by €0.80 million. The team needs the movement explained against the original allowance.
03
Include the cost of the programme decision. The scenario adds an approved €0.15 million for mitigation. The total forecast cost movement becomes €0.95 million. Changing the delivery approach has a commercial consequence that belongs in the estimate.
04
Show the consequence for profit. In this example, forecast cost rises from €106 million to €106.95 million. Holding the €112 million selling price leaves €5.05 million forecast gross profit, down from €6 million. That is an option for assessment, not a recorded loss.
05
Keep price and programme decisions accountable. The executive reviews the selling-price options. The planner tests the potential ten-working-day equipment shift against dependencies, calendars, float and mitigation before confirming any completion impact.
Agents prepare the evidence and comparisons. Calculations must reconcile under explicit rules, and specialists confirm the assumptions. The team can then see what accepting the price would mean for the business.
[ Reviewing and Approving ]
A useful settlement review brings together the scope, selected quotes, estimate, programme, qualifications and unresolved conditions. The proposed controls make that position explicit before the company commits to it.
01
Check that the return is complete. Verify mandatory documents, pricing schedules, declarations, signatures and current supporting evidence against the invitation. A file being uploaded is not evidence that its contents meet the requirement. Keep missing items visible with a named owner.
02
Check that the commercial and delivery positions agree. Reconcile quotes, allowances, exclusions, risk treatment and price. Review the method, milestones, procurement and resources behind the programme. Distinguish confirmed impacts from questions still awaiting assessment, and an internal qualification from one accepted by the employer.
03
Check authority and the release version. The authorised person approves the defined scope and price. Material changes reopen the affected review. The submission snapshot preserves the authorised documents, qualifications, programme assumptions and approvals, so the record shows exactly what was released.
Agents can also prepare draft minutes and actions from meeting transcripts. A discussion does not itself constitute approval, and an ambiguous action needs confirmed ownership.
In the proof of concept, portal operations and receipts are simulated. A production workflow must verify actual acknowledgement before describing a tender as received. These release controls require implementation and testing.
[ Why the Approach Works ]
Tender preparation involves repeated transitions: a requirement reaches the estimator, an exclusion reaches the commercial reviewer, a changed dependency reaches the planner. Each transition needs the source, the current position and the question requiring judgement.
A shared workspace can make those connections available to everyone responsible for the bid. The proposed record explains why a quotation was selected, how missing scope was treated and which assumptions supported the programme. Reviewers have a common basis to examine, with responsibilities attached to unresolved items.
That matters across several tenders. The company view can expose competing deadlines and shared workloads while each bid retains its own evidence, pricing and approvals. Suppliers must not gain access to competitors’ prices or internal margins. Permissions need to follow the source through search and generated briefs.
Reviewed answers, project evidence and lessons can become reusable material with ownership and validity checks. A previous answer still needs to fit the new opportunity.
If the work is won, a separate handover should compare the submitted bid with the accepted contract and carry forward agreed scope, supplier assumptions, programme, risks and open obligations.
The opportunity extends beyond preparing a meeting pack: preserve the reasoning so the next person can act on a supported position. Live permissions, integrations and handover behaviour still need validation.
[ Results ]
The film shows a 20-page review pack spanning scope, pricing, programme, risks and responsibilities. Its addendum example connects a changed requirement, supplier exclusion and estimate allowance for human assessment. It establishes a practical preparation pattern; the wider operating design has yet to be validated in production.
Less repeated preparation
Agents can take on the initial gathering and assembly that precedes specialist review: finding documents, comparing revisions, organising quotations and preparing evidence. A pilot should measure active team hours, including checking, corrections and system upkeep, to establish the net time returned.
Better-supported commercial decisions
The synthetic example shows why scope coverage matters to the price. A pilot should test whether reviewers can open the cited versions, reconcile adjustments and identify unresolved exposure. Margin protection is an intended benefit, not a measured Harbor outcome.
Clearer coordination across bids
The wider design makes competing deadlines, shared workloads and changed assumptions visible for responsible people to address. Test whether that visibility arrives early enough to change the work, with each bid’s records kept separate.
The next step is a focused trial on representative tenders. Compare preparation effort, late scope gaps and rework against the current process. If agents return useful time and improve the evidence behind decisions, the contractor has a grounded case for extending their role in tender preparation.